key: cord-0018007-owmp1ukp authors: Shi, Chunhu; Dumville, Jo C; Cullum, Nicky; Rhodes, Sarah; McInnes, Elizabeth title: Foam surfaces for preventing pressure ulcers date: 2021-05-06 journal: Cochrane Database Syst Rev DOI: 10.1002/14651858.cd013621.pub2 sha: 786707800ab6be5919f2af13a8705d138cc9df85 doc_id: 18007 cord_uid: owmp1ukp BACKGROUND: Pressure ulcers (also known as pressure injuries) are localised injuries to the skin or underlying soft tissue, or both, caused by unrelieved pressure, shear or friction. Foam surfaces (beds, mattresses or overlays) are widely used with the aim of preventing pressure ulcers. OBJECTIVES: To assess the effects of foam beds, mattresses or overlays compared with any support surface on the incidence of pressure ulcers in any population in any setting. SEARCH METHODS: In November 2019, we searched the Cochrane Wounds Specialised Register; the Cochrane Central Register of Controlled Trials (CENTRAL); Ovid MEDLINE (including In‐Process & Other Non‐Indexed Citations); Ovid Embase and EBSCO CINAHL Plus. We also searched clinical trials registries for ongoing and unpublished studies, and scanned reference lists of relevant included studies as well as reviews, meta‐analyses and health technology reports to identify additional studies. There were no restrictions with respect to language, date of publication or study setting. SELECTION CRITERIA: We included randomised controlled trials that allocated participants of any age to foam beds, mattresses or overlays. Comparators were any beds, mattresses or overlays. DATA COLLECTION AND ANALYSIS: At least two review authors independently assessed studies using predetermined inclusion criteria. We carried out data extraction, 'Risk of bias' assessment using the Cochrane 'Risk of bias' tool, and the certainty of the evidence assessment according to Grading of Recommendations, Assessment, Development and Evaluations methodology. If a foam surface was compared with surfaces that were not clearly specified, then the included study was recorded and described but not considered further in any data analyses. MAIN RESULTS: We included 29 studies (9566 participants) in the review. Most studies were small (median study sample size: 101 participants). The average age of participants ranged from 47.0 to 85.3 years (median: 76.0 years). Participants were mainly from acute care settings. We analysed data for seven comparisons in the review: foam surfaces compared with: (1) alternating pressure air surfaces, (2) reactive air surfaces, (3) reactive fibre surfaces, (4) reactive gel surfaces, (5) reactive foam and gel surfaces, (6) reactive water surfaces, and (7) another type of foam surface. Of the 29 included studies, 17 (58.6%) presented findings which were considered at high overall risk of bias. Primary outcome: pressure ulcer incidence Low‐certainty evidence suggests that foam surfaces may increase the risk of developing new pressure ulcers compared with (1) alternating pressure (active) air surfaces (risk ratio (RR) 1.59, 95% confidence interval (CI) 0.86 to 2.95; I(2) = 63%; 4 studies, 2247 participants), and (2) reactive air surfaces (RR 2.40, 95% CI 1.04 to 5.54; I(2) = 25%; 4 studies, 229 participants). We are uncertain regarding the difference in pressure ulcer incidence in people treated with foam surfaces and the following surfaces: (1) reactive fibre surfaces (1 study, 68 participants); (2) reactive gel surfaces (1 study, 135 participants); (3) reactive gel and foam surfaces (1 study, 91 participants); and (4) another type of foam surface (6 studies, 733 participants). These had very low‐certainty evidence. Included studies have data on time to pressure ulcer development for two comparisons. When time to ulcer development is considered using hazard ratios, the difference in the risk of having new pressure ulcers, over 90 days' follow‐up, between foam surfaces and alternating pressure air surfaces is uncertain (2 studies, 2105 participants; very low‐certainty evidence). Two further studies comparing different types of foam surfaces also reported time‐to‐event data, suggesting that viscoelastic foam surfaces with a density of 40 to 60 kg/m(3) may decrease the risk of having new pressure ulcers over 11.5 days' follow‐up compared with foam surfaces with a density of 33 kg/m(3) (1 study, 62 participants); and solid foam surfaces may decrease the risk of having new pressure ulcers over one month's follow‐up compared with convoluted foam surfaces (1 study, 84 participants). Both had low‐certainty evidence. There was no analysable data for the comparison of foam surfaces with reactive water surfaces (one study with 117 participants). Secondary outcomes Support‐surface‐associated patient comfort: the review contains data for three comparisons for this outcome. It is uncertain if there is a difference in patient comfort measure between foam surfaces and alternating pressure air surfaces (1 study, 76 participants; very low‐certainty evidence); foam surfaces and reactive air surfaces (1 study, 72 participants; very low‐certainty evidence); and different types of foam surfaces (4 studies, 669 participants; very low‐certainty evidence). All reported adverse events: the review contains data for two comparisons for this outcome. We are uncertain about differences in adverse effects between foam surfaces and alternating pressure (active) air surfaces (3 studies, 2181 participants; very low‐certainty evidence), and between foam surfaces and reactive air surfaces (1 study, 72 participants; very low‐certainty evidence). Health‐related quality of life: only one study reported data on this outcome. It is uncertain if there is a difference (low‐certainty evidence) between foam surfaces and alternating pressure (active) air surfaces in health‐related quality of life measured with two different questionnaires, the EQ‐5D‐5L (267 participants) and the PU‐QoL‐UI (233 participants). Cost‐effectiveness: one study reported trial‐based cost‐effectiveness evaluations. Alternating pressure (active) air surfaces are probably more cost‐effective than foam surfaces in preventing pressure ulcer incidence (2029 participants; moderate‐certainty evidence). AUTHORS' CONCLUSIONS: Current evidence suggests uncertainty about the differences in pressure ulcer incidence, patient comfort, adverse events and health‐related quality of life between using foam surfaces and other surfaces (reactive fibre surfaces, reactive gel surfaces, reactive foam and gel surfaces, or reactive water surfaces). Foam surfaces may increase pressure ulcer incidence compared with alternating pressure (active) air surfaces and reactive air surfaces. Alternating pressure (active) air surfaces are probably more cost‐effective than foam surfaces in preventing new pressure ulcers. Future research in this area should consider evaluation of the most important support surfaces from the perspective of decision‐makers. Time‐to‐event outcomes, careful assessment of adverse events and trial‐level cost‐effectiveness evaluation should be considered in future studies. Trials should be designed to minimise the risk of detection bias; for example, by using digital photography and by blinding adjudicators of the photographs to group allocation. Further review using network meta‐analysis will add to the findings reported here. Sauvage 2017 presented data for the questionnaire's subscales as numbers and percentages of responders with the specific subscales, and reported no significant difference in the overall satisfaction between study groups (P = 0.21). - (1 RCT) Very low e,f It is uncertain whether there is any difference in support surface associated patient comfort between alternating pressure (active) air surfaces and foam surfaces. All reported adverse events Follow-up: range 30 days to 6 months Nixon 2019 and Sauvage 2017 reported similar rates of adverse events between their study arms. Rosenthal 2003 reported 1 death but did not specify which study group the death was associated with. -2181 (3 RCTs) It is uncertain whether there is any difference in all reported adverse events between alternating pressure (active) air surfaces and foam surfaces. Health-related quality of life (90-day EQ-5D-5L, The mean health-related quality of life (90-MD 0 -267 (1 RCT) expressed as utility values ranging from −1 to 1 with 1 representing perfect health, 0 representing death, and −1 representing worse than death) Follow-up: 90 days day EQ-5D-5L) was 0.52. (0.05 lower to 0.05 higher) life measured using EQ-5D-5L at 90-day follow-up between foam surfaces and alternating pressure (active) air surfaces. Health-related quality of life (90-day PU-QoL-UI, expressed as utility values ranging from −1 to 1 with 1 representing perfect health, 0 representing death, and −1 representing worse than death) Follow-up: 90 days The mean health-related quality of life (90day PU-QoL-UI) was 0.60. MD 0 (0.03 lower to 0.03 higher) It is uncertain if there is a difference in health-related quality of life measured using PU-QoL-UI at 90-day follow-up between foam surfaces and alternating pressure (active) air surfaces. Cost-effectiveness Follow-up: 90 days Incremental cost-effectiveness ratio (ICER) = GBP -101,699 and net-monetary benefit (NMB) = GBP -2114 in the probabilistic analysis, meaning alternating pressure (active) air surfaces have lower costs and higher quality-adjusted life-years (QALY) values. Alternating pressure (active) air surfaces had a 99% probability of being cost-effective at a threshold of GBP 20,000 and alternating pressure (active) air surfaces dominated reactive foam surfaces. Comments Foam surfaces may increase the proportion of participants developing a new pressure ulcer compared with reactive air surfaces. Included studies did not report this outcome. Support surface associated patient comfort Follow-up: 13 days More people using reactive air surfaces had comfort increased than using foam surfaces on top of an alternating pressure (active) air surface; less had comfort decreased (P = 0.04). - It is uncertain if there is a difference in patient comfort responses between reactive air surfaces and foam surfaces on top of an alternating pressure (active) air surface. All reported adverse events Follow-up: 13 days There appeared to be similar rates of patients having adverse events between those using foam surfaces and those using reactive air surfaces. - (1 RCT) It is uncertain if there is a difference in adverse events between foam surfaces and reactive air surfaces. Proportion of participants developing a new pressure ulcer Follow-up: unspecified Hoshowsky 1994, involving a totality of 135 individuals (270 halves of bodies), indicated no pressure ulcers developed in either group. (1 RCT) Time to pressure ulcer development The included study did not report this outcome. Support surface associated patient comfort The included study did not report this outcome. All reported adverse events The included study did not report this outcome. Health-related quality of life The included study did not report this outcome. Cost-effectiveness The included study did not report this outcome. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. a Downgraded twice for high risk of bias in domains other than performance bias. b Downgraded twice for imprecision due to the small sample size and the low event rate. Foam surfaces compared to reactive foam and gel surfaces for pressure ulcer prevention Patient or population: pressure ulcer prevention Setting: operating room Intervention: foam surfaces Comparison: reactive foam and gel surfaces Outcomes Impact № of participants (studies) Proportion of participants developing a new pressure ulcer Follow-up: unspecified Hoshowsky 1994 compared foam surfaces and reactive foam and gel surfaces in 91 participants (with 182 halves of bodies) using a split body design. The study authors found that no pressure ulcers developed in either group. (1 RCT) Library Trusted evidence. Informed decisions. Better health. Cochrane Database of Systematic Reviews Pressure ulcers -also known as pressure injuries, pressure sores, decubitus ulcers and bed sores -are localised injuries to the skin or underlying so tissue (or both), caused by unrelieved pressure, shear or friction (NPIAP 2016) . Pressure ulcer severity is generally classified as follows, using the National Pressure Injury Advisory Panel (NPIAP) system (NPIAP 2016). • Stage 1: intact skin with a local appearance of non-blanchable erythema. • Stage 2: partial-thickness skin loss with exposed dermis. • Stage 3: full-thickness skin loss. • Stage 4: full-thickness skin and tissue loss with visible fascia, muscle, tendon, ligament, cartilage or bone. • Unstageable pressure injury: full-thickness skin and tissue loss that is obscured by slough or eschar so that the severity of injury cannot be confirmed. • Deep tissue pressure injury: local injury of persistent, nonblanchable deep red, maroon, purple discolouration or epidermal separation revealing a dark wound bed or bloodfilled blister. The stages described above are consistent with those described in another commonly used system, the International Classification of Diseases for Mortality and Morbidity Statistics (World Health Organization 2019). Pressure ulcers are complex wounds that are relatively common, a ecting people across di erent care settings. A systematic review found that prevalence estimates for people a ected by pressure ulcers in communities of the UK, USA, Ireland and Sweden ranged from 5.6 to 2300 per 10,000 depending on the nature of the population surveyed (Cullum 2016) . A subsequent cross-sectional survey of people receiving community health services in one city in the UK estimated that 1.8 people per 10,000 have a pressure ulcer (Gray 2018 ). Pressure ulcers confer a heavy burden in terms of personal impact and use of health-service resources. Having a pressure ulcer may impair physical, social and psychological activities (Gorecki 2009). Ulceration impairs health-related quality of life (Essex 2009); can result in longer institution stays (Theisen 2012); and increases the risk of systemic infection (Espejo 2018) . There is also substantial impact on health systems: a 2015 systematic review of 14 studies across a range of care settings in Europe and North America showed that costs related to pressure ulcer treatment ranged from EUR 1.71 to EUR 470.49 per person, per day (Demarré 2015). In the UK, the annual average cost to the National Health Service for managing one person with a pressure ulcer in the community was estimated to be GBP 1400 for a Stage 1 pressure ulcer and more than GBP 8500 for more severe stages (2015 /2016 Guest 2018) . In Australia, the annual cost of treating pressure ulcers was estimated to be AUD 983 million (95% confidence interval (CI) 815 million to 1151 million) at 2012/2013 prices (Nguyen 2015) . The serious consequences of pressure ulceration have led to an intensive focus on their prevention. Pressure ulcers are considered largely preventable. Support surfaces are specialised medical devices designed to relieve or redistribute pressure on the body, or both, in order to prevent pressure ulcers (NPIAP S3I 2007) . Types of support surface include, but are not limited to, integrated bed systems, mattresses and overlays (NPIAP S3I 2007) . The NPIAP Support Surface Standards Initiative (S3I) system can be used to classify types of support surface (NPIAP S3I 2007) . According to this system, support surfaces may: • be powered (i.e. require electrical power to function) or nonpowered; • passively redistribute body weight (i.e. reactive pressure redistribution), or mechanically alternate the pressure on the body to reduce the duration of pressure (i.e. active pressure redistribution); • be made of a range of materials, including but not limited to: air cells, foam materials, fibre materials, gel materials, sheepskin for medical use and water-bags; and • be constructed of air-filled cells that have small holes on the surface for blowing out air to dry skin (i.e. low-air-loss feature) or have fluid-like characteristics via forcing filtered air through ceramic beads (i.e. air-fluidised feature), or have neither of these features. Full details of classifications of support surfaces are listed in Appendix 1. A widely used type of support surface is the foam bed or mattress. These beds or mattresses are commonly non-powered and are made of foam materials which confer reactive pressure redistribution over a larger contact area (NPIAP S3I 2007 ) . The aim of using support surfaces to prevent pressure ulceration is to redistribute pressure beneath the body, thereby increasing blood flow to tissues and relieving distortion of the skin and so tissue (Wounds International 2010) . Reactive support surfaces (e.g. foam surfaces) achieve pressure redistribution by passive mechanisms, including immersion (i.e. 