key: cord-1032980-amfxrvih
authors: Neville, Thanh H.; Hays, Ron D.; Tseng, Chi-Hong; Gonzalez, Cynthia A.; Chen, Lucia; Hong, Ashley; Yamamoto, Myrtle; Santoso, Laura; Kung, Alina; Schwab, Kristin; Chang, Steve Y.; Qadir, Nida; Wang, Tisha; Wenger, Neil S.
title: Survival After Severe COVID-19: Long-Term Outcomes of Patients Admitted to an Intensive Care Unit
date: 2022-04-05
journal: J Intensive Care Med
DOI: 10.1177/08850666221092687
sha: dc1368f6d71aa6af58a30e0059f2d41c4992f8f1
doc_id: 1032980
cord_uid: amfxrvih

BACKGROUND: Understanding the long-term sequelae of severe COVID-19 remains limited, particularly in the United States. OBJECTIVE: To examine long-term outcomes of patients who required intensive care unit (ICU) admission for severe COVID-19. DESIGN, PATIENTS, AND MAIN MEASURES: This is a prospective cohort study of patients who had severe COVID-19 requiring an ICU admission in a two-hospital academic health system in Southern California. Patients discharged alive between 3/21/2020 and 12/31/2020 were surveyed approximately 6 months after discharge to assess health-related quality of life using Patient-Reported Outcomes Measurement Information System (PROMIS®)-29 v2.1, post-traumatic stress disorder (PTSD) and loneliness scales. A preference-based health utility score (PROPr) was estimated using 7 PROMIS domain scores. Patients were also asked their attitude about receiving aggressive ICU care. KEY RESULTS: Of 275 patients admitted to the ICU for severe COVID-19, 205 (74.5%) were discharged alive and 132 (64%, median age 59, 46% female) completed surveys a median of 182 days post-discharge. Anxiety, depression, fatigue, sleep disturbance, ability to participate in social activities, pain interference, and cognitive function were not significantly different from the U.S. general population, but physical function (44.2, SD 11.0) was worse. PROPr mean score of 0.46 (SD 0.30, range −0.02 to 0.96 [<0 is worse than dead and 1 represents perfect health]) was slightly lower than the U.S. general population, with an even distribution across the continuum. Poor PROPr was associated with chronic medical conditions and receipt of life-sustaining treatments, but not demographics or social vulnerability. PTSD was suspected in 20% and loneliness in 29% of patients. Ninety-eight percent of patients were glad they received life-saving treatment. CONCLUSION: Most patients who survive severe COVID-19 achieve positive outcomes, with health scores similar to the general population at 6 months post-discharge. However, there is marked heterogeneity in outcomes with a substantial minority reporting severely compromised health.

cognitive, and psychological impairments that arise from critical illness, persisting long after a patient leaves the ICU. [13] [14] [15] PICS can contribute to decreased health-related quality of life (HRQOL). 16, 17 PICS can also lead to financial strain, with 40% of ICU survivors unable to return to work a year after hospital discharge. 18, 19 Admission to the ICU for COVID-19 involves isolation from family, uncertainty, and limited human interaction-factors that are predictive of long-term psychological harm for ICU patients and their families. [20] [21] [22] [23] [24] Delirium, which is associated with persisting cognitive impairment after discharge, 25 is common in COVID-19 patients in the ICU. 20, 26 Furthermore, infection prevention mandates make coordination of care such as inpatient rehabilitation more challenging for COVID-19 patients, placing them at a higher risk for deconditioning. Taken together, these factors might be expected to cause severe PICS for the survivors of COVID-19 ICU care, but this has not been clearly elucidated.

We aimed to characterize post-discharge HRQOL approximately 6-months after discharge for patients admitted to the ICU with severe COVID-19. We also surveyed whether patients were glad they received aggressive critical care given their current state-information that might help providers better engage future patients and families in informed decisionmaking regarding invasive life-sustaining interventions. With an ever-increasing number of COVID-19 survivors, it is critical to understand the trajectory of severe COVID-19 and optimize recovery after hospital admission.

