key: cord-0798962-75tykglp
authors: Clare, Tom; Twohig, Katherine A.; O’Connell, Anne-Marie; Dabrera, Gavin
title: Timeliness and completeness of laboratory-based surveillance of COVID-19 cases in England
date: 2021-04-01
journal: Public Health
DOI: 10.1016/j.puhe.2021.03.012
sha: 5bd2f0756d9bf106ba279f55be7a936fdcf08193
doc_id: 798962
cord_uid: 75tykglp

Objective Evaluate completeness and timeliness of the rapidly developed surveillance of SARS-CoV-2 in England using patient-level case data. Study Design Observational study. Evaluation of public health surveillance systems. Methods Data was collected in the Public Health England Second-Generation Surveillance System through routine laboratory reporting processes, as well as enhanced testing in collaboration with commercial partners. Three periods were chosen to present developments in disease surveillance around the first pandemic wave in England. Completeness of valid entries for key demographic and epidemiological fields were summarised. Timeliness was assessed using recorded date intervals; the sample collection to the laboratory reporting a positive result; the positive result being received by the national surveillance system; and the data being available for epidemiological analysis. Results In each period, demographic variables were more than 95% complete, and enhanced ethnicity more than 85%, allowing a rich understanding of the general characteristics of COVID-19 cases in England. The proportion of cases completing all reporting stages of the national system within 3 days of when the specimen was taken increased from 69.1% in period 1 to 76.6% in period 3. In period 3, the median number of days to complete all reporting stages decreased to two, from three in previous periods. Analysis of each reporting stage offers suggestive evidence that timeliness of the system has improved as reporting has become established over time. Conclusions Timely processing of data for epidemiological use was consistent and rapid once received by the national system. Delays in timeliness were most likely to occur in the first stage of the reporting process, prior to laboratory input to the surveillance platform. Existing national surveillance mechanisms enhanced during the response have succeeded in providing rapid collection and reporting of case data to facilitate epidemiological monitoring and analysis, and guide public health policy and strategy.

Existing national surveillance mechanisms enhanced during the response have succeeded 24 in providing rapid collection and reporting of case data to facilitate epidemiological 25 monitoring and analysis, and guide public health policy and strategy. 26

Surveillance of the novel coronavirus disease, COVID-19, was escalated in England in early 28 2020, with initial cases reported in January 2020 [1] . Rapid detection of new incident cases 29 was a key priority, and initial processes were built into existing laboratory reporting systems 30 including the Second-Generation Surveillance System (SGSS) and Respiratory Data Mart [2] . 31

Urgent need to improve case ascertainment, as well as alleviate testing capacity challenges, 32 resulted in the UK government's deployment of a strategy to scale up testing for COVID-19 33 in April 2020 [3] . This policy referred to testing "pillars", with three that contributed to the 34 detection of cases with current infection: pillar 1, aiming to strengthen established testing 35 pathways, such as National Health Service (NHS) and Public Health England (PHE) 36 laboratories; pillar 2, initiating testing capacity through commercial partners; and pillar 4, 37 swab testing for surveillance studies. This expansion of testing aimed to provide more rapid 38 results to improve data collection to better understand the epidemiological characteristics of 39 infection, and to support key workers ability to return to work with reduced risk. Based on key 40 priorities of data completeness and timeliness, we evaluated the rapidly developed and 41 expanded laboratory surveillance system for SARS-CoV-2 around the first pandemic wave in 42

England. 43 44 Data on laboratory-confirmed cases of SARS-CoV-2 infection are legally required to be 45 submitted to PHE by the operators of diagnostic laboratories; submitted laboratory data is 46 managed within the SGSS. Three periods were chosen to present developments in the surveillance system's timeliness and completeness. Cases were assigned to a period by lab 48 report date, which has 100% completeness and validity. The periods were: 30 January -26 49

April (the set-up period; ends the week with the highest number of cases reported); 27 April -50 always represent homogenous populations. Pillar 1 includes testing of patients in hospitals 56 (through routine diagnostic investigations or due to COVID-19 symptoms), as well as testing 57 of health and social care workers. Pillar 2 testing broadly represents community testing in 58 the wider population, including mildly symptomatic cases and testing from mobile units. Both 59 pillar 1 and pillar 2 contains some outbreak investigations and care home testing, where 60 reporting is based on whether the testing is carried out by a PHE/NHS (pillar 1) or 61 commercially contracted (pillar 2) laboratory. Pillar 4 swab tests can be reported into either 62 pillar 1 or pillar 2 depending on the diagnostic laboratory contracted for the study, and pillar 4 63 results are not consistently distinguishable within the surveillance system. While pillar 1 built 64 upon existing laboratory reporting pathways with established data flows, pillar 2 and 4 65 required new processes to be created for both data collection and submission. This can lead 66 to differences in both timeliness and completeness of reported data fields by reporting pillar. 67 criterion [7] . Where there are differing ethnicities for the same personally identifiable 81 information, priority is given according to: a) a valid ethnicity (ie not including "Unknown" or 82 "Prefer not to say"); b) the most recent date; and c) higher ranked datasets. The datasets 83 rank, starting with the highest, as follows: SUS live feed; HES Admitted Patient Care; 84 Outpatient HES; and HES Accident and Emergency. Where this linkage did not result in a 85 valid ethnicity for cases reported through pillar 2, the self-reported pillar 2 ethnicity was used. 86

Postcodes that were indicated as being populated with laboratory or GP default information 87 were considered missing for the purposes of assigning patient residential information. 88