'sinking' of the body into a support surface) and envelopment (i.e. conforming of a support surface to the irregularities in the body). These devices distribute the pressure over a greater area, thereby reducing the magnitude of the pressure at specific sites (Clark 2011). Support surfaces are widely used for preventing pressure ulcers and are the focus of recommendations in international and national guidelines (EPUAP/NPIAP/PPPIA 2019; NICE 2014). Since the publication of the Cochrane Review, 'Support surfaces for pressure ulcer prevention' (McInnes 2015) , there has been a substantial increase in the number of relevant randomised controlled trials published in this area. The NPIAP S3I 2007 support surface-related terms and definitions have also been internationally recognised, and Cochrane has developed new methodological requirements, such as the use of GRADE assessments (Guyatt 2008) . These developments necessitate an update of the evidence base. In considering this evidence update, we took into account the size and complexity of the published review (McInnes 2015) , which includes all types of support surface. An alternative approach is to Library Trusted evidence. Informed decisions. Better health. Cochrane Database of Systematic Reviews split the review into multiple new titles, each with a narrower focus. We consulted on this splitting option via an international survey in August 2019. The potential new titles suggested were based on clinical use, the new terms and definitions related to support surfaces (NPIAP S3I 2007) , a relevant network meta-analysis (Shi 2018a) , and current clinical practice guidelines (EPUAP/NPIAP/ PPPIA 2019; NICE 2014). We received responses from 29 health professionals involved in pressure ulcer prevention activity in several countries (Australia, Belgium, China, Italy, the Netherlands and the UK). In total, 83% of respondents supported splitting the review into the suggested titles and 17% were unsure (no respondent voted against splitting). The new review titles are as follows. • Alternating pressure (active) air surfaces for preventing pressure ulcers. • Foam surfaces for preventing pressure ulcers. • Reactive air surfaces for preventing pressure ulcers. • Alternative reactive support surfaces (non-foam and non-airfilled) for preventing pressure ulcers. We will bring the results of these new reviews together in an overview with a network meta-analysis (Salanti 2012) , in order to simultaneously compare all support surfaces and to rank them based on the probabilities of each being the most e ective for preventing pressure ulcers. This particular review compares foam beds, mattresses or overlays with any surface. To assess the e ects of foam beds, mattresses or overlays compared with any support surface on the incidence of pressure ulcers in any population in any setting. We included published and unpublished randomised controlled trials (RCTs), including multi-armed studies, cluster-RCTs and cross-over trials, regardless of the language of publication. We also included RCTs with particular designs (factorial design, nof-1 trials). We excluded studies using quasi-random allocation methods (e.g. alternation). We included studies in any population, including those defined as being at risk of ulceration, as well as those with existing pressure ulcers at baseline (when the study measured pressure ulcer incidence). Eligible studies included foam beds, overlays or mattresses. We included studies where two or more mattresses were used sequentially over time or in combination, where the mattress(es) of interest were included in one of the study arms. We included studies comparing eligible foam beds, overlays or mattresses against any comparator defined as a support surface. Comparators could be: • non-foam surfaces, including: alternating pressure (active) air surfaces such as alternating pressure (or dynamic) air mattresses, reactive air surfaces (e.g. static air overlays, dry flotation mattresses, air-fluidised beds), and non-foam and nonair-filled surfaces (e.g. reactive gel surfaces such as a gel pad used on an operating table, reactive fibre surfaces such as Silicore fibre overlay, reactive water surfaces, reactive sheepskin surfaces such as Australian Medical Sheepskins overlay), or • a di erent type of foam surface. We included studies in which co-interventions (e.g. repositioning) were delivered, provided that co-interventions were the same in all arms of the study (i.e. interventions randomised were the only systematic di erence). We considered the following primary and secondary outcomes. If a study did not report any review-relevant outcomes but was otherwise eligible (i.e. eligible study design, participants and interventions), we contacted the study authors (where possible) to clarify whether they measured a relevant outcome but did not report it. We considered the study as 'awaiting classification' if we could not establish whether it measured an outcome or not. We excluded the study if the study authors confirmed that they did not measure any review-relevant outcomes. If a study measured an outcome at multiple time points, we considered outcome measures at three months as being of primary interest to this review (Schoonhoven 2007) , regardless of the time points specified as being of primary interest by the study. If the study did not report three-month outcome measures, we considered those closest to three months. Where a study only reported a single time point, we considered these data in this review. Where a study did not specify a time point for its outcome measurement, we assumed this was the final duration of follow-up noted. Our primary outcome was pressure ulcer incidence. We recorded two outcome measures (the proportion of participants developing a new pressure ulcer; and time to pressure ulcer incidence), where available. However, we considered the proportion of participants developing a new pressure ulcer as the primary outcome for this review. Our preferred measure was time to pressure ulcer incidence. However, we did not expect it to be reported in many studies. We extracted and analysed time-to-event data but we focused on the binary outcome in our conclusions. We accepted authors' definitions of an incident ulcer regardless of which pressure ulcer severity classification was used to measure or grade new pressure ulcers. We also considered the outcome of pressure ulcer incidence irrespective of whether studies reported ulcers by stages or as a non-stratified value. We did not consider subjective outcome measures (e.g. 'better' or 'worse' skin condition) as measures of pressure ulcer incidence. • Patient support-surface-associated comfort. We considered patient comfort outcome data in this review only if the evaluation of patient comfort was pre-planned and was systematically conducted across all participants in the same way Library Trusted evidence. Informed decisions. Better health. Cochrane Database of Systematic Reviews in a study. The definition and measurement of this outcome varied from one study to another; for example, the proportion of participants who report comfort, or comfort measured by a scale with continuous (categorical) numbers. We planned to include these data with di erent measurements in separate meta-analyses when possible. • All reported adverse events (measured using surveys or questionnaires, other data capture process or visual analogue scale). We included data where study authors specified a clear method for collecting adverse event data. Where available, we extracted data on all serious and all non-serious adverse events as an outcome. We recorded where it was clear that events were reported at the participant level or whether multiple events per person were reported, in which case appropriate adjustments were required for data clustering (Higgins 2019a) . We considered the assessment of any event in general defined as adverse by participants, health professionals, or both. • Health-related quality of life (measured using a standardised generic questionnaire such as EQ-5D (Herdman 2011) , 36item Short Form (SF-36; Ware 1992), or pressure ulcer-specific questionnaires such as the PURPOSE Pressure Ulcer Quality of Life (PU-QOL) questionnaire (Gorecki 2013), at noted time points). We did not include ad hoc measures of quality of life or qualitative interviews of quality of life because these measures were unlikely to be validated. • Cost e ectiveness: within-trial cost-e ectiveness analysis comparing mean di erences in e ects with mean cost di erences between the two arms. We extracted data on incremental mean cost per incremental gain in benefit (incremental cost-e ectiveness ratio (ICER)). We also considered other measures of relative cost-e ectiveness (e.g. net monetary benefit, net health benefit). We searched the following electronic databases to identify reports of relevant clinical trials: • Search strategies for clinical trials registries can be found in Appendix 2. For previous versions of McInnes 2015, the review authors of McInnes 2015 contacted experts in the field of wound care to enquire about potentially relevant studies that are ongoing, or recently published. In addition, the review authors of McInnes 2015 contacted manufacturers of support surfaces for details of any studies manufacturers were conducting. This approach did not yield any additional studies; therefore, we did not repeat it for this review. We identified other potentially eligible studies or ancillary publications by searching the reference lists of retrieved included studies, as well as relevant systematic reviews, meta-analyses and health technology assessment reports. When necessary, we contacted authors of key papers and abstracts to request further information about their trials. We did not perform a separate search for adverse e ects of interventions used. We considered adverse e ects described in included studies only. We carried out data collection and analysis according to the methods stated in the published protocol (Shi 2020) , which were based on the Cochrane Handbook for Systematic Reviews of Interventions (Li 2019). Changes from the protocol or previous published versions of the review are documented in Di erences between protocol and review. One review author re-checked the RCTs included in McInnes 2015 for eligibility (CS). Two review authors or researchers (CS and Asmara Jammali-Blasi, or JCD) independently assessed the titles and abstracts of the new search results for relevance using Rayyan (Ouzzani 2016) (Di erences between protocol and review), and then independently inspected the full text of all potentially eligible studies. The two review authors or researchers (CS and Asmara Jammali-Blasi, or JCD) resolved disagreements through discussion and by involving another review author, if necessary. One review author checked data from the studies included in McInnes 2015 and extracted additional data where necessary (CS). A second review author or researcher (SR, EM, Zhenmi Liu, Gill Norman, or Melanie Stephens) checked any new data extracted. For new included studies, one review author (CS) independently extracted data and another review author or researcher (SR, EM, Library Trusted evidence. Informed decisions. Better health. Zhenmi Liu, Gill Norman, or Melanie Stephens) checked all data (Di erences between protocol and review). Any disagreements were resolved through discussion and, if necessary, with the involvement of another review author. Where necessary, we contacted the authors of included studies to clarify data. We extracted these data using a pre-prepared data extraction form: • basic characteristics of studies (first author, publication type, publication year and country); • funding sources; • care setting; • characteristics of participants (trial eligibility criteria, average age in each arm or in a study, proportions of participants by gender and participants' baseline skin status); • support surfaces being compared (including their descriptions); • details on any co-interventions; • duration of follow-up; • the number of participants enrolled; • the number of participants randomised to each arm; • the number of participants analysed; • participant withdrawals with reasons; • the number of participants developing new ulcers (by ulcer stages where possible); • data on time to pressure ulceration; • patient support-surface-associated comfort; • adverse event outcome data; • health-related quality of life outcome data; and • cost-e ectiveness outcome data. We (CS and NC) classified specific support surfaces in the included studies into intervention groups using the NPIAP S3I support surface-related terms and definitions (NPIAP S3I 2007) . Therefore, to accurately assign specific support surfaces to intervention groups, we extracted full descriptions of support surfaces from included studies, and when necessary supplemented the information with that from external sources such as other publications about the same support surface, manufacturers' or product websites, and expert clinical opinion (Shi 2018b ). If we were unable to define any of specific support surfaces evaluated in an included study, we extracted available data and reported these as additional data outside the main review results. Two review authors or researchers (CS and SR, EM, Zhenmi Liu, Gill Norman, or Melanie Stephens) independently assessed risk of bias of each included study using the Cochrane 'Risk of bias' tool (see Appendix 3). This tool has seven specific domains: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete data (attrition bias), selective outcome reporting (reporting bias), and other issues (Higgins 2017) . We assessed performance bias, detection bias, and attrition bias separately for each of the review outcomes (Higgins 2017) . We noted that it is o en impossible to blind participants and personnel in device trials. In this case, performance bias may be introduced if knowledge of treatment allocation results in deviations from intended interventions, di erential use of co-interventions or care between groups not specified in the study protocol that may influence outcomes. We attempted to understand if, and how, included studies compensated for challenges in blinding; for example, implementing strict protocols to maximise consistency of co-interventions between groups to reduce the risk of performance bias. We also noted that pressure ulcer incidence is a subjective outcome. Compared with blinded assessment, nonblinded assessment of subjective outcomes tends to be associated with more optimistic e ect estimates of experimental interventions in RCTs (Hróbjartsson 2012). Therefore, we judged non-blinded outcome assessment as being at high risk of detection bias. In this review, we included the issues of di erential diagnostic activity and unit of analysis under the domain of 'other issues'. For example, unit of analysis issues occurred where a cluster-randomised trial had been undertaken but analysed at the individual level in the study report. For the studies included in McInnes 2015, one review author (CS) checked the 'Risk of bias' judgements and, where necessary, updated them. A second review author or researcher (SR, EM, Zhenmi Liu, Gill Norman, or Melanie Stephens) checked any updated judgement. We assigned each 'Risk of bias' domain a judgement of high, low, or unclear risk of bias. We resolved any discrepancy through discussion and by involving another review author where necessary. Where possible, useful and feasible, when a lack of reported information resulted in a judgement of unclear risk of bias, we planned to contact study authors for clarification. We present our assessment of risk of bias for the proportion of participants developing a new pressure ulcer outcome using two 'Risk of bias' summary figures. One is a summary of bias for each item across all studies, and the second shows a cross-tabulation of each study by all of the 'Risk of bias' items. Once we had given our judgements for all 'Risk of bias' domains, we judged the overall risk of bias for each outcome across studies as: • low risk of bias, if we judged all domains to be at low risk of bias; • unclear risk of bias, if we judged one or more domains to be at unclear risk of bias and other domains were at low risk of bias but no domain was at high risk of bias; or • high risk of bias, as long as we judged one or more domains as being at high risk of bias, or all domains had unclear 'Risk of bias' judgements, as this could substantially reduce confidence in the result. We resolved any discrepancy between review authors through discussion and by involving another review author where necessary. For studies using cluster randomisation, we planned to consider the risk of bias in relation to recruitment bias, baseline imbalance, loss of clusters, incorrect analysis and comparability with individually randomised studies (Eldridge 2019; Higgins 2019b) (Appendix 3). However, we did not include any studies with a cluster design. For meta-analysis of pressure ulcer incidence data, we present the risk ratio (RR) with its 95% confidence interval (CI). For continuous outcome data, we present the mean di erence (MD) with 95% CIs for studies that use the same assessment scale. If studies reporting continuous data used di erent assessment scales, we planned to report the standardised mean di erence (SMD) with 95% CIs. However, this was not undertaken in the review. Trusted evidence. Informed decisions. Better health. For time-to-event data (time to pressure ulcer development), we present the hazard ratio (HR) with its 95% CI. If included studies reporting time-to-event data did not report an HR, when feasible we estimated this using other reported outcomes (such as numbers of events) through employing available statistical methods (Parmar 1998; Tierney 2007) . We noted whether studies presented outcomes at the level of cluster (e.g. ward, research site) or at the level of participants. We also recorded whether the same participant was reported as having multiple pressure ulcers. Unit of analysis issues may occur if studies randomise at the cluster level but the incidence of pressure ulcers is observed and data are presented and analysed at the level of participants (clustered data). We noted whether data regarding participants within a cluster were (incorrectly) treated as independent within a study, or were analysed using within-cluster analysis methods. If clustered data were incorrectly analysed, we recorded this as part of the 'Risk of bias' assessment. If a cluster-RCT was not correctly analysed, we planned to use the following information to adjust for clustering ourselves where possible, in accordance with guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019b) . • The number of clusters randomly assigned to each intervention, or the average (mean) number of participants per cluster. • Outcome data ignoring the cluster design for the total number of participants. • Estimate of the intra-cluster (or intra-class) correlation coe icient (ICC). However, we did not identify any n-of-1 trials in this review. We did not adjust for clustering for the two studies with treatment sessions of each participant as the unit of analysis because they did not report su icient information to facilitate this (Bliss 1995a; Hoshowsky 1994). For cross-over trials, we only considered outcome data at the first intervention phase (i.e. prior to cross-over) as eligible. If a study had more than two eligible study groups, where appropriate we combined results across these arms to make single pair-wise comparisons (Higgins 2019b) . Data are commonly missing from study reports. Reasons for missing data could be the exclusion of participants a er randomisation, withdrawal of participants from a study, or loss to follow-up. The exclusion of these data from analysis may break the randomisation and potentially introduces bias. Where there were missing data and where relevant, we contacted study authors to pose specific queries about these data. In the absence of other information, for pressure ulcer incidence we assumed that participants with missing data did not develop new pressure ulcers for the main analysis (i.e. we added missing data to the denominator but not the numerator). We examined the impact of this assumption through undertaking a sensitivity analysis (see Sensitivity analysis). When a study did not specify the number of randomised participants prior to dropout, we used the available number of participants as the number randomised. Assessing heterogeneity can be a complex, multifaceted process. Firstly, we considered clinical and methodological heterogeneity; that is, the extent to which the included studies varied in terms of participant, intervention, outcome, and other characteristics including duration of follow-up, clinical settings, and overall studylevel 'Risk of bias' judgement (Deeks 2019). In terms of the duration of follow-up, in order to assess the relevant heterogeneity, we recorded and categorised assessment of outcome measures as follows: • up to eight weeks (short-term); • more than eight weeks to 16 weeks (medium-term); and • more than 16 weeks (long-term). We supplemented this assessment of clinical and methodological heterogeneity with information regarding statistical heterogeneity assessed using the Chi 2 test. We considered a P value of less than 0.10 to indicate statistically significant heterogeneity given that the Chi 2 test has low power, particularly in the case where studies included in a meta-analysis have a small sample size. We carried out this statistical assessment in conjunction with the I 2 statistic (Higgins 2003) , and the use of prediction intervals for randome ects meta-analyses (Borenstein 2017; Riley 2011). The I 2 statistic is the percentage of total variation across studies due to heterogeneity rather than chance (Higgins 2003) . Very broadly, we considered that I 2 values of 25% or less may indicate a low level of heterogeneity and values of 75% or more may indicate very high heterogeneity (Higgins 2003) . For random-e ects models where the meta-analysis has more than 10 included studies and no clear funnel plot asymmetry, we also planned to present 95% prediction intervals (Deeks 2019). We planned to calculate prediction intervals following methods proposed by Borenstein 2017. Random-e ects analyses produce an average treatment e ect, with 