This is a prospective cohort study conducted at a two-hospital academic health system in Southern California. The study was approved by the health system's Institutional Review Board (IRB #20-000817) and is reported in accordance with the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) reporting guideline. 27 Participants Patients were included in the study if they were at least 18 years old, admitted to the ICU for laboratory-confirmed COVID-19, and discharged between March 21 and December 31, 2020. Patients transferred to another acute care hospital and those who incidentally tested positive but were admitted to the ICU for unrelated reasons were excluded. Demographics, comorbidities, highest Sequential Organ Failure Assessment (SOFA) 28 score within the first 24 hours of ICU admission, clinical course, and discharge status were retrospectively abstracted from the electronic medical record (EMR) and stored in REDCap. 29 Elixhauser comorbidity index (ECI)a set of comorbidities that predict in-hospital mortalitywas calculated for each patient. 30 Each patient's Social Vulnerability Index (SVI) was computed based on their address. The SVI uses 15 census variables including poverty and housing, and is associated with a community's capacity to respond and recover when confronted by external stresses on human health. 31, 32 Patients who survived to discharge were eligible for posthospital surveys. Surveys were mailed to the patient's home two weeks prior to their 3-month and 6-month post-hospital discharge dates. For the analyses reported, we used the survey completed closest to 6-months after discharge. The mailing included an introductory letter, an informational sheet that described risks, benefits, and alternatives to participating, the survey itself, and a stamped pre-addressed return envelope. A Spanish-language version of the survey was mailed to patients for whom the EMR specified that Spanish was their preferred language. Reminder cards were sent one week later. After 2 weeks, study personnel called the patient to prompt completion, offer an online option if preferred, or conduct the survey over the telephone. Non-responders received at least two telephone calls, one after 5pm. Survey participants received a $25 gift card. For patients who did not respond to the survey, postdischarge clinical status was abstracted from the EMR including information obtained from the institution's Ambulatory Monitoring Program, a readmission prevention clinical program in which nurses call COVID-19 patients to query symptoms and whether patients have returned to baseline health.

The survey (eSupplement 1) included Patient-Reported Outcomes Measurement Information System (PROMIS®) measures. 33 We incorporated the PROMIS-29 v2.1, which includes a pain intensity item and 4 items assessing each of 7 domains: physical function, anxiety, depression, fatigue, sleep disturbance, social participation, and pain interference; 34 the PROMIS Cognitive Function Short Form 8a V2.0; 35 and four PROMIS global health questions (2 physical health and 2 mental health). 36 PROMIS measures have undergone extensive development and evaluation in the general population and a variety of chronic diseases. 33,37-39 PROMIS scores are reported on a T-score metric, with a mean of 50 and standard deviation of 10 in the general U.S. population. We used pattern-based item response theory scoring for the PROMIS measures. 40 PROMIS-29 domains had excellent internal consistency reliability in this study: coefficient alpha was 0.95 for physical function, 0.96 for anxiety, 0.94 for depression, 0.96 for fatigue, 0.88 for sleep disturbance, 0.98 for social participation, 0.99 for pain interference, and 0.99 for cognitive function. We also estimated the PROMIS-29 physical and mental health summary scores. 41 In addition, we estimated the PROMIS-Preference (PROPr) preference-based utility score from 7 PROMIS domains: depression, fatigue, pain interference, physical function, sleep disturbance, ability to participate in social activities, and cognitive function. The PROPr was calculated using SAS (version 9.4) and ranges from −0.022 to 1.0 where < 0 is worse than dead and 1.0 is perfect health in a representative U.S. sample. [42] [43] [44] Finally, seven questions were added in month three of the study to assess current physical function, cognition, mood, fatigue, sleep, social activities, and pain compared to baseline prior to COVID-19 infection.

Post-traumatic stress disorder (PTSD) was assessed by the short form of PCL-5 (PTSD Checklist). 45 , 46 The four questions are scaled 0-4 and a sum of ≥ 8 was used to indicate probable PTSD. 46 To evaluate the potential effect of isolation, we used a 3-item Loneliness Scale, in which a score ≥ 6 indicates loneliness. 47 These three items were derived from the revised UCLA Loneliness Scale and are a subset of another short-form measure. 48 The survey also asked whether the patient had a paying job prior to hospitalization and whether they returned to work. Lastly, the survey asked whether the patient was glad that they received aggressive life-sustaining treatment for COVID-19 and whether they would be willing to receive this type of treatment again.

We characterized and compared patients who died during the hospitalization and those discharged alive. Among patients who survived to discharge, we compared survey responders and non-responders. Bivariate comparisons between patient groups were performed using χ2, Fisher's exact, two-sample t-tests, and Wilcoxon rank-sum tests as appropriate. One-sample t-tests were used to compare standardized PROMIS scores to a general U.S. population average of 50. All analyses were two-tailed and a p-value < 0.05 was considered statistically significant. Analyses were performed using STATA/SE 15.1 and R statistical software (version 4.1.0).