Timeliness was assessed using four key date fields to construct three intervals: (1) specimen 89 date to lab report date, which is the time between the sample being collected and the 90 laboratory reporting positive results to its systems; (2) lab report date to SGSS received 91 date, indicating the time taken from the positive result being available to it being received by 92 the national surveillance system; (3) SGSS received to import date, the time between receipt 93 in SGSS and the data being imported so it can be used by epidemiologists, statisticians and 94 modellers. Some of these intervals occur on the same day; for instance, intervals 2 and 3 95 could occur on the same day. The timeliness analysis included only case records from 14 96

April for pillar 1 and from 26 May for pillar 2 due to limitations on available date fields prior to 97 this and the end of the analysis period was 6 September 2020. 98 public health to inform policy decision making (including local public health restrictions) [8, 9] , 108 modelling to provide forecasting and tracking of the pandemic in real-time [10, 11, 12] , routine 109 surveillance reporting of official statistics [4, 13] , and peer-reviewed research [14, 15] . 110

The least complete demographic field was the NHS number, an identifier linked to a patient's 111 electronic health record. This field is routinely enhanced in SGSS through matching to the 112 Demographic Batch Service (DBS) [16] . Low completeness is likely due to matching requiring 113 a high level of precision that is not always available for self-reported information (such as 114 through pillar 2). Completeness for this field decreased across the study period from 92.9% 115 to 80.8%. While part of this decrease reflects an increasing proportion of national COVID-19 116 cases being reported through the pillar 2 reporting pathway, pillar 1 completeness also 117 decreased from approximately 94% in periods 1 and 2, to 80.9% in period 3. 118

Key epidemiological surveillance variables reported by laboratories were mainly incomplete. 119

Availability of symptom onset date decreased from 2.2% to 0.2% from period 1 to 3, as the 120 proportion of cases detected through pillar 2 increased, with almost entirely incomplete data 121 for this field, after its inclusion in pillar 2 data collection in May. The asymptomatic indicator 122 has shown the greatest completeness improvement, increasing from 1.4% to 88.5% across 123 the analysis periods. This is almost entirely due to improvements in completeness for pillar 2 124 testing, in which this became a mandatory variable in late June 2020. Other indicators, such 125 as travel exposure and hospital-acquired infection status, were generally unavailable through 126 pillar 1 and not collected through pillar 2. 127 cases in each period. The timeliness of the second reporting stage, lab report to SGSS 134 received, improved significantly over time, completing within 1 day from 41% to 74.5% of 135 reports between periods 1 and 3. The final reporting stage, SGSS received to import date, 136 occurred within 1 day for 90% of cases in all periods demonstrating that processing for 137 epidemiological use was consistent and rapid once data was received by the national 138 system. 139

The two primary COVID-19 case reporting pathways (ie pillars) show distinct patterns in 140 reporting by interval. The first interval, from the specimen date to lab report date, is typically 141 shorter for those within the pillar 1 system with 95% processed within 3 days, while it takes 142 up to 4 days to see that level of completeness for pillar 2. Conversely, reporting from the 143 laboratory to SGSS is quicker through pillar 2, with more than three-quarters of cases 144 received by SGSS on the same day as the lab report (77.3%, compared with 54.4% of cases 145 from pillar 1 laboratories). 146

Combining the three reporting stages describes the overall timeliness of case data being 147 reported through the surveillance system from the date a patient is tested. The largest 148 improvements in timely reporting occurred between days 1 to 3. The proportion of cases 149 completing all reporting stages within 2 days increased from 27.2% in period 1 to 53% in 150 period 3, and within 3 days increased from 69.1% to 76.6% over the same time. The 151 proportion completing within 4 days was relatively stable in each period (from 84.4% to 152 86.8%). In period 3, the median number of days to complete all reporting stages decreased 153 to 2, from 3 in previous periods. 154

Analysis of each reporting stage of the new surveillance system offers suggestive evidence 155 that the timeliness of the system has improved as COVID-19 reporting has become 156 The COVID-19 pandemic has changed the landscape of public health surveillance in 172

England. Existing surveillance mechanisms that have been enhanced during the response, 173 such as SGSS, have succeeded in providing rapid collection and reporting of case data to 174 facilitate epidemiological monitoring and analysis, and guide public health policy and 175 strategy. Larger-scale health service or diagnostic laboratory reporting improvements may 176 be required to address remaining limitations, as well as an emphasis on high-quality data 177 collection. The surveillance and health information structures that have been developed, and 178 will continue to be refined, will allow public health services to better characterise the 179 pandemic to the benefit of healthcare professionals and the public, with potential learning 180 and application for the surveillance of other infectious diseases in the future. The data analysed during this study are not publicly available due to a need to protect 197 individual's anonymity. These data are confidential, but fully anonymised data may be 198 available from the corresponding author on reasonable request. Aggregated and 199 anonymised output from the dataset described is publicly available at 

COVID-19: public health management of the first two confirmed cases 203 identified in the UK. Epidemiology and Infection

Department of Health & Social Care. Coronavirus (COVID-19): scaling up testing 211 programmes 2020

National flu and COVID-19 surveillance weekly reports 2020

Hospital Episode Statistics (HES)

Available at Hospital Episode Statistics and the Office for National Statistics

Department of Health & Social Care. COVID-19 contain framework: a guide for local 232 decision-makers

Scientific Advisory Group for Emergencies. Scientific evidence supporting the 251 government response to coronavirus (COVID-19): Evidence considered by the Scientific 252

2020: 26

Statistics at PHE. 2021: 2021. Available at

SARS-CoV-2 infections in children following the full re-opening of 261 schools and the impact of national lockdown: Prospective, national observational cohort 262 surveillance