95% confidence intervals indicating where the true population average value is likely to lie. Prediction intervals quantify variation away from this average due to between-study heterogeneity. The interval conveys where a future study treatment e ect estimate is likely to fall based on the data analysed to date (Riley 2011). Prediction intervals are always wider than confidence intervals (Riley 2011). It is important to note that prediction intervals will reflect heterogeneity of any source, including from methodological issues as well as clinical variation. For this reason, some authors have suggested that prediction intervals are best calculated for studies at low risk of bias to ensure intervals that have meaningful clinical interpretation (Riley 2011). We had planned to calculate prediction intervals for all studies to assess heterogeneity and then to explore the impact of risk of bias in subgroup analysis stratified by study risk of bias assessment as detailed below. However, we did not calculate any prediction intervals because all conducted metaanalyses contained fewer than 10 studies. We followed the systematic framework recommended by Page 2019 to assess risk of bias due to missing results (non-reporting bias) in the meta-analysis of pressure ulcer incidence data. To make an overall judgement about risk of bias due to missing results we: • identified whether pressure ulcer incidence data were unavailable by comparing the details of outcomes in trials registers, protocols or statistical analysis plans (if available) with reported results. If the above information sources were unavailable, we compared outcomes in the conference abstracts or in the methods section of the publication, or both, with the reported results. If we found non-reporting of study results, we then judged whether the non-reporting was associated with the nature of findings by using the 'Outcome Reporting Bias In Trials' (ORBIT) system (Kirkham 2018). • assessed the influence of definitely missing pressure ulcer incidence data on meta-analysis. • assessed the likelihood of bias where a study had been conducted but not reported in any form. For this assessment, we considered whether the literature search was comprehensive and planned to produce a funnel plot for meta-analysis for seeking more evidence about the extent of missing results, provided there were at least 10 included studies (Peters 2008; Salanti 2014). However, we did not produce a funnel plot for any meta-analysis because all analyses in this review had fewer than 10 included studies. We summarised the included studies narratively and synthesised included data using meta-analysis where applicable. We structured comparisons according to type of comparator and then by outcomes, ordered by follow-up period. We considered clinical and methodological heterogeneity and undertook pooling when studies appeared appropriately similar in terms of participants, support surfaces and outcome type. Where statistical synthesis of data from more than one study was not possible or considered inappropriate, we conducted a narrative review of eligible studies. Once the decision to pool was made, we used a random-e ects model, which estimated an underlying average treatment e ect from studies. Conducting meta-analysis with a fixed-e ect model in the presence of even minor heterogeneity may provide overly narrow confidence intervals. We used the Chi 2 test and I 2 statistic to quantify heterogeneity but not to guide choice of model for meta-analysis (Borenstein 2009). We exercised caution when metaanalysed data were at risk of small-study e ects because use of a random-e ects model may be unsuitable in this situation. In this case, or where there were other reasons to question the choice of a fixed-e ect or random-e ects model, we assessed the impact of the approach using sensitivity analyses to compare results from alternate models (Thompson 1999). We performed meta-analyses largely using Review Manager 5.4 (Review Manager 2020). We presented data using forest plots where possible. For dichotomous outcomes, we presented the summary estimate as a RR with 95% CI. Where continuous outcomes were measured, we presented the MD with 95% CIs. We planned to report SMD estimates where studies measured the same outcome using di erent methods. For time-to-event data, we presented the summary estimates as HRs with 95% CIs. When important heterogeneity occurred, we planned to follow steps proposed by Cipriani 2013 and Deeks 2019 to investigate further: • check the data extraction and data entry for errors and possible outlying studies; • if outliers existed, perform sensitivity analysis by removing them; and • if heterogeneity was still present, we planned to perform subgroup analyses for study-level characteristics (see below) in order to explain heterogeneity as far as possible. However, we did not undertake any subgroup analysis because metaanalyses in this review included fewer than 10 studies. We investigated heterogeneity using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2019). We planned to perform subgroup analyses for binary and categorical factors (or meta-regression for continuous factors) to determine whether the size of treatment e ects was influenced by these four study-level characteristics: • risk of bias (binary: low or unclear risk of bias; and high risk of bias (Schulz 1995)); • settings (categorical: acute care and other hospital settings; long-term care settings; operating theatre setting; and intensive care unit); • baseline skin status (categorical: participants at risk, of mixed skin status or non-reporting; non-blanchable erythema; existing ulcers of Stage 2 or serious (Shi 2018c)); and • follow-up duration (continuous). We did not perform subgroup analysis or meta-regression when the number of studies included in the meta-analysis was not reasonable (i.e. fewer than 10). We planned to compare subgroup findings using the 'Test for Subgroup Di erences' in Review Manager 5.4 (Review Manager 2020). We conducted sensitivity analyses for the following factors, to assess the robustness of meta-analysis of data on pressure ulcer incidence. • Impact of the selection of pressure ulcer incidence outcome measure. The proportion of participants developing a new pressure ulcer was the primary outcome measure for this review but we also analysed time to pressure ulcer development, where data were available. • Impact of missing data. The primary analysis assumed that participants with missing data did not develop new pressure ulcers. We also analysed pressure ulcer incidence by only Cochrane Database of Systematic Reviews including data for the participants for whom we had endpoint data (complete cases). We noted that when a study only had complete case data (i.e. missing data or the numbers of participants randomised were not reported), complete case data were considered in the related main analysis (see Di erences between protocol and review). • Impact of altering the e ects model used. We used a randome ects model for the main analysis followed by a fixed-e ect analysis. We presented the main, pooled results of the review in 'Summary of findings' tables, which we created using GRADEpro GDT so ware. These tables present key information concerning the certainty of evidence, the magnitude of the e ects of the interventions examined and the sum of available data for the main outcomes (Schünemann 2019). The tables also include an overall grading of the certainty of the evidence associated with each of the main outcomes that we assessed using the GRADE approach. The GRADE approach defines the certainty of a body of evidence as the extent to which one can be confident that an estimate of e ect or association is close to the true quantity of specific interest. The GRADE assessment involves consideration of five factors: within-trial risk of bias, directness of evidence, heterogeneity, precision of e ect estimates, and risk of publication bias (Schünemann 2019). The certainty of evidence can be assessed as being high, moderate, low or very low; RCT evidence has the potential to be high-certainty. We did not downgrade the certainty of evidence for the risk of bias factor in a specific circumstance. That is if the blinding of participants and personnel was the only domain resulting in our judgement of overall high risk of bias for the included studies; however for these studies it was impossible to blind participants and personnel. When downgrading for imprecision, we followed the methods described in Guyatt 2011: either considering both the optimal information size (OIS) and the 95% CI of each meta-analysis if they were estimable; or considering the sample size, the number of events and other e ectiveness indicators if the calculation of OIS and undertaking a meta-analysis were not applicable. Where necessary, we used the GRADE 'default' minimum important di erence values (RR = 1.25 and 0.75) as the thresholds to judge if a 95% CI was wide (imprecise) so as to include the possibility of clinically important harm and benefit (Guyatt 2011) . We presented a separate 'Summary of findings' table for all but two comparisons evaluated in this review. The two exceptions were the comparison of foam surfaces versus another type of foam surface, and the comparison of foam surfaces versus reactive water surfaces; see Di erences between protocol and review. We present these outcomes in the 'Summary of findings' tables: • proportion of participants developing a new pressure ulcer; • time to pressure ulcer development; • support-surface-associated patient comfort; • all reported adverse events; • health-related quality of life; and • cost-e ectiveness. We prioritised the time points and method of outcome measurement specified in Types of outcome measures for presentation in 'Summary of findings' tables. Where we did not pool data for some outcomes within a comparison, we conducted a GRADE assessment for each of these outcomes and presented these assessments in a narrative format in 'Summary of findings' tables (see Di erences between protocol and review). See Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification; Characteristics of ongoing studies. The electronic searches identified 1624 records. including 1164 from electronic databases and 460 from trial registries. We excluded 218 duplicate records and screened 1406 records, of which 233 were identified as potentially eligible and obtained as full-text. Following full-text screening, we considered 42 records of 28 studies eligible for inclusion in this review (Berthe 2007; Bliss 1995a; Bueno de Camargo 2018; Collier 1996; Feuchtinger 2006; Gray 1994; Gray 2000; Gunningberg 2000; Hofman 1994; Hoshowsky 1994; Kemp 1993; Laurent 1998; Nixon 2019; Ozyurek 2015; Park 2017; Rosenthal 2003; Russell 2003a; Santy 1994; Sauvage 2017; Schultz 1999; Stapleton 1986; Takala 1996; Van Leen 2011; Van Leen 2013; Van Leen 2018; Vyhlidal 1997; Whitney 1984; Whittingham 1999 In total we included 29 studies in the review, of which one was an unpublished report (Santy 1994), and two were conference abstracts (Laurent 1998; Whittingham 1999) . See Figure 1 . Cochrane Database of Systematic Reviews Included studies Of the 29 included RCTs, 25 had a parallel group design: 21 with two arms, one with three arms (Stapleton 1986), two with six arms (Santy 1994; Whittingham 1999) , and one with eight arms (Collier 1996) . Four studies had particular design features: • one study appeared to be a multi-arm, multi-stage trial design with eight arms, of which seven were randomised and eligible for this review (Bliss 1995a); • one study was a split body design (that is, it randomly allocated di erent support surfaces to either the right or le half of the body of the same person) and three of its six arms included foam surfaces (Hoshowsky 1994); • one study applied 2 × 2 factorial design (Laurent 1998), including the comparison of foam mattresses versus standard hospital surfaces; and • one study used cross-over design (Van Leen 2013). Of the 29 studies, six were conducted at more than one research site (Kemp 1993; Nixon 2019; Rosenthal 2003; Russell 2003a; Sauvage 2017; Van Leen 2018 (Bliss 1995a; Collier 1996; Gray 1994; Gray 2000; Nixon 2019; Russell 2003a; Santy 1994; Stapleton 1986; Whittingham 1999) and the USA Hoshowsky 1994; Kemp 1993; Rosenthal 2003; Schultz 1999; Vyhlidal 1997; Whitney 1984) . The included studies were published between 1986 and 2018. Of the 26 studies that clearly stated duration of follow-up, the median was 14.5 days (range: 5.0 days to 12.0 months). The 29 included studies enrolled a total of 9566 participants (median study sample size: 101 participants; range: 40 to 2029). The average participant age was specified for 25 studies and ranged between 47.0 and 85.3 years (median: 76 years). The sex of the participants was specified in 24 studies; and within these 2659 (43.4%) of participants were male and 3466 (56.6%) were female. Of the 29 studies, 25 (8601 participants) recruited people at risk of having a new ulcer with risk assessed largely using the Waterlow, Norton or Braden scales. In 21 of these studies, 5512 (64.1%) participants were free of pressure ulcers at baseline. In four studies, 3089 (35.9%) participants with superficial ulcers were enrolled (Bliss 1995a; Nixon 2019; Santy 1994; Whitney 1984) . Two studies (817 participants; Hoshowsky 1994; Laurent 1998) did not specify the skin status at baseline; and two studies (148 participants ; Rosenthal 2003) recruited people with severe fullthickness pressure ulcers alone. Participants were recruited from a variety of settings, including: • a mixture of secondary and community in-patient facilities (n = 2; Kemp 1993; Nixon 2019); • acute care settings (including accident and emergency departments, and hospitals in general) (n = 16; Berthe 2007; Bliss 1995a; Collier 1996; Gray 1994; Gray 2000; Gunningberg 2000; Hofman 1994; Hoshowsky 1994; Laurent 1998; Park 2017; Russell 2003a; Santy 1994; Stapleton 1986; Vyhlidal 1997; Whitney 1984 Cochrane Database of Systematic Reviews The studies investigated a wide range of foam surfaces. Of the 29 studies, 14 described characteristics of foam surfaces used (e.g. foam thickness, foam density, viscoelastic foam; Bueno de Camargo 2018; Collier 1996; Gray 1994; Gray 2000; Gunningberg 2000; Hofman 1994; Laurent 1998; Nixon 2019; Park 2017; Santy 1994; Sauvage 2017; Takala 1996; Vyhlidal 1997; Whittingham 1999) and 15 did not specify the types of foam surfaces they used. Full details of foam surfaces and comparators are listed in Appendix 4 and in results below. Eight studies used comparator group surfaces defined by the study authors as 'standard hospital surfaces' that could not be classified further using the NPIAP S3I support surface terms and definitions (Berthe 2007; Feuchtinger 2006; Gunningberg 2000; Hofman 1994; Laurent 1998; Park 2017; Russell 2003a; Schultz 1999 Twelve studies specified co-interventions they applied (e.g. repositioning, cushions) Bueno de Camargo 2018; Hofman 1994; Ozyurek 2015; Park 2017; Rosenthal 2003; Schultz 1999; Van Leen 2011; Van Leen 2013; Van Leen 2018; Vyhlidal 1997; Whitney 1984) . All twelve stated or indicated that the same cointerventions were applied in all study groups. Of the 29 included studies, 19 specified the details of funding sources. Eleven of these were completely or partly funded by industry or received mattresses under evaluation from industries Bliss 1995a; Bueno de Camargo 2018; Gray 1994; Gray 2000; Gunningberg 2000; Russell 2003a; Schultz 1999; Takala 1996; Van Leen 2018; Vyhlidal 1997) ; four were supported by public funding (Nixon 2019; Ozyurek 2015; Santy 1994; Stapleton 1986) ; one was funded by charity foundations (Kemp 1993) ; and three noted no funding support (Berthe 2007; Laurent 1998; Van Leen 2011) . We excluded 142 studies (with 177 records). The main reasons for these 142 exclusions were: irrelevant and ineligible interventions (55 studies); ineligible study design (e.g. non-RCT, reviews, commentary articles; 52 studies); studies focused on the treatment rather than prevention of pressure ulcers (20 studies); incorrect randomisation and non-randomised methods (eight studies); studies with ineligible outcomes (four studies); clinical trials that were withdrawn (two studies; NCT02634892; NCT02735135); and ineligible participants (healthy subjects; one study). We also identified eight duplicates in screening full texts (see Figure 1 ). We did not identify any ongoing studies. There were six studies (six records) for which we could not make eligibility decisions. For Gardner 2008, we were unable to determine whether the study used foam surfaces. For the remaining five studies, we were unable to obtain the full-texts (in part due to more limited access to intra-library loans during the COVID-19 period) despite extensive e orts (Chaloner 2000b; Henn 2004; Knight 1999; Mastrangelo 2010a; Melland 1998 We judged 12 of the 29 studies to have an unclear overall risk of bias for the primary outcome Berthe 2007; Feuchtinger 2006; Gray 1994; Gray 2000; Gunningberg 2000; Kemp 1993; Rosenthal 2003; Schultz 1999; Van Leen 2011; Van Leen 2013; Vyhlidal 1997) . We judged all the remaining 17 studies as having findings at a high overall risk of bias, of which two had an unclear risk of bias judgements for all domains (Stapleton 1986; Whittingham 1999) , and 15 had one or more domains with a high risk of bias judgement (Bliss 1995a; Bueno de Camargo 2018; Collier 1996; Hofman 1994; Hoshowsky 1994; Laurent 1998; Nixon 2019; Ozyurek 2015; Park 2017; Russell 2003a; Santy 1994; Sauvage 2017; Takala 1996; Van Leen 2018; Whitney 1984) . Of these 15 studies, 10 had a high risk of bias judgement for the primary outcome in the domains of blinding of participants and personnel, blinding of outcome assessment, or both (Bueno de Camargo 2018; Collier 1996; Hofman 1994; Hoshowsky 1994; Laurent 1998; Nixon 2019; Russell 2003a; Sauvage 2017; Takala 1996; Whitney 1984) . We ran a comprehensive search and were able to locate one eligible study from other resources. We considered the risk of having missed published reports to be low. We were unable to assess for the risk of non-publication of studies with negative findings as we could not present funnel plots given the small number of included studies in each analysis. See: Summary of findings 1 Foam surfaces compared with alternating pressure (active) air surfaces for pressure ulcer prevention; Summary of findings 2 Foam surfaces compared with reactive air surfaces for pressure ulcer prevention; Summary of findings 3 Foam surfaces compared with reactive fibre surfaces for pressure ulcer prevention; Summary of findings 4 Foam surfaces compared with reactive gel surfaces for pressure ulcer prevention; Summary of findings 5 Foam surfaces compared with reactive foam and gel surfaces for pressure ulcer prevention See Summary of findings 1; Summary of findings 2; Summary of findings 3; Summary of findings 4; Summary of findings 5. Unless otherwise stated, random-e ects analysis was used throughout. Each pooled result presented is an average e ect, rather than a common e ect and should be interpreted as such. We have not reported data from the nine studies with comparator surfaces that we could not classify in the main body of the results (Berthe 2007; Feuchtinger 2006; Gunningberg 2000; Hofman 1994; Laurent 1998; Park 2017; Russell 2003a; Schultz 1999; Van Leen 2018) . For completeness, we summarise the results of these studies in Appendix 5. We performed data analyses for the following comparisons and outcomes. Where applicable, we performed