Kruskal-Wallis tests were used to test the association between PROPr and categorical variables, and Pearson's correlation coefficient for continuous variables. Baseline characteristics and outcomes were compared between patients who had PROPr > 0.2 and those with PROPr ≤ 0.2, representing patients with very low health utilities.

Between March 11, 2020 (start of the pandemic in Los Angeles) and December 31, 2020, 275 patients were admitted to the health system's ICUs for COVID-19 and were discharged (n = 205) or died inpatient (n = 70) ( Figure 1 ). Compared to those who died, the 75% of patients who survived ICU admission for severe COVID were younger, had lower ICU admission Sequential Organ Failure Assessment (SOFA) scores, lower ECI, shorter ICU stays, and were less likely to receive mechanical ventilation (p < 0.05 for all comparisons, Table 1 ). Survivors were more likely to have received dexamethasone (recommendations favoring dexamethasone were announced in July 2020). 49 They were also more likely to have private insurance and have English as their primary language, although they had similar race and ethnicity and SVI to those who died.

Of the 205 patients who were discharged from the ICU, 10 (4.9%) died before receiving 3-month surveys and 132 (64.3%) completed either or both 3-month and 6-month surveys. Surveys used in this analysis were completed a median of 182 (IQR 173-188) days after hospital discharge. Those who completed surveys had higher Social Vulnerability Index scores (on 0-1 scale, 0.7 vs. 0.4, p = 0.004), lower Elixhauser comorbidity index (8.0 vs. 10.0, p = 0.003), and were less likely to have chronic kidney disease (16.7% vs. 30.1%, p = 0.038) ( Table 1 ) than those who did not return surveys. Other variables in Table 1 were not significantly different between survey responders and non-responders.

Of the 63 non-respondents who survived three months after discharge, 31 (49.2%) reported being "back to baseline" health and 12 (19.0%) reported continued symptoms according to Ambulatory Monitoring Program outpatient notes ( Figure 1 ). Five (7.9%) were in extremely compromised health states (eg, advanced dementia in long term care) before COVID-19 infection, 1 (1.6%) became severely compromised due to COVID-19 (discharged obtunded to a long-term acute care facility), and 14 (22.2%) were unable to be characterized (3 were known dead within one year of discharge).

The surveyed sample of 132 survivors of COVID-19 in the ICU had a median age of 59 years with the majority being male, Hispanic and having English as their preferred language. Insurance was nearly evenly split between commercial and public. About half were working, 47% had a family income less than $35,000, 41% had a high school education or less, and median SVI was 0.7. Most participants had at least one co-morbid condition and nearly 10% had received an organ transplant. Mean body mass index (BMI) was 29 (overweight, nearly obese). Nearly all were treated with at least one COVID-19-specific therapy with over one-third requiring ventilator support and vasopressors during the ICU stay. The median length of stay in the ICU and the hospital was 6 days and 13 days, respectively. (Tables 1 and 2 

Six months after hospital discharge, 94% of respondents were living at home, 16% required help from a caregiver that was not needed before COVID-19, and 17% were on supplemental oxygen (2 of whom were previously on oxygen). Of the 68 respondents who were employed before hospitalization, 40 (59%) were back to work, with a median hiatus of 6 weeks, and 32 (47%) employed at their pre-morbid level. Twelve (9%) had been readmitted to the hospital. (Table 2) 

Six months after hospitalization, anxiety, depression, fatigue, sleep disturbance, ability to participate in social roles and activities, pain interference, and cognitive function scores in survivors were not significantly different from the U.S. general population (Figure 2 ). Physical function (44.2, SD 11.0), however, was significantly worse than that of the population. PROMIS-29 physical and mental health summary scores were 45.2 (SD 11.6) and 50.5 (SD 11.4), respectively. PROMIS Global Physical and Mental Health scores were lower than the U.S. mean of 50 (47.1, SD 9.8 and 43.1, SD 10.8, respectively). Of the 117 patients who answered questions regarding their physical function, cognition, mood, fatigue, sleep, social activities, and pain compared to before COVID-19, the majority of patients felt "about the same" or "better" in all domains except physical function and fatigue, where the majority felt that they were worse (Table E1 ).