pre-specified sensitivity analyses as noted in Sensitivity analysis. Four studies (2247 participants) reported data for this outcome that were pooled (Nixon 2019; Rosenthal 2003; Sauvage 2017; Stapleton 1986) . Foam surfaces (117/1122 (10.4%)) may increase the proportion of participants developing incident pressure ulcers compared with alternating pressure (active) air surfaces (83/1125 (7.4%)). The RR is 1.59 (95% CI 0.86 to 2.95; I 2 = 63%; Analysis 1.1). The evidence is of low certainty. Evidence certainty was downgraded once for risk of bias (two studies contributing 50% weight in the meta-analysis had either one domain other than performance bias at high risk of bias or all domains at unclear risk of bias; two studies contributing 50% of weight in the meta-analysis had domains other than performance bias at low or unclear risk of bias), and once for imprecision as, despite the fact that the OIS was met, the wide confidence interval crossed RR = 1.25. We considered the studies included in Analysis 1.1 heterogeneous in terms of all pre-specified subgroup factors (overall 'risk of bias', care settings, skin status at baseline, and follow-up) and there was some indication of statistical heterogeneity (Chi 2 test P value = 0.07; Tau 2 = 0.18; I 2 = 63%). We noticed that, of the four studies, Sauvage 2017 reported a greater treatment e ect than the other three, and that once that study data were removed, I 2 was reduced from 63% to 0% but the overall estimate remained consistent with the main analysis (RR 1.27, 95% CI 0.97 to 1.67; Chi 2 test P value = 0.83; Tau 2 = 0.00; I 2 = 0%). Of the four studies, Sauvage 2017 was di erent from others in terms of care settings: Sauvage 2017 was conducted at long-term care settings whilst others studies were conducted in acute care settings. However, as noted in Subgroup analysis and investigation of heterogeneity, because there were fewer than 10 studies, we did not undertake a subgroup analysis. We performed sensitivity analyses for the following factors but did not use complete case data for sensitivity analysis because the four included studies did not report missing data. • Sensitivity analysis with fixed-e ect (rather than randome ects) model . The use of a fixed-e ect model resulted in a RR of 1.41 (95% CI 1.08 to 1.83; I 2 = 63%). The results suggest that the e ect size of our outcome of interest is sensitive to the type of e ect model chosen and there is a possibility that foam surfaces increase the proportion of participants developing a new pressure ulcer in comparison with alternating pressure (active) air surfaces (Appendix 6). • Post-hoc sensitivity analysis of using pressure ulcer incidence data from Nixon 2019 only . In Analysis 1.1, Nixon 2019 was the largest study (with data for 2029 participants) and was the only study having all domains other than performance bias at low risk of bias for this outcome. Using pressure ulcer incidence data from Nixon 2019 made little di erence to the pooled e ect estimate (RR 1.29, 95% CI 0.96 to 1.74; I 2 = 0%; Appendix 6). • Sensitivity analysis with time to pressure ulcer development as pressure ulcer incidence measure (median follow-up duration 60 days, minimum 30 days, maximum 90 days) . Two studies (2105 participants) reported this outcome measure (Nixon 2019; Sauvage 2017), and these data were pooled. Analysis 1.2 resulted in a HR of 2.46 (95% CI 0.61 to 9.88; I 2 = 86%). It is uncertain whether there is a di erence in the risk of developing a new pressure ulcer, over 60 days' follow-up, between foam surfaces and alternating pressure (active) air surfaces. Evidence is of very low certainty, downgraded once for high risk of bias in one study with 40% of analysis weight, twice for substantial inconsistency, and once for imprecision (Appendix 6). Only Sauvage 2017 (76 participants) reported this outcome, defined by the study authors as the perception of patient comfort and measured using a satisfaction questionnaire. Sauvage 2017 reported no significant di erence in the overall satisfaction between study groups (P = 0.21); no other information was reported. We are uncertain whether there is any di erence between foam surfaces and alternating pressure (active) air surfaces in positive patient comfort responses. Evidence is of very low certainty, downgraded twice for high risk of detection bias, and once for imprecision. Three studies (2181 participants) reported this outcome (Nixon 2019; Rosenthal 2003; Sauvage 2017). We did not pool these data as the definitions of adverse events varied between studies (Table 1) . It is uncertain if there is any di erence in adverse e ects between foam surfaces and alternating pressure (active) air surfaces. Evidence is of very low certainty, downgraded once for unclear risk of bias in two studies and twice for inconsistency. Only Nixon 2019 (2029 participants) reported health-related quality of life, measured using the EQ-5D-5L (with 267 participants only) and PU-QoL-UI (with 233 participants only). It is uncertain if there is a di erence in health-related quality of life (measured using either the EQ-5D-5L or the PU-QoL-UI) at 90 days follow-up in those allocated to foam surfaces or alternating pressure (active) air surfaces (low-certainty evidence). Nixon 2019 reported a MD in the 90-day EQ-5D-5L of 0.00 (95% CI -0.05 to 0.05) between 149 participants using foam surfaces and 118 using alternating pressure (active) air surfaces; and a MD in 90-day PU-QoL-UI of 0.00 (95% CI -0.03 to 0.03) between 126 participants using foam surfaces and 107 using alternating pressure (active) air surfaces (Analysis 1.3). Evidence certainty was downgraded twice for imprecision due to small sample sizes for this outcome. Only Nixon 2019 (2029 participants) reported the incremental cost per quality-adjusted life-year (QALY) gained based on within-trial analyses. Moderate-certainty evidence suggests that alternating pressure (active) air surfaces have a 99% probability of being coste ective at a threshold of GBP 20,000 compared with foam surfaces. Evidence certainty was downgraded once for imprecision for the EQ-5D-5L outcome from which QALY scores were calculated. Four studies (236 participants) compared foam surfaces with reactive air surfaces Takala 1996; Van Leen 2011; Van Leen 2013) . Of these studies, Allman 1987 applied a foam mattress on top of an alternating pressure (active) air surface in comparison with a reactive air surface that had an air-fluidised feature. Proportion of participants developing a new pressure ulcer (follow-up duration minimum 13 days, maximum six months) All four studies (236 participants) reported this outcome and the data of 229 participants were available for analysis. Foam surfaces (32/116 (27.6%)) may increase the proportion of participants developing a new pressure ulcer compared with reactive air surfaces (12/113 (10.6%); low-certainty evidence). The RR is 2.40 (95% CI 1.04 to 5.54; I 2 = 25%; Analysis 2.1). Evidence certainty was downgraded once for risk of bias (one study contributing 8% weight in the meta-analysis had domains other than performance bias at high risk of bias and all the remaining studies had domains other than performance bias at low or unclear risk of bias) and once for imprecision as, despite the fact that the OIS was met, the 95% CI crossed RR = 1.25. The included studies did not report data on time to pressure ulcer incidence. We considered the studies in Analysis 2.1 heterogenous in terms of follow-up durations, care settings, and overall 'risk of bias' and there was an indication of small statistical heterogeneity (Chi 2 test P = 0.26; Tau 2 = 0.21; I 2 = 25%). We did not perform any prespecified subgroup analysis because, as noted in Subgroup analysis and investigation of heterogeneity, the number of included studies was fewer than 10, meaning it would be di icult to meaningfully interpret the results. • Sensitivity analysis with fixed-e ect (rather than randome ects) model . The use of a fixed-e ect model resulted in a RR of 2.47 (95% CI 1.40 to 4.38; I 2 = 25%). The result remained consistent with the main analysis (Appendix 6). Support-surface-associated patient comfort (follow-up duration 13 days) Only Allman 1987 (72 participants) reported this outcome in which participants were asked to choose a response to a comfortrelated question from categories: 'Very comfortable', 'Comfortable', 'Uncomfortable', or 'Very uncomfortable'. It is uncertain if there is a di erence in patient comfort responses between those using foam surfaces on top of an alternating pressure (active) air surface and those using reactive air surfaces (P = 0.04; very low-certainty evidence). Evidence certainty was downgraded once for unclear risk of bias, and twice for imprecision due to the small sample size. Only Allman 1987 (72 participants) reported this outcome (see Table 1 ). It is uncertain if there is a di erence in adverse events between foam surfaces and reactive air surfaces (very low-certainty evidence). Evidence certainty was downgraded once for unclear risk of bias, and twice for imprecision due to the small sample size. Not reported. Not reported. Bliss 1995a and Stapleton 1986 compared foam surfaces with reactive fibre surfaces. Bliss 1995a had no outcomes directly relevant to this review and so none of the data were analysable. Stapleton 1986 (68 participants) reported data for this outcome. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces (14/34 (41.2%)) and reactive fibre surfaces (12/34 (35.3%)). The RR is 1.17 (95% CI 0.64 to 2.14; Analysis 3.1). The evidence is of very low certainty, downgraded twice for unclear risk of bias in all domains, and twice for imprecision as the OIS was not met and the wide 95% CI crossed RRs = 0.75 and 1.25, failing to exclude important benefits or harms. The included study did not report data on time to pressure ulcer incidence. None reported. Hoshowsky 1994 was a study with a split body design. It compared foam surfaces with two study arms that both applied reactive gel surfaces on top of another type of surface. We combined these into a single study arm. Hoshowsky 1994 (135 participants) reported this outcome but indicated that no pressure ulcers developed. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces and reactive gel surfaces. The evidence is of very low certainty, downgraded twice for high Cochrane Database of Systematic Reviews risk of bias in domains other than performance bias, and twice for imprecision due to the small sample size and the low event rate. The included study did not report data on time to pressure ulcer incidence. None reported. Comparison 5: Foam surfaces versus reactive foam and gel surfaces (one study, 91 participants) Using a split body design, Hoshowsky 1994 compared foam surfaces with reactive foam and gel surfaces. Proportion of participants developing a new pressure ulcer (follow-up duration unspecified) Hoshowsky 1994 (91 participants) reported this outcome but indicated that no pressure ulcers developed. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces and reactive foam and gel surfaces. The evidence is of very low certainty, downgraded twice for high risk of bias in domains other than performance bias, and twice for imprecision due to the small sample size and the low event rate. The included study did not report data on time to pressure ulcer incidence. None reported. Bliss 1995a compared foam surfaces with reactive water surfaces but reported no outcomes directly relevant to this review and so none of the data were analysable. Nine studies compared two di erent types of foam surface (Bueno de Camargo 2018; Collier 1996; Gray 1994; Gray 2000; Kemp 1993; Ozyurek 2015; Santy 1994; Vyhlidal 1997; Whittingham 1999) . Of these, two studies compared six types of foam surfaces (Santy 1994; Whittingham 1999) , and one included eight foam surfaces (Collier 1996). We did not pool data from the nine studies as it was not possible to interpret this as a single comparison. We summarised study findings narratively below with key outcome data presented in Table 2 and Table 3 . Proportion of participants developing a new pressure ulcer (follow-up duration minimum 10 days, maximum 12 months or unspecified) Six studies (733 participants) reported data for this outcome (Bueno de Camargo 2018; Collier 1996; Gray 2000; Kemp 1993; Ozyurek 2015; Vyhlidal 1997 ; see Table 2 ). Overall, it is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between the two types of foam surface. Evidence is of very low certainty, downgraded once for risk of bias (three studies contributing half the data for this outcome were at high risk of bias and the remaining three studies were at unclear risk of bias in at least one domain), twice for substantial inconsistency that we could not explain, and once for imprecision as the sample sizes were small for all six studies. Two studies (146 participants) reported time to pressure ulcer development (follow-up duration 11.5 days and one month). Bueno de Camargo 2018 (62 participants) reported an unadjusted HR of 0.33 (95% CI 0.17 to 0.64) for a comparison of viscoelastic foam surfaces with a density of 40 to 60 kg/m 3 versus foam surfaces with a density of 33 kg/m 3 in an intensive care unit setting. Kemp 1993 (84 participants) reported an adjusted HR of 0.40 (95% CI 0.20 to 0.80) for a comparison of solid foam surfaces versus convoluted foam surfaces at acute care and long-term care settings. See Table 2 . Overall, low-certainty evidence suggests that viscoelastic foam surfaces with a density of 40 to 60 kg/m 3 and solid foam surfaces may decrease the risk of developing incident pressure ulcers at any point over one month's follow-up compared with the control foam surfaces. Evidence certainty was downgraded once for risk of bias (the two studies were at either high or unclear risk of bias) and once for imprecision as both studies were very small. Support-surface-associated patient comfort (follow-up duration minimum 10 days, maximum 12 months or unspecified) Four studies (669 participants) reported this outcome (Collier 1996; Gray 1994; Gray 2000; Whittingham 1999) . The studies report a range of di erent measures and outcome data cannot be easily interpreted (see Table 3 ). We are uncertain if there is a di erence in positive patient comfort responses between di erent types of foam surface. Evidence is of very low certainty, downgraded once for risk of bias (two studies were at high risk of bias and another two studies were at unclear risk of bias), twice for substantial inconsistency, and once for imprecision due to small sample sizes in these studies. Not reported. Not reported. Not reported. We report evidence from 29 RCTs on the e ects of foam surfaces compared with any alternative support surface on the incidence of pressure ulcers in any population in any setting. We did not analyse data reported in the nine studies that compared foam surfaces with surfaces that could not be classified. We analysed data for seven comparisons in the review and we summarise key findings for these comparisons below. Trusted evidence. Informed decisions. Better health. Cochrane Database of Systematic Reviews alternating pressure (active) air surfaces (four studies with 2247 participants; low-certainty evidence). It is uncertain whether there is any di erence in support-surface-associated patient comfort between these types of support surfaces (one study; 76 participants), as well as in the number of all reported adverse events (three studies; 2181 participants). It is uncertain if there is a di erence in health-related quality of life (measured using either the EQ-5D-5L or the PU-QoL-UI) at 90 days' followup between these surfaces (one study with 2029 participants; low-certainty evidence). We found moderate-certainty coste ectiveness evidence that alternating pressure (active) air surfaces are probably more cost-e ective than foam surfaces. • Foam surfaces versus reactive air surfaces. Foam surfaces may increase the proportion of participants developing a new pressure ulcer compared with reactive air surfaces (four studies with 229 participants; low-certainty evidence). It is uncertain if there is a di erence in patient comfort responses and in adverse event rates between people using reactive air surfaces and those using foam surfaces on top of alternating pressure (active) air surfaces (one study with 72 participants; very lowcertainty evidence). • Foam surfaces versus reactive fibre surfaces. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces with reactive fibre surfaces (one study with 68 participants). there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces and reactive gel surfaces (one study with 135 participants). • Foam surfaces versus reactive foam and gel surfaces. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between foam surfaces and reactive gel surfaces (one study with 91 participants). • Foam surfaces versus reactive water surfaces. There are no analysable data for this comparison. • Foam surfaces versus another type of foam surface. It is uncertain if there is a di erence in the proportion of participants developing a new pressure ulcer between di erent types of foam surfaces (six studies with 733 participants). When we considered time to pressure ulcer incidence as our primary outcome, we found that viscoelastic foam surfaces with a density of 40 to 60 kg/m 3 may decrease the risk of developing incident pressure ulcers at time points up to 11.5 days' follow-up compared with foam surfaces with a density of 33 kg/m 3 . Solid foam surfaces may also decrease the risk of developing incident pressure ulcers at time points up to one month's follow-up compared with convoluted foam surfaces. It is also uncertain if there is a di erence in support-surface-associated patient comfort between di erent types of foam surface (four studies with 669 participants). As detailed in Search methods for identification of studies, we ran a comprehensive set of literature searches to maximise the relevant research included here. Whilst use of foam surfaces is relevant to adults and children in any settings, all participants in the included studies were adults (with the reported average age ranging from 47 to 85.3 years, median of 76 years). Across the included studies, more than half (56.6%) of enrolled participants were female. Almost all of the studies enrolled people who were at (high) risk of pressure ulceration, with risk assessed using a risk assessment tool (e.g. the Braden scale), and who were ulcer-free at the time of recruitment. Four included studies (with 3089 participants) did include participants with superficial pressure ulcers at baseline. Most of the included studies were small (half had fewer than 100 participants), whilst eleven studies enrolled more than 200 participants, and seven studies more than 400. These seven trials together accounted for 71.6% (6853/9566) of the participants in the review. The geographical scope of included studies was limited. Almost all the studies were from Europe and North America. One small study was from South Korea (Park 2017), and one small study was from Turkey (Ozyurek 2015). The included studies recruited participants from a variety of care settings including: acute care settings (16 studies), community and long-term care settings (six studies), or both (two studies); intensive care units (three studies); and operating room (two studies). Whilst three of the seven comparisons included studies from a variety of care settings, due to a limited number of included studies for these three comparisons we could not perform pre-specified subgroup analysis by di erent care settings. Thus, for these comparisons, we are unable to drawn conclusions about potential modification of treatment e ects in di erent care settings. The remaining four comparisons included data that were only from either acute care settings or nursing home settings and almost all of these four comparisons only included one study. Therefore, their evidence is very limited. These comparisons are foam surfaces compared with reactive water surfaces, reactive fibre surfaces, reactive gel surfaces, or reactive foam and gel surfaces. Additionally, the included data were limited for intensive care units and operating rooms. We recognise that foam surfaces have evolved over decades and can have a range of features (e.g. foam density, foam thickness, layers of foam). The included studies were published from 1986 to 2018, and the specific foam surface types applied in the studies inevitably varied (see Appendix 4). In this review, we considered all specific foam types as foam surfaces because they have the same underlying mechanism of redistributing pressure activity (i.e. distributing the pressure over a greater area via immersion and envelopment). We did not analyse data reported in the nine studies that compared foam surfaces with surfaces that we could not define using the NPIAP S3I 2007 support surfaces terms and definitions. However, for completeness of all relevant evidence, we reported the data of these studies in Appendix 5. Another limitation in the included studies was the large variation in terms of follow-up durations (with a range from five days to 12 months, median of 14.5 days). This is partly because di erent follow-up durations are appropriate in di erent care settings. For example, participants staying at acute care settings are more likely to be discharged a er a short-term hospital stay whilst those staying at community and long-term care settings will typically stay for longer. The short median duration of follow-up may contribute to an under-estimation of pressure ulcer incidence across study groups of the included studies because most pressure ulcers would occur in the first two to four weeks a er hospital admission Cochrane Database of Systematic Reviews (Schoonhoven 2007) , and some incident pressure ulcers may have been missed in these studies. We implemented the GRADE approach for assessing the certainty of the evidence and found that most of the included evidence from our 15 meta-analyses or syntheses across seven comparisons was of low or very low certainty. Downgrading of evidence was largely due to the high risk of bias of findings, and imprecision due to small study sizes in terms of participants or event numbers, or both. There was also some inconsistency across studies for some comparisons. We downgraded once or twice for study limitations for almost all evidence. We assessed risk of bias according to seven domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, selective outcome reporting, incomplete follow-up, and other potential biases. Of the 29 studies, we judged 17 as being at high overall risk of bias; and 12 at unclear overall risk of bias. The prevalence of high overall risk of bias is partly due to the nonblinding of participants and personnel for most comparisons. We acknowledged that such blinding of participants and personnel is impractical for most comparisons. Therefore, we did not downgrade certainty of evidence for studies at high overall risk of bias solely due to the possible presence of performance bias. Nine studies were also at high risk of bias due to unblinded outcome assessment. Unblinded assessment has been found to exaggerate odds ratios (from subjective binary outcomes) by, on average, 36% (Hróbjartsson 2012). The outcome assessment of pressure ulcer incidence is subjective and blinded assessment -whilst operationally challenging -can be undertaken (for example, through masked adjudication of photographs of pressure areas; Baumgarten 2009). Therefore, we considered unblinded pressure ulcer incidence assessment could substantially bias e ect estimates in the included studies and downgraded the certainty of evidence for detection bias on a study-by-study basis. We did not downgrade for all evidence. This was because we considered that the participants, interventions, and outcomes in the included studies were within the scope of the published review protocol and there was no indirectness. Statistical heterogeneity was low for most of the evidence synthesis (11/15) we performed and we did not downgrade for inconsistency for these pieces of evidence. The low statistical heterogeneity was partly because eight of the 11 syntheses included only one study. We downgraded for inconsistency for the rest of the meta-analyses or narrative syntheses. None of these four analyses included more than six studies. Despite the fact that we found heterogeneity in terms of overall risk of bias, care settings, outcome measurement methods, or follow-up durations between the included studies, we did not investigate their heterogeneity using subgroup analysis and we considered their heterogeneity (inconsistency) unexplained. We have to note that although we had planned to calculate prediction intervals to understand the implications of heterogeneity, all analyses included a small number (up to seven) of included studies, which was fewer than the 10 needed for this calculation. We downgraded once or twice for imprecision for 14 of 15 syntheses. Study sample sizes are small in most cases (median sample size: 101; range: 40 to 2029) with o en small numbers of events and wide associated confidence intervals around e ect estimates. Confidence intervals o en crossed the line of null e ect, thus meaning we could not discern whether the true population e ect was likely to be beneficial or harmful. We did not downgrade the certainty of evidence for publication bias in all meta-analyses. This is because (1) we have confidence in the comprehensiveness of our literature searches; and (2) we did not find any clear evidence of non-reporting bias of study results. Although we planned to perform funnel plots for meta-analysis to visually inspect for publication bias, there was no analysis including more than ten studies. We followed pre-specified methods to review evidence in order to prevent potential bias in the review process. For example, we ran comprehensive electronic searches, searched trials registries, and checked references of systematic reviews identified in electronic searches. This review also has limitations. Firstly, some included studies may have considered co-interventions as 'usual care' but did not fully describe them. We assumed that all studies had provided cointerventions equally to participants in their study groups if there was nothing to indicate that this was not the case. Secondly, we did not implement pre-specified subgroup analysis, as mentioned above, mainly because no analysis included more than 10 studies. Thirdly, we included a factorial design study in this review (Laurent 1998), but did not consider the potential interaction between interventions. Fourthly, only Nixon 2019 fully reported HRs and CIs related to time-to-event data. The HR and CI for Sauvage 2017 we used in Analysis 1.2 were calculated using the methods described in Tierney 2007; we recognised those calculated data (and associated meta-analyses) might be inaccurate. We noted that the time-to-event data analysis using the HR and CI we calculated tended to agree with the associated binary data analysis (Analysis 1.1) as we expected. Fi hly, eight studies described their controls as 'standard hospital surfaces' but did not specify construction materials of these surfaces. Although we made e orts to collect information on these surfaces, we were not able to classify them. Traditionally, 'standard hospital surfaces' meant foam surfaces, but we felt adopting that assumption was unwarranted. Thus, we did not classify them as foam surfaces and we did not perform any analysis for the comparison of di erent types of foam surface. Finally, we were not able to pre-specify the comparisons included in this review. This is because specific support surfaces applied could only be known and defined once eligible studies were included. However, we pre-planned to use the NPIAP S3I 2007 support surface terms and definitions to define specific support surfaces in order to avoid any potential bias. To our knowledge, among the 14 systematic reviews or metaanalyses we identified in the electronic searches for this review ( This review di ers from Shi 2018a and McInnes 2015 in how specific support surfaces (including foam surfaces) are classified and labelled. As mentioned above, the types of foam surface used in the included studies varied, and we labelled all these types as 'foam surfaces'. However, Shi 2018a and McInnes 2015 used the term 'high specification foam' surfaces. Whilst this term is used in pressure ulcer guidelines and there is an Australian consensus on characteristics that constitute a high specification foam mattress (e.g. foam density, thickness), it has been deprecated by the NPIAP S3I. NPIAP S3I 2007 noted that the term 'high specification foam surfaces' "potentially limits clinical options because it is based on materials not system performance characteristics". Additionally, the characteristics of foam surfaces used in the included studies were not always given (see Appendix 4). Some studies specified the foam density of foam surfaces whilst others only specified thickness and foam materials (e.g. viscoelastic foam, or polyurethane foam). It is inappropriate to group all specific foam surfaces across studies as high specification foam surfaces. In terms of the included comparators, Shi 2018a considered reactive air-fluidised surfaces, reactive air surfaces and reactive low-air-loss surfaces as separate groups whilst we considered them a single generic group, 'reactive air surfaces'. Likewise, Shi 2018a considered alternating pressure (active) low-air-loss surfaces, alternating pressure (active) air surfaces, and hybrid air surfaces as separate groups whilst we considered them a single generic group, 'alternating pressure (active) air surfaces'. Shi 2018a grouped some interventions under the term 'standard hospital surfaces' but concluded that the types of surfaces labelled in this way varied over time, and by setting. McInnes 2015 applied the terms 'standard hospital foam' and 'standard hospital mattresses' in one specific comparison. We noted that the NPIAP S3I 2007 recommends that the term 'standard hospital surfaces' should be avoided for use and the surface characteristics should be specified. In this review, we made great e orts to define surfaces where these surfaces were described as a 'standard hospital surface' in the included studies to ensure they were placed in the correct comparisons. We classified 'standard hospital surfaces' that had no characteristic details or could not fit the NPIAP S3I 2007 support surfaces terms and definitions as undefined surfaces. These above re-definitions and re-classifications of specific support surfaces can explain some of the inconsistency between these reviews, but importantly, Shi 2018a was a network meta-analysis. Shi 2018a considered pressure ulcer incidence and supportsurface-associated patient comfort outcomes only whilst this review adds cost-e ectiveness evidence to the evidence base and suggests that alternating pressure (active) air surfaces are probably more cost-e ective than foam surfaces. Shi 2018a indicated an evidence gap around the comparison of foam surfaces versus alternating pressure (active) air surfaces, and expected to tackle this gap by including a large, then ongoing study -Nixon 2019 -in data analysis. This review did include this study, but this inclusion still resulted in some uncertain evidence with the use of pairwise meta-analysis methods. Further planned review work using network meta-analysis will add to the findings reported here. McInnes 2015 suggested that the so-called 'high specification foam mattress' can reduce pressure ulcer incidence compared with standard hospital surfaces. We did not perform any analysis for the comparison of foam surfaces versus 'standard hospital surfaces'. The current evidence base is full of uncertainties about the di erence in pressure ulcer incidence between using foam surfaces and some other surfaces (i.e. reactive fibre surfaces, reactive gel surfaces, reactive foam and gel surfaces, or reactive water surfaces). Foam surfaces may increase the risk of pressure ulcer development in comparisons with alternating pressure (active) air surfaces and with reactive air surfaces. Alternating pressure (active) air surfaces are probably more cost-e ective than foam surfaces. When considering di erent types of foam surface, viscoelastic foam surfaces with a density of 40 to 60 kg/m 3 may reduce the risk of developing incident ulcers over 11.5 days' follow-up compared with foam surfaces with a density of 33 kg/m 3 in people treated in the intensive care unit setting. Solid foam surfaces may also reduce the risk of developing pressure ulcers over one month's follow-up compared with convoluted foam surfaces in people treated in acute care and long-term care settings. Given the large number of di erent support surfaces available, future studies should prioritise which support surfaces to evaluate on the basis of the priorities of decision-makers. For example, foam surfaces versus reactive gel surfaces may be a high priority for future evaluation, particularly in operating rooms. All interventions used should be clearly described using the current classification system. Researchers should avoid the use of some terms such as 'high specification foam surfaces' and 'standard hospital surfaces' without further detail about the specific nature of the support surfaces being evaluated. Limitations in included studies are largely due to small sample size and sub-optimal RCT design. The incidence of pressure ulcers can be low in certain settings and this needs to be considered in sample size calculations and when considering the feasibility of trial conduct. Under-recruitment or over-estimation of event rates that then fail to occur, or both, can lead to imprecision and less robust e ect estimates. Future studies should also consider carefully the choice of outcomes they report. Time-to-event data for pressure ulcer incidence should be used in studies. Careful and consistent assessment and reporting of adverse events needs to be undertaken to generate meaningful data that can be compared between studies. Likewise, patient comfort is an important outcome but poorly defined and reported, and this needs to be considered in future research studies. Further studies should aim to collect and report health-related quality of life using validated Any future studies must be undertaken to the highest standard possible. Whilst it is challenging to avoid the risk of performance bias in trials of support surfaces as blinding of participants and personnel is seldom possible, stringent protocols -for example, in terms of encouraging consistent care and blinded decision-making -can help to minimise risk. It is also important to fully describe co-interventions (e.g. repositioning) and ensure protocols mandate balanced use of these across trial arms. The risk of detection bias can also be minimised with the use of digital photography and adjudicators of the photographs being masked to support surfaces (Baumgarten 2009). Follow-up periods should be for as long as possible and clinically relevant in di erent settings. Where possible and useful, data collection a er discharge from acute settings may be considered. The authors are grateful to the following peer reviewers who provided feedback on both the protocol and the review: Julie Bruce and Zena Moore. Thanks are also due to Jessica Sharp for copyediting the protocol, to Denise Mitchell for additional copy-edit feedback, to Faith Armitage for copy-editing the review and to Nicole Pitcher for writing the Plain Language Summary. We also would like to thank Asmara Jammali-Blasi for screening search records; and to thank Zhenmi Liu, Gill Norman, and Melanie Stephens for double-checking data extraction and risk of bias assessment for this review. Thanks also to Cochrane Musculoskeletal, Oral, Skin and Sensory Network Editors Peter Tugwell and Jennifer Hilgart for feedback and final approval of the review for publication. Shi C, Westby M, Norman G, Dumville J, Cullum N. Node-making processes in network meta-analysis of non-pharmacological interventions should be well planned and reported. Journal of Clinical Epidemiology 2018;101:124-5. Shi C, Dumville JC, Cullum N. Skin status for predicting pressure ulcer development: a systematic review and meta-analyses. : patients with change in comfort from baseline. Level of comfort assessed by asking the patient to respond to a second question scored from 1 to 4: "Which of the following best describes the bed you are using here in the hospital: very comfortable, comfortable, uncomfortable, or very uncomfortable?" • Dropouts and reasons: 7 withdrew prior to follow-up and excluded from analysis (6 died, 1 withdrew due to nausea and dislike of the air-fluidised bed) • Data and results: 8 comfort increased, 4 no change and 1 decreased on air-fluidised bed; 3 increased, 4 no change and 6 decreased on conventional therapy (P = 0.04) • Notes (e.g. other results reported): All reported adverse events Cochrane Database of Systematic Reviews • Dropouts and reasons: 7 withdrew prior to follow-up and excluded from analysis (6 died, 1 withdrew due to nausea and dislike of the air-fluidised bed) • Data and results: 8 died, 2 pneumonia, 10 urinary tract infections, 6 hypotension, 5 hypernatraemia, 5 oliguria, 7 sepsis, 16 fever, and 3 heart failure on air-fluidised bed; 7 died, 4 pneumonia, 7 urinary tract infections, 7 hypotension, 5 hypernatraemia, 8 oliguria, 6 sepsis, 22 fever, and 6 heart failure on conventional therapy • Notes (e.g. other results reported): some patients appeared to have multiple adverse events • Not reported • Not reported Outcomes that are not considered in this review but reported in trials: • Ulcer healing • Change in total surface area • Patients improved • 50% reduction in total surface area • Pain response Random sequence generation (selection bias) Low risk Quote: "Patients were randomly allocated to treatment groups in two strata in balanced blocks of six with stratification … The randomization sequence was determined using a table of random numbers …" Comment: low risk of bias due to the use of a proper randomisation method. Unclear risk Quote: "… treatment allocations were placed in envelopes sealed and numbered sequentially. After establishing eligibility, one of the investigators selected the unopened envelope with the lowest number in the appropriate strata and allocated the patient to the treatment indicated on the enclosed card" Comment: unclear risk of bias because it is unclear if the envelopes are opaque. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Comment: low risk of bias because of the low rate of attrition (7/72, 9.7%). Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to determine the effectiveness in pressure-sore prevention of an interface pressure-decreasing mattress, the Kliniplot® mattress Outcomes that are not considered in this review but reported in trials: • None Random sequence generation (selection bias) Unclear risk Quote: "... were freely assigned to a bed which has been randomly equipped in advance either with a Kliniplot® mattress, or with a standard mattress" Comment: unclear risk of bias because the sequence generation method is not specified. Cochrane Database of Systematic Reviews Allocation concealment (selection bias) Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Unclear risk Comment: no information provided. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to identify inexpensive and, if possible, non-mechanical constant low pressure overlays effective for patients at long-term risk in continuing-care wards for elderly people Study design: randomised controlled trial (a poorly designed multi-arm multi-stage trial, with re-randomisation) Duration of follow-up: not given; assessment with a mean of 17.7 days Number of arms: 7 (the trial had a Vaperm as control arm but its participants were not randomised. Vaperm data were not extracted for this review) Single centre or multi-sites: not specified Study start date and end date: not described Outcomes that are not considered in this review but reported in trials: • None Random sequence generation (selection bias) Low risk Quote: "the patient was randomly allocated to an experimental overlay by the researcher writing the names of all those available at the time on slips of paper which were folded and offered to the nurse to choose one blind" Comment: low risk of bias because drawing of lots is applied to generate random sequence. Foam surfaces for preventing pressure ulcers ( Allocation concealment (selection bias) High risk Quote: "the patient was randomly allocated to an experimental overlay by the researcher writing the names of all those available at the time on slips of paper which were folded and offered to the nurse to choose one blind. The designated overlay was then placed on the bed" Comment: high risk of bias because it appears difficult to conceal the allocation process as the authors described. The nurse would have knowledge of which overlays were available at the time of consent. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Unclear risk Comment: no information provided. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias High risk Comment: high risk of bias because some individuals may be repeatedly observed and included in analysis (i.e. correlation issue in analysis). For example, Bliss stated "there were no written criteria determining the decision to stop a trial [i.e. using an overlay as the experimental intervention]. This depended mainly on these experienced nurses' unwillingness to allow it to continue because of enlargement of an existing sore, a new blister, discolouration, oedema ... Patients who developed pressure damage between assessments might also be taken o their overlay ... if they later improved ... they were re-randomized for another trial period [i.e. comparisons of new overlays]." Additionally, overlays were observed for unequally periods of time. Treatments were discontinued or introduced without prespecified stopping rules. Some comparisons are not parallel. Methods Study objective: to analyse whether a viscoelastic mattress support surface can reduce the incidence of stage 2 pressure injuries compared to a standard hospital mattress with pyramidal overlay in critically ill patients The standard hospital mattress is a 12-centimetre cold foam with a density of 33 measuring 188 by 80 centimetres. The pyramidal overlay is a 5-centimetre layer of polyurethane foam, density 33, whose surface looks like egg carton. Cochrane Database of Systematic Reviews • Definition (including ulcer stage): incidence of stage 2 pressure injuries: partial-thickness loss of skin with exposed dermis • Dropouts: intention-to-treat (ITT) analysis • Notes (e.g. other results reported): ulcers occurred in 35 patients; higher in pyramidal overlay (25 of 31; 80.6%) compared to viscoelastic foam mattress (10 of 31; 32.2%) P < 0.001 Time to pressure ulcer incidence Outcomes that are not considered in this review but reported in trials: • Mortality rate (mentioned in NCT02844166 but not reported in the study's paper). Random sequence generation (selection bias) Low risk Quote: "Randomization was performed using a computerized table, and patients were allocated into two groups" Comment: low risk of bias because a proper randomisation method used. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Quote: "the blinding of the health team was not possible" Comment: high risk of bias as the authors stated no blinding. Blinding of outcome assessment (detection bias) All outcomes Quote: "the blinding of the health team was not possible" Cochrane Database of Systematic Reviews Comment: high risk of bias as the authors stated no blinding. Incomplete outcome data (attrition bias) All outcomes Comment: low risk of bias because the paper clearly states ITT analysis performed. Selective reporting (reporting bias) High risk Comment: high risk of bias because even though the study protocol is available but it is clear that the published report does not include mortality outcome that was pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to compare 8 new foam mattresses with a new standard 180 mm hospital mattress, and to define their ability to reduce the incidence of pressure sore formation and to provide comfort sual rating scale (1 = poor, 10 = excellent) • Definition: not described • Dropouts: not described • Notes: range of patient comfort assessments 5 to 7 in Clinifloat (n = 11); 0 to 0 in NHS Standard (n = 9); 3 to 8 in Omnifoam (n = 11); 8 to 11 in Softform (n = 12); 9 to 9 in STM5 (n = 10); 8 to 8 in Therarest (n = 13); 2 to 8 in Transfoam (n = 10); 10 to 10 in Vapourlux (n = 14) All reported adverse events using allocated support surfaces Blinding of outcome assessment (detection bias) All outcomes Quote: "Patients were periodically reassessed ... and any evidence of skin deterioration was documented ... conducted at least weekly throughout their period in hospital" Comment: high risk of bias for both pressure ulcer and comfort outcomes because it is unlikely that blinding was implemented for participants and personnel given the information provided. Self-reported comfort outcome cannot be measured in a blinded way. Incomplete outcome data (attrition bias) All outcomes Comment: no attrition identified. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to assess the effect of a 4 cm thermoactive viscoelastic foam overlay with a water-filled warming mattress on the operating room-table compared with the standard operating roomtable (a water-filled warming mattress, no pressure-reducing device) on the postoperative pressure ulcer incidence in cardiac surgery patients. Cochrane Database of Systematic Reviews Comment: low risk of bias because ITT analysis was conducted. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to fully evaluate and define the clinical abilities of the standard 130 mm contract mattress and the Softform mattress in regards to their ability to provide the patient with adequate pressure reduction, so as to prevent pressure sore formation, and provide the patient with adequate comfort. Cochrane Database of Systematic Reviews Random sequence generation (selection bias) Unclear risk Quote: "Allocation of mattresses was by patient randomisation on admission ... randomly allocated to one of the two types of mattress using unmarked envelopes" Comment: unclear if a proper randomisation method was applied. Allocation concealment (selection bias) Unclear risk Quote: "randomly allocated to one of the two types of mattress using unmarked envelopes" Comment: unclear if allocation was appropriately concealed. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Comment: no information provided. Comment: high risk of bias because it is unlikely that patients who self-reported their comfort responses are blinded. Incomplete outcome data (attrition bias) All outcomes Quote: "A number of patients were excluded from the study because the Waterlow score awarded by the ward sta differed greatly from that of the researcher" Comment: unclear risk of bias because the number of exclusions is unclear and unclear if this exclusion was post-randomisation. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to evaluate the ability of 2 pressure-reducing mattresses to prevent pressure sores in a population who were deemed to be at high risk of sore development. Random sequence generation (selection bias) Unclear risk Quote: "Individuals who met the entry criteria were randomised to a control or trial mattress using an opaque envelope" Comment: unclear if a proper randomisation method was applied. Unclear risk Quote: "Individuals who met the entry criteria were randomised to a control or trial mattress using an opaque envelope" Comment: unclear if allocation was concealed. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Quote: "Tissue damage was assessed by sta who were unaware which mattress the subject was using" Comment: low risk of bias because blinded outcome assessors were used for the comparison of 2 foam mattresses. This blinding is feasible. Comment: high risk of bias because it is unlikely that it was possible to blind patient self-reported outcome assessment. Cochrane Database of Systematic Reviews All outcomes Comment: unclear risk of bias because the number of individuals with data observed was not specified. Comment: low risk of bias because in total 5 of 100 missed. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Study objective: to investigate if viscoelastic foam mattresses are more effective than standard hospital mattresses in reducing the incidence of pressure ulcers in patients with hip fractures. Cochrane Database of Systematic Reviews Random sequence generation (selection bias) Unclear risk Quote: "On arrival in A&E patients with a suspected hip fracture were randomised to an experimental or a control group with concealed allocation" Comment: the method of randomisation was not reported. Unclear risk Quote: "On arrival in A&E patients with a suspected hip fracture were randomised to an experimental or a control group with concealed allocation" Comment: the method of concealing allocation was not reported. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Quote: "The pressure ulcer nurse on the ward usually performed the assessments on the fourth postoperative day and at discharge. The pressure ulcers were photographed ... The ulcers in these photos were graded by an expert nurse ... who was blinded to treatment, and compared with the classifications performed by the nurses in A&E and on the wards ... an excellent agreement" Comment: low risk of bias because the expert nurse who was blinded to treatment had assessments consistent with the ward nurses, meaning ward nurses' outcome assessment was unlikely to be influenced by treatment. Comment: high risk of bias because it is impossible to blind patients to self-reported outcome measure. Incomplete outcome data (attrition bias) All outcomes Comment: no missing data. Quote: "Forty-one patients (21 in the experimental and 20 in the control group) with a mean age of 84 years (SD: 7.6, 67-102) answered this question" Comment: high risk of bias because 27 of 48 in viscoelastic foam group and 33 of 53 in standard hospital mattress group missed. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Foam surfaces for preventing pressure ulcers ( Methods Study objective: to determine the effectiveness in pressure-sore prevention of the DeCube mattress versus standard mattress in patients with a femoral-neck fracture and a concomitant high risk for the development of pressure sores. Outcomes that are not considered in this review but reported in trials: • Pressure ulcer incidence by sacrum, trochanters, shoulders, le hip fracture and right hip fracture (reported by authors but not extracted) Random sequence generation (selection bias) Unclear risk Quote: "Each group of 6 consecutively admitted patients was randomly divided into 3 patients nursed preoperatively and postoperatively on the standard Vredestein polyproleen [polypropylene] SG 40 hospital mattress (Vredestein, Netherlands) and 3 nursed on the Comfortex DeCube" Comment: the method of randomisation was not described. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Quote: "The study was not blinded with respect to observer or nurse" Comment: high risk of bias because clearly blinding was not implemented. Blinding of outcome assessment (detection bias) All outcomes Quote: "The study was not blinded with respect to observer or nurse" Comment: high risk of bias because clearly blinding was not implemented. Incomplete outcome data (attrition bias) All outcomes Comment: high risk of bias because 10 of 46 individuals missed at 2 weeks. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Unclear risk Comment: the study appears to have been stopped early. It is not clear whether this interim analysis was pre-planned in advance of data collectionthe sample size calculation doesn't seem to take this into account. Methods Study objective: to examine the effects of 2 operating room (OR) • Not reported • Not reported • Not reported • Not reported Random sequence generation (selection bias) Unclear risk Comment: unclear risk of bias because each patient served as their own control but within the patient, the allocation of interventions was unspecified. Unclear risk Comment: no information provided. High risk Outcome group: ulcer incidence. Quote: "Use of the overlay in this manner prevented the investigators from being blinded at the time of postoperative assessment whenever the overlay was used." Comment: high risk of bias because non-blinding is clearly stated. Blinding of outcome assessment (detection bias) All outcomes High risk Outcome group: ulcer incidence. Quote: "Use of the overlay in this manner prevented the investigators from being blinded at the time of postoperative assessment whenever the overlay was used." Comment: high risk of bias because non-blinding is clearly stated. Incomplete outcome data (attrition bias) All outcomes Unclear risk Comment: no information provided. Selective reporting (reporting bias) Unclear risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes. No data are reported on the number or rate of pressure ulcers by group and this would be expected. Only statistically significant odds were reported. Other bias High risk Comment: the study appears to consider parts of a person's body as unit of analysis. However, the logistic regression as described does not appear to take into account the multiple measures per person. Foam surfaces for preventing pressure ulcers ( Incomplete outcome data (attrition bias) All outcomes Low risk Comment: no missing data. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to assess the effectiveness of 3 prevention strategies and compare them to the standard mattress Cochrane Database of Systematic Reviews Random sequence generation (selection bias) Unclear risk Quote: "Patients were randomised by blocks" Comment: unclear risk of bias because the randomisation method was not stated. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Quote: "Given the kind of material tested, blinding was not possible" Comment: high risk of bias as the above statement suggests. Blinding of outcome assessment (detection bias) All outcomes Quote: "Given the kind of material tested, blinding was not possible" Comment: high risk of bias as the above statement suggests. Incomplete outcome data (attrition bias) All outcomes Comment: no attrition identified. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias High risk Comment: the study appears not to consider the interaction between the effects of the different interventions that results from the factorial design used. Methods Study objective: to compare clinical and cost-effectiveness of 2 mattress types: alternating pressure mattresses (APMs) or high specification foam (HSF) Inclusion criteria: inpatient with evidence of acute illness; ≥ 18 years; expected stay ≥ 5 days; expected to comply with follow-up; on electric profiling bed-frame; high pressure ulcer risk due to at least 1 of following: Braden activity score 1/2 and mobility score 1/2; category 1 ulcers; localised skin pain on a healthy/altered/category 1 pressure area Interventions Alternating pressure air mattress (APM) • Description of interventions: fully automatic; some may have dual therapy; for example, the mattress comprises a combination of alternating pressure or low-air-loss. The trial will include only those participants nursed on the alternating pressure mode of action, with a 7.5 to 30 minute cycle time. High-specification foam mattress (HSF) • Description of interventions: be a high-density foam, viscoelastic (memory) foam or a combination of both, and can be castellated (for ventilation and profiling and Personal Social Services (PSS). The NICE £20,000 per QALY gained threshold was used to determine cost-effectiveness. Utility values were derived from the EQ-5D-5L, and costs were estimated using the UK tariff. Costs and outcomes were adjusted for baseline imbalances. Sampling uncertainty was determined via a probabilistic sensitivity analysis (PSA) using a non-parametric bootstrap. Estimates indicate that APM has a 99% probability of being costeffective at a threshold of GBP 20,000 (APMs dominate HSFM, as APM has lower costs and higher QALY values). Lifetime decision-analytic model developed for lifetime cost-effectiveness analysis but data not extracted for this review. Finding is: APM to be cost-effective over both the short and the long term. • Time to healing of all pre-existing category 2 ulcers • Mattress compliance Random sequence generation (selection bias) Low risk Quote: "Participants were randomised centrally (24 h automated telephone system, ensuring allocation concealment) on a 1:1 basis using minimisation (with random element) and minimisation factors: centre, PU status, type of facility, and type of consent" Comment: low risk of bias because of the use of a proper randomisation method. Low risk Quote: "Participants were randomised centrally (24 h automated telephone system, ensuring allocation concealment) on a 1:1 basis using minimisation (with random element) and minimisation factors: centre, PU status, type of facility, and type of consent" Comment: low risk of bias because allocation is properly concealed. Blinding of participants and personnel (performance bias) All outcomes High risk Quote: "Blinding of the research and clinical sta or patients was not possible due to the appearance of the mattresses" Comment: high risk of bias because non-blinding is clearly stated. Blinding of outcome assessment (detection bias) All outcomes Low risk Quote: "Assessment of risk of bias of the primary endpoint was done with central blind review of photographs and a 10% sample of patients who had skin assessments by a practitioner blinded to previous assessments was performed" Comment: low risk of bias because attempts were made to mask outcome assessment. Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "All participants recruited were included using Intention-To-Treat (ITT) and analysed by randomised allocation" Comment: low risk of bias because ITT analysis was performed. Selective reporting (reporting bias) Low risk Comment: the study protocol is available and it is clear that the published reports include all outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to compare whether differences exist between 2 viscoelastic foam support surfaces in the development of new pressure ulcers Total number of participants: 357 randomised; 105 analysed Interventions Viscoelastic foam 1 • Description of interventions: viscoelastic polyurethane foam 1, composed of 2 layers, a 7 cm support surface with 8 cm of high-flexibility foam • NPIAP S3I classification: non-powered, reactive foam surface; multi-layered, viscoelastic polyurethane, high-flexibility foam • Co-interventions: repositioning, nutrition support Cochrane Database of Systematic Reviews Random sequence generation (selection bias) Low risk Quote: "Randomization was performed through an independent, secure, 24hour randomization automated telephone system, ensuring allocation concealment. We used minimization so that groups were parallel" Comment: low risk of bias due to the use of a proper randomisation method. Low risk Quote: "Randomization was performed through an independent, secure, 24hour randomization automated telephone system, ensuring allocation concealment" Comment: low risk of bias due to the proper concealment. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Quote: "Skin follow-up evaluations were completed daily" Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Comment: high risk of bias because "FIGURE. Flow of patients through the trial" shows that of 357 individuals who were randomised, only 105 are included in analysis Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Ozyurek 2015 (Continued) Interventions Viscoelastic foam overlay (VEFO) • Description of interventions: viscoelastic polyurethane polyester foam overlay (Viscosafe Overlay Yellow/Pink 111-45; Safe4Care ApS, Soro, Denmark), placed on top of our standard hospital mattress ... its indentation hardness was 40%, its length was 191 cm, and its width was 90 cm. The core was an open-cell foam with characteristic viscosity and elasticity of 3 cm, respectively. The outer cover of the VEFO is also made of an elastic polyester material designed to be waterproof, breathable and reduce friction. Outcomes that are not considered in this review but reported in trials: • Interface pressure outcome Random sequence generation (selection bias) Low risk Quote: "Participants were randomly allocated to groups using a 1:1 allocation generated via a computer-based program" Comment: low risk of bias because of the use of a proper randomisation method. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes High risk Quote: "We enrolled 122 subjects; 59 were randomly allocated to the experimental group and 63 to the control group ... the final sample comprised 110 subjects; 55 were allocated to the experimental group and 55 in the control group" Quote: "5 subjects transferred to different nursing units during data collection, 3 were found to have PI, IAD, or other skin diseases during the study ..." Comment: high risk of bias because even though the overall dropout rate (9.8%) is not high, some missed participants had incident pressure ulcers during the study. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to compare the rate of healing when patients are treated with low-air-loss (LAL) bed, pressure-relieving bed overlays, and generic total contact seat surface Inclusion criteria: those being alert, able to sit in the 6 months before the study, still sit up with assistance, with a stage III or IV ulcer on the coccyx, trochanter, or ischial tuberosities Exclusion criteria: those with sacral pressure ulcers; previously in a trial to treat their current pressure ulcer; already on low-air-loss, or transfer to low-air-loss planned; skin grafting planned within 1 week; with an active sinus tract or fistula; poor nutrition; requiring antibiotics to treat methicillin-resistant Staphylococcus aureus, vancomycin-resistant Enterococci, or active skin infection; osteomyelitis diagnosed; body weight below 60 kg; unable to flex both hip and knee at least 90 degree Interventions Low-air-loss (LAL) bed • Description of interventions: low-air-loss suspension bed (TheraPulse bed) attaching a rack of inflatable fabric pillows to a modified bed frame to provide pulsating air support intended to increase capillary blood flow and to lower interface pressure. These beds are covered with the manufacturer's Gore-Tex fabric surface to reduce friction. • NPIAP S3I classification: powered, alternating pressure (active), low air loss air surface All reported adverse events using allocated support surfaces • Notes: 1 death in this study but the authors did not specify which group the death was in; 3 participants withdrawn at 4 weeks due to worsened condition, all in overlay group • Not reported • Not reported Outcomes that are not considered in this review but reported in trials: • Ulcer healing • Time to ulcer healing Random sequence generation (selection bias) Low risk Quote: "Randomization was performed by placing a number corresponding to each experimental condition into a sealed envelope with an equal number of Cochrane Database of Systematic Reviews envelopes per condition. A research assistant with no clinical experience drew envelopes by lot as eligible subjects were identified" Comment: low risk of bias because the sequence generation process seems proper. Allocation concealment (selection bias) Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Unclear risk Comment: unclear risk of bias because the dropout rate is low but unbalanced (1 death was excluded from analysis and it was unclear which group the death was in; 3 participants withdrawn at 4 weeks due to worsened condition, all in overlay group). Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Study objective: to determine whether a viscoelastic polymer (energy absorbing) foam mattress was superior to a standard hospital mattress for pressure ulcer prevention and to analyse the cost-effectiveness in comparison with standard hospital mattresses. Duration of follow-up: median days 11 (25th to 75th percentile 6 to 20) in CONFORM-Med; 12 (7 to 22) in standard mattress Single centre or multi-sites: multi-sites Setting: elderly acute care, rehabilitation, and orthopedic wards of hospitals. Inclusion criteria: all patients admitted to acute elderly care and orthopedic wards at hospital 1; elderly rehabilitation wards at hospital 2; and acute elderly care wards at hospital 3 within the preceding 72 hours, who are aged 65 years and older; a pressure ulcer (PrU) risk of 15 to 20 on the Waterlow score, which is based on physiologic, demographic, and disease-specific features; consent to regular examination of pressure areas Foam surfaces for preventing pressure ulcers ( • Notes: no significant differences in comfort assessment were found. The average assessment of comfort for both mattress types ... with levels of 2.33 ± 0.98 and 2.46 ± 1.0 (P = NS) on a 1 to 10 scale. • Reporting: not reported • Reporting: not reported • Outcome type: continuous • Reporting: partially reported • Measurement method (e.g. scale, self-reporting): 2 cost-effectiveness ratios were calculated: (1) a cost per any PrU avoided; and (2) a cost per non-blanching erythema (or worse) avoided. A costeffectiveness acceptability curve was also generated. • Definition: cost-effectiveness acceptability curve plots the probability of the cost-effectiveness of the new mattress against a range of cost-effectiveness ratios. • Notes: an approximately 88% chance that the experimental equipment is the dominant option (i.e. more effective and less costly) ... a 95% chance that the experimental equipment produces a cost per averted non-blanching erythema area of GBP 100 (i.e. USD 140) or less (see Figure 3 ). • Development of blanching erythema • Length of time spent on secondary equipment • Nursing intervention Random sequence generation (selection bias) Low risk Quote: "On admission, participants were randomised to the standard equipment group or the experimental equipment group" Quote: "Equipment allocation at 2 sites was made by converting random numbers (Excel; Microsoft Corp, Redmond, WA) on a 50:50 basis ..." Foam surfaces for preventing pressure ulcers ( Cochrane Database of Systematic Reviews Comment: low risk of bias because study used a proper randomisation method. Low risk Quote: "At site 3, trial numbers were allocated sequentially and the patient chose from 1 of 2 opaque envelopes" Quote: "At sites 1 and 2, each trial ward kept sealed, opaque envelopes containing a trial number and equipment allocation" Quote: "All patients were enrolled into the trial by a research nurse, who carried out the randomization by taking an envelope" Comment: low risk of bias because a proper concealment was likely used. Blinding of participants and personnel (performance bias) All outcomes Quote: "Because ... the experimental mattress surface is distinctive, data collection could not be blinded" Quote:"Although ... it as impossible to blind the research nurses to mattress assignment" Comment: high risk of bias because it is unlikely participants and personnel were blinded. Blinding of outcome assessment (detection bias) All outcomes Quote: "The participants' pressure areas were assessed daily by ward nurses ... A research nurse was immediately notified of any significant deterioration ... completed data collection proformas weekly" Quote: "Because the data collection team examined participants at bedside and the experimental mattress surface is distinctive, data collection could not be blinded" Comment: high risk of bias because outcome assessment was not blinded. Incomplete outcome data (attrition bias) All outcomes Quote: "The primary analysis was intention-to-treat and involved all randomised participants other than the 2 excluded participants" Comment: low risk of bias because ITT analysis is done. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to evaluate the effect of 6 types of hospital mattress on the development of pressure damage Cochrane Database of Systematic Reviews Therarest • Description of interventions: Therarest (KCI Therapeutic Services) with 3 layer therapeutic fire retardant foam core, absorbing and dispersing pressure from high pressure points • NPIAP S3I classification: non-powered, reactive foam surface • Co-interventions: not described • Number of participants randomised: not described Cochrane Database of Systematic Reviews Outcomes that are not considered in this review but reported in trials: • Price of mattresses Random sequence generation (selection bias) Low risk Quote: "Mattresses were randomly allocated to patients using random number tables" Comment: low risk of bias because a proper randomisation method was applied. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Comment: skin assessment by a research nurse but no information as to whether they were blinded. Incomplete outcome data (attrition bias) All outcomes Comment: no information provided. Selective reporting (reporting bias) High risk Comment: 6 types of mattresses were evaluated initially; however, the data collected on the Omnifoam mattress were not analysed because there were insufficient numbers for the results to be significant and they could possibly adversely affect the analysis. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to compare Axtair One, an alternating pressure air mattress (APAM), with a viscoelastic foam mattress (VFM) in elderly patients at moderate to high risk of developing pressure ulcers (PUs) Inclusion criteria: males and females aged 70 and over, bedridden for at least 15 hours per day, with reduced mobility due to medical problems (such as malnutrition, low blood pressure, urinary incontinence, neurological diseases and sensory disorders), a low to zero positioning capability, a Karnofsky score ≤ 40% and a planned period of hospitalisation of at least 2 weeks. Had no PUs at the time of enrolment but had a medium to high risk for developing PUs, as defined by a Braden score ≤ 14. Exclusion criteria: a weight > 120kg, body mass index (BMI) < 12kg/m 2 , a nutritional status score < 12 according to the Mini Nutritional Assessment (MNA), uncompensated nutritional insufficiency and ongoing participation, or within 15 days before, in another clinical research study Interventions Alternating pressure air mattress (APAM) • Description of interventions: APAM (Axtair One, Asklé Santé, Nîmes, France) consisted of therapeutic air cells with a height of 12 cm, supplied by a compressor, which adjusts the pressure based on the patient's weight and whose mode of operation allows alternating inflation of 1 out of 2 cells, with a 6 minute cycle time. All reported adverse events using allocated support surfaces • Notes: the serious adverse events (SAEs) reported in the APAM group were 2 deaths, a massive septic shock with acute pulmonary oedema and a decompensation of an insulin-dependent diabetes. No SAE was reported in the VFM group. There were 20 adverse events reported in each group, including 2 discomforts in the APAM group and one hyperalgesia in the VFM group. The other events did not involve the mattresses. • Reporting: not reported • Reporting: not reported Outcomes that are not considered in this review but reported in trials: • The duration of bed rest • The duration of sitting in a chair • The frequency of preventative interventions • Any therapeutic change Random sequence generation (selection bias) Low risk Quote: "Randomisation was centralised (RANDLIST software v1.2) and globally balanced intracentre with random block sizes established from two possibilities (2 and 4)" Comment: low risk of bias because of the use of a proper randomisation method. Allocation concealment (selection bias) Unclear risk Quote: "Randomisation was centralised (RANDLIST software v1.2) and globally balanced intracentre with random block sizes established from two possibilities (2 and 4)" Comment: unclear risk of bias because even though central randomisation was performed, the small block size means that the allocation in the subsequent block is predictable if a prior randomisation sequence has already been known. Blinding of participants and personnel (performance bias) All outcomes Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "The population selected for the main analysis were all randomised patients in intention-to-treat (ITT)." Comment: low risk of bias because ITT analysis was performed. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to evaluate a special operating room (OR) mattress overlay in preventing pressure ulcer development Single centre or multi-sites: single centre Study start date and end date: not described Outcomes that are not considered in this review but reported in trials: • Risk factors of ulcer development analysed but not extracted Random sequence generation (selection bias) Low risk Quote: "Again, using a random number table, patients were then assigned to the control or the experimental group by a principal investigator" Comment: low risk of bias due to the use of a proper randomisation method. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Quote: "... the study group designation was blinded to all nursing personnel" Comment: unclear because no information provided on participants' blinding. Blinding of outcome assessment (detection bias) All outcomes Quote: "Beginning on the day after surgery and continuing for six days, two research assistants, blinded to the study group of the patient, examined the skin over the bony prominences of each patient for any evidence of skin changes" Comment: low risk of bias because outcome assessors were blinded. Incomplete outcome data (attrition bias) All outcomes Comment: no attrition. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Random sequence generation (selection bias) Unclear risk Quote: "patients for the first two groups were selected by lottery, and thereafter patients were allocated to each group systematically, in rotation" Comment: unclear risk of bias because it is unclear if a proper randomisation method was applied. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) Outcomes that are not considered in this review but reported in trials: • Interface pressure Cochrane Database of Systematic Reviews Random sequence generation (selection bias) Unclear risk Quote: "Those with an expected ICU stay exceeding five days were randomly assigned to be treated on either ..." Comment: unclear risk of bias because a proper randomisation criteria is unspecified. Allocation concealment (selection bias) High risk Comment: randomisation influenced by mattress availability; therefore, allocation not concealed. Blinding of participants and personnel (performance bias) All outcomes High risk Outcome group: pressure ulcer outcome Quote: "The study was not blinded, since the severity of illness of the patients precluded their transfer for evaluation of the skin condition by a blinded reviewer, and the type of mattress in the bed could not be blinded" Comment: high risk of bias because this statement implies blinding of participants and personnel was likely impossible. Blinding of outcome assessment (detection bias) All outcomes Quote: "The study was not blinded, since the severity of illness of the patients precluded their transfer for evaluation of the skin condition by a blinded reviewer, and the type of mattress in the bed could not be blinded" Comment: high risk of bias as it is clearly stated. Incomplete outcome data (attrition bias) All outcomes Quote: "Sequential analysis of the primary outcome variable (pressure sore formation) on an intention-to-treat basis was done after each block of four patients had completed the treatment" Comment: low risk of bias because ITT analysis was conducted. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to evaluate the clinical efficacy of combining a standard 15 cm cold foam mattress with a static air overlay mattress versus a cold foam mattress alone in preventing pressure ulcers Outcomes that are not considered in this review but reported in trials: • Treatment data on the new ulcers reported but not extracted Random sequence generation (selection bias) Unclear risk Quote: "Randomization into two groups was performed after informed consent using numbered envelopes" Comment: unclear risk of bias because the randomisation method used is not sufficiently clear. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Quote: "A weekly inspection of the skin to assess the possible occurrence of a skin lesion was done by an independent nurse" Comment: low risk of bias because the attempt was made to blind outcome assessment. Incomplete outcome data (attrition bias) All outcomes Comment: no attrition identified. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Foam surfaces for preventing pressure ulcers ( Outcomes that are not considered in this review but reported in trials: • Treatment data on the new ulcers reported but not extracted Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote: "Patients were randomised into 2 groups using numbered envelopes" Comment: low risk of bias because, although the randomisation method is not sufficiently presented in the paper, author response suggests remote computer randomisation sequence generation. Unclear risk Comment: unclear risk of bias because author responded that sealed envelopes were opened by nurse but its unclear if envelopes were sequentially numbered and opaque. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Quote: "Patients' skin was inspected weekly to assess the possible occurrence of a skin lesion" Comment: no information provided on the blinding of outcome assessment. Foam surfaces for preventing pressure ulcers ( Incomplete outcome data (attrition bias) All outcomes Comment: no attrition identified; 2 cases were transferred to low-air-loss bed treatments after they developed category III ulcers. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Random sequence generation (selection bias) Low risk Quote: "randomization into 2 groups was performed by using the Castor randomization software (version 1.44; Mionix, Malmo¨, Sweden) ." Comment: low risk of bias because of the use of a proper randomisation method. Allocation concealment (selection bias) Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Quote: "Data were collected weekly, controlled by an independent research nurse." Comment: unclear risk of bias because of insufficient information. Incomplete outcome data (attrition bias) All outcomes Low risk Comment: low risk of bias because it appears to include all 206 patients in analysis. Selective reporting (reporting bias) High risk Comment: high risk of bias because the study protocol is available from https://www.trialregister.nl/trial/4435 and it is clear that the pre-specified costs outcome is not presented. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: this study compares these 2 foam products [MAXIFLOAT foam mattresses and the Iris 3000 foam overlay] based on pressure ulcer incidence in an at-risk population Outcomes that are not considered in this review but reported in trials: • Cost analysis Random sequence generation (selection bias) Low risk Quote: "Subjects meeting the admission criteria were randomly assigned by lot by the investigator who obtained the consent to use either the Iris 3000 or the MAXIFLOAT ... subjects were randomly assigned by research interviewer by drawing assignment out of a hat" Comment: low risk of bias because of the use of a proper randomisation method. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes Comment: no information provided. Blinding of outcome assessment (detection bias) All outcomes Comment: no information provided. Incomplete outcome data (attrition bias) All outcomes Comment: no attrition identified. Selective reporting (reporting bias) Low risk Comment: the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified. Other bias Low risk Comment: the study appears to be free of other sources of bias. Methods Study objective: to provide data that will assist nurses in determining which mattress is the best choice for pressure sore prevention, and under which circumstances Random sequence generation (selection bias) Unclear risk Quote: "26 were selected at random and placed in the foam mattress group, 25 in the AP mattress group" Comment: unclear risk of bias because it is unclear how random sequence was generated. Unclear risk Comment: no information provided. Blinding of participants and personnel (performance bias) All outcomes High risk Outcome group: primary outcome Quote: "... the investigators, who assessed the patient and placed him/her in one of the two mattress groups" Comment: high risk of bias because it is likely the investigators, i.e. key study personnel who operated the study, were not blinded. Blinding of outcome assessment (detection bias) All outcomes Quote: "In most cases patients were assessed by two investigators as a team, and occasionally by only one of the investigators" Quote: "The investigators who rated patient risk and evaluated skin condition knew the mattress assignment of each patient, making investigator bias possible" Comment: high risk of bias because non-blinding of outcome assessment is clearly stated. Cochrane Database of Systematic Reviews of developing a pressure ulcer was greater for patients nursed on convoluted foam than for patients nursed on solid foam when the averaged mobility score was also taken into account. Ozyurek 2015 Overarching class of support surface (as used in this review) Powered/non-powered reactive air surfaces A group of support surfaces constructed of air cells, which redistribute body weight over a maximum surface area (i.e. has reactive pressure redistribution mode), with or without the requirement for electrical power Static air mattress overlay, dry flotation mattress (e.g. Roho, Sofflex), static air mattress (e.g. EHOB), and static mode of Duo 2® mattress Powered/non-powered reactive lowair-loss air surfaces A group of support surfaces made of air cells, which have reactive pressure redistribution modes and a low-airloss function, with or without the requirement for electrical power Low-air-loss Hydrotherapy Powered reactive air-fluidised surfaces A group of support surfaces made of air cells, which have reactive pressure redistribution modes and an air-fluidised function, with the requirement for electrical power Air-fluidised bed (e.g. Clinitron) Non-powered reactive foam surfaces A group of support surfaces made of foam materials, which have a reactive pressure redistribution function, without the requirement for electrical power S24 TI decubitus or AB decubitus S23 TI ( bed sore* or bedsore* ) or AB ( bed sore* or bedsore* ) S22 TI ( pressure ulcer* or pressure sore* ) or AB ( pressure ulcer* or pressure sore* ) S21 (MH "Pressure Ulcer") S20 S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15 OR S16 OR S17 OR S18 OR S19 S19 TI net bed* or AB net bed* S18 TI ( kinetic therapy or kinetic table* ) or AB ( kinetic therapy or kinetic table* ) S17 TI ( turn* bed* or tilt* bed* ) or AB ( turn* frame* or tilt* frame* ) Cochrane Database of Systematic Reviews S16 TI sheepskin OR AB sheepskin S15 TI water suspension or AB water suspension S14 TI air bag* or AB air bag* S13 TI air suspension or AB air suspension S12 TI alternat* pressure or AB alternat* pressure S11 TI static air or AB static air S10 Appendix 3. Risk of bias 1 'Risk of bias' assessment in individually randomised controlled trials The study authors describe a random component in the sequence generation process, such as referring to a random number table, using a computer random number generator, coin tossing, shu ling cards or envelopes, throwing dice, drawing of lots. Trusted evidence. Informed decisions. Better health. Any one of the following. • Insu icient information to permit a judgement of low or high risk of bias. • The study did not address this outcome. Any one of the following. • No missing outcome data. • Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias). • Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. • For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk is not su icient to have a clinically relevant impact on the intervention e ect estimate. • For continuous outcome data, the plausible e ect size (di erence in means or standardised di erence in means) among missing outcomes is not su icient to have a clinically relevant impact on observed e ect size. • Missing data have been imputed using appropriate methods. Any one of the following. • Reason for missing outcome data is likely to be related to the true outcome, with either imbalance in numbers or reasons for missing data across intervention groups. • For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk is su icient to induce clinically relevant bias in intervention e ect estimate. • For continuous outcome data, the plausible e ect size (di erence in means or standardised di erence in means) among missing outcomes is su icient to induce clinically relevant bias in the observed e ect size. • 'As-treated' analysis done, with substantial departure of the intervention received from that assigned at randomisation. • Potentially inappropriate application of simple imputation. Any one of the following. • Insu icient reporting of attrition/exclusions to permit judgement of low or high risk of bias (e.g. number randomised not stated; no reasons for missing data provided). • The study did not address this outcome. Any of the following. • The study protocol is available and all of the study's prespecified (primary and secondary) outcomes that are of interest in the review have been reported in the prespecified way. • The study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were prespecified (convincing text of this nature may be uncommon). Any one of the following. • Not all of the study's prespecified primary outcomes have been reported. • One or more primary outcomes are reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not prespecified. • One or more reported primary outcomes were not prespecified (unless clear justification for their reporting is provided, such as an unexpected adverse e ect). • One or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta-analysis. • The study report fails to include results for a key outcome that would be expected to have been reported for such a study. Three types of foam mattresses, each served as an arm in Bliss 1995a: • Groove (a contoured 10-cm thick foam overlay) • Propad (an 8.5 cm thick foam pad • Preventix (a 16-cm thick mat of 8-cm square foam modules of different densities) • NPIAP S3I classification: non-powered, reactive, foam surfaces; foam characteristics unspecified Two types of fibre-filled mattresses, each served as an arm in Bliss 1995a: • Spenco (cotton hollow-core fibre-filled) • Surgicgoods Hollowcore Mattress fibre-filled pad • NPIAP S3I classification: non-powered, reactive, fibre surfaces Polyether foam pad (2 feet x 2 feet x 3-inch thickness) • Support-surface-associated patient comfort (follow-up duration minimum 11.5 days, maximum 14.0 days) Two studies (1269 participants) that compared foam surfaces with undefined 'standard hospital surfaces' reported this outcome (Gunningberg 2000; Russell 2003a ). The two studies reported different measures and outcome data: Gunningberg 2000 measured comfort using a five point scale (higher score = better comfort) and reported a mean rating of comfort of 4.2 for foam surfaces and 4.0 for standard hospital mattress. Russell 2003a measured this using a ten point scale (higher score = poorer comfort) but reported no significant differences in comfort between foam mattresses (mean 2.33 and SD 0.98) and standard hospital mattress (mean 2.46 and SD 1.0). All reported adverse events (follow-up duration 12 weeks) Van Leen 2018 (206 participants) compared foam surfaces with Bedcare surfaces. The study reported this outcome but stated that there was no reported adverse events in either study group. It is uncertain if there is a difference in the adverse effects between foam surfaces and the undefined reactive surfaces. Evidence was of very low certainty, downgraded twice for high risk of bias in a domain other than performance bias, and once for imprecision as the sample size was small and the number of events was relatively low. Cost-effectiveness (follow-up duration 11.5 days) Russell 2003a (1168 participants) compared foam surfaces with undefined 'standard hospital surfaces'. The study reported this outcome using two measures: cost per any pressure ulcer (including blanching erythema) prevented; and cost per non-blanching erythema (or worse) avoided. The results suggest that foam surfaces have a 88% probability of being cost effective compared with standard hospital surfaces in preventing any pressure ulcer (including blanching erythema); and have a 95% probability of being cost effective in preventing non-blanching erythema or worse. Hoshowsky 1994 ? ? --? ? -Kemp 1993 + ? ? ? + + + Laurent 1998 ? ? --+ + -Nixon 2019 + + -+ + + + Ozyurek 2015 + + ? ? -+ + Park 2017 + ? ? ? -+ + Rosenthal 2003 + ? ? ? ? + + Russell 2003a + + --+ + + Santy 1994 + ? ? ? ? -+ Sauvage Air-fluidized beds or conventional therapy for pressure sores. A randomized trial GRADE: an emerging consensus on rating quality of evidence and strength of recommendations GRADE guidelines: 4. Rating the quality of evidencestudy limitations (risk of bias) Development and preliminary testing of the new fivelevel version of EQ-5D (EQ-5D-5L) Measuring inconsistency in meta-analyses Cochrane Handbook for Systematic Reviews of Interventions version 5.2.0 (updated Cochrane Handbook for Systematic Reviews of Interventions version 6 Cochrane Handbook for Systematic Reviews of Interventions version 6 Foam surfaces for preventing pressure ulcers #1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 AND INREGISTER *:ti,ab,kw #3 (foam or transfoam):ti,ab Copyright © 2021 The Authors. Cochrane Database of Systematic Reviews published by John Wiley & Sons, Ltd. on behalf of The Cochrane Collaboration low pressure" near/2 device*):ti,ab,kw #11 ("low pressure" near/2 support):ti,ab,kw #12 (constant near/2 pressure):ti,ab,kw #13 "static air":ti,ab,kw #14 (alternat* next pressure):ti,ab,kw #15 (air next suspension*):ti,ab,kw #16 (air next bag*):ti,ab,kw #17 (water next suspension*):ti,ab,kw #18 sheepskin:ti,ab,kw #19 (turn* or tilt*) next (bed* or frame*):ti,ab,kw #20 kinetic next Pressure Ulcer] explode all trees #24 (pressure next (ulcer* or sore* or injur*)):ti,ab,kw #25 (decubitus next (ulcer* or sore*)):ti,ab,kw #26 ((bed next sore*) or bedsore*):ti,ab Copyright © 2021 The Authors. Cochrane Database of Systematic Reviews published by John Wiley & Sons, Ltd. on behalf of The Cochrane Collaboration Copyright © 2021 The Authors. Cochrane Database of Systematic Reviews published by John Wiley & Sons, Ltd. on behalf of The Cochrane Collaboration Copyright © 2021 The Authors conceived the review; designed the review; coordinated the review; extracted data; analysed or interpreted data; undertook quality assessment; performed statistical analysis; produced the first dra of the review review; analysed or interpreted data; checked quality of statistical analysis; produced the first dra of the review; contributed to writing or editing the review; advised on the review; secured funding conceived the review; designed the review; coordinated the review; checked quality of data extraction; contributed to writing or editing the review; advised on the review; secured funding conceived the review; designed the review; checked quality of data extraction; checked quality assessment; checked quality of statistical analysis Elizabeth McInnes: conceived the review; designed the review; coordinated the review; checked quality of data extraction; checked quality assessment Editor): edited the protocol; advised on methodology, interpretation and content; approved the final protocol prior to publication Trusted evidence. Informed decisions. Better health. Alternating pressure (active) air surfaces A group of support surfaces made of air cells, which offer both reactive and active pressure redistribution modes as well as a low air loss function, with the requirement for electrical power Stand-alone bed unit with alternating pressure, static modes and low air-loss (e.g. TheraPulse) A group of support surfaces made of any materials, used as-usual in a hospital and without reactive or active pressure redistribution capabilities, nor any other functions (e.g. low air loss, or air-fluidised) The study authors describe a non-random component in the sequence generation process. Usually, the description would involve some systematic, non-random approach, for example, sequence generated by odd or even date of birth, sequence generated by some rule based on date (or day) of admission, sequence generated by some rule based on hospital or clinic record number.Cochrane Database of Systematic Reviews Insu icient information to permit judgement of low or high risk of bias. It is likely that the majority of studies will fall into this category. The study appears to be free of other sources of bias. There is at least one important risk of bias. For example, the study:• had a potential source of bias related to the specific study design used; or • has been claimed to have been fraudulent; or • had some other problem. There may be a risk of bias, but there is either:• insu icient information to assess whether an important risk of bias exists; or • insu icient rationale or evidence that an identified problem will introduce bias. Recruitment bias (or identification bias) is the bias that occurs in cluster-RCTs if the personnel recruiting participants know individuals' allocation, even when the allocation of clusters has been concealed appropriately. The knowledge of the allocation of clusters may lead to bias because the individuals' recruitment in cluster trials is o en behind the clusters' allocation to di erent interventions; and the knowledge of allocation can determine whether individuals are recruited selectively.This bias can be judged through considering the following questions.• Were all the individual participants identified/recruited before randomisation of clusters?• Is it likely that selection of participants was a ected by knowledge of the intervention?• Were there baseline imbalances that suggest di erential identification or recruitment of individual participants between arms? Baseline imbalance between intervention groups can occur due to chance, problems with randomisation, or identification/recruitment bias. The issue of recruitment bias has been considered above.In terms of study design, the risk of chance baseline imbalance can be reduced by the use of stratified or pair-matched randomisation.Minimisation -an equivalent technique to randomisation -can be used to achieve better balance in cluster characteristics between intervention groups if there is a small number of clusters.Concern about the influence of baseline imbalance can be reduced if studies report the baseline comparability of clusters, or statistical adjustment for baseline characteristics. Similar to missing outcome data in individually randomised trials, bias can occur if clusters are completely lost from a cluster-RCT, and are omitted from the analysis.The amount of missing data, the reasons for missingness and the way of analysing data given the missingness should be considered in assessing the possibility of bias. Data analyses, which do not take the clustering into account, in cluster-RCTs will be incorrect. Such analyses lead to a 'unit of analysis error' and over-precise results (overly small standard error) and overly small P values. Though these analyses will not result in biased estimates of e ect, they (if not correctly adjusted) will lead to too much weight allocated to cluster trials in a meta-analysis.Note that the issue of analysis may not lead to concern any more and will not be considered substantial if approximate methods are used by review authors to address clustering in data analysis.Cochrane Database of Systematic Reviews In the case that a meta-analysis includes, for example, both cluster-randomised and individually randomised trials, potential di erences in the intervention e ects between di erent trial designs should be considered. This is because the 'contamination' of intervention e ects may occur in cluster-RCTs, which would lead to underestimates of e ect. The contamination could be known as a 'herd e ect': that is, within clusters, individuals' compliance with using an intervention may be enhanced, which in return a ects the estimation of e ect. • Two review authors independently assessed the titles and abstracts of the new search results for relevance using Rayyan rather than using Covidence. • For new included studies, one review author independently extracted data and another review author checked all data, rather than two review authors independently carrying out data extraction. • When a study only had complete case data, we considered complete case data in the related main analysis (i.e. assuming no missing data issue). This was not pre-planned. • We presented separate 'Summary of findings' tables for five of the seven comparisons evaluated in this review. We did not present the tables for the comparison between di erent types of foam surfaces and the comparison of foam surfaces versus reactive water surfaces. • Where we did not pool data, we conducted a GRADE assessment and presented these assessments in a narrative format in 'Summary of findings' tables. This was not pre-planned.