The mean PROPr preference-based utility score of patients who survived severe COVID-19 was 0.46 (SD 0.30) and ranged from −0.02 to 0.96 (a negative score indicates a state worse than dead and 1 indicates perfect health). In contrast to PROPr scores of a general population in which the mean is 0.52 and approximates a normal distribution, 50 patients who survived severe COVID-19 had PROPr scores reflecting a relatively uniform (rather than normal) distribution ( Figure 3 ).

In bivariate analyses, PROPr was unrelated to demographics or SVI. PROPr was significantly lower for patients with underlying medical conditions (hypertension, diabetes, heart failure, kidney disease and lung disease), transplant, and immunosuppression. Although not associated with ECI or SOFA score at ICU admission, PROPr was associated with mechanical ventilator and vasopressor requirement and a longer hospital length of stay (Table E2) .

PROPr was lower in patients who required oxygen or a caregiver at follow up, those readmitted to the hospital, and those who did not return to work or returned to a lower level of employment (Table E2) . Nearly a quarter (23.8%) of the ICU survivors had PROPr scores of ≤ 0.2. This group of patients reported fatigue, anxiety, pain, depression and social function more than one standard deviation worse than the general population and physical function nearly two standard deviations worse than the general population average. Nearly 40% required supplemental oxygen, 32% needed a caregiver and 23% had been readmitted to the hospital (Table E3) .

The mean sum of the four questions on the PCL-5 was 3.48 (SD 4.33), with 26 (20.3%) having scores ≥ 8, indicating that PTSD is probable ( Table 2 ). The mean score on the 3-item loneliness score was 4.4 (SD 2.0), with 38 (29.2%) patients indicating loneliness (score ≥ 6). Both PTSD and loneliness were significantly negatively correlated with PROPr (Table E3) . Nearly all surveyed, 125 (97.7%), were glad they received aggressive life-sustaining treatment, and 118 (89.4%) would choose to receive it again, if necessary (Table 2) . Including the 3 patients who wished they had received comfort care during their severe COVID-19 illness, there were 9 (6.8%) patients who would not want aggressive life-sustaining treatment again if they had recurrent severe COVID-19. Preference for aggressive care was not associated with PROPr (Table E3 ).

Few studies have evaluated the post-ICU course of adults in the U.S. who were critically ill due to COVID-19. Using both survey and medical record data, we found that survivors of severe COVID-19 had worse physical function, but similar anxiety, depression, fatigue, sleep disturbance, and pain interference scores compared to that of the U.S. general population. Moreover, the mean physical function score was less than one standard deviation below normal. In contrast to other studies, 7-11 we found evidence for debilitating physical and mental disability in only a small portion of our patient sample. The PROPr score was 0.46 (SD 0.3), which is worse than the mean PROPr score (0.52) observed in the US population; 44,50 however, this difference is less than the effect of chronic illness on PROPr (ie, emphysema 50 and dialysis 51 decrease PROPr by about −0.2). Taking into consideration that the majority of this patient population had at least one baseline co-morbid condition in addition to their risk for PICS, these findings are reassuring. Furthermore, PTSD and loneliness were uncommon, and most patients were able to return home to live independently. Thus, most patients who survive admission to the ICU for severe COVID-19 are in health states 6 months post-discharge that are not very different from that of the general U.S. population.

However, these overall statistics belie the substantial minority of survivors with significant health impacts from COVID-19 following intensive care, reflected by PROPr scores close to zero. Thirty-one patients had health utility scores ≤ 0.2; of which 21 patients had scores less than 0.1 (a score of 0 is comparable to dead). This substantial minority are often tethered to oxygen, newly require a caregiver, and commonly are unable to return to work. Demographic characteristics and social vulnerability do not distinguish this group of patients; compromised outcomes appear to be associated with underlying medical conditions and level of illness during the ICU stay (but not at entry). It is also notable that both Global Physical and Mental Health scores are lower than that of the U.S general population, indicating that the patients' overall perception of their health is lower than what is indicated by the PROMIS-29 domains. These findings may reflect the more difficult to define symptoms due to "long-COVID". 52, 53 Identification of factors that can prospectively identify patients who do poorly after severe COVID-19 remains urgently needed.

The overwhelming majority of patients who survived severe COVID were appreciative of the aggressive ICU care they received and would choose to undergo it again. This attitude contrasts starkly with the perspective of intensive care being largely futile commonly displayed in the lay media. [54] [55] [56] Interpretation of this finding should recognize that we were unable to obtain the perspectives of those who had the worst PROPr is a preference-based utility score for health states defined by 7 PROMIS domains (depression, fatigue, pain interference, physical function, sleep disturbance, ability to participate in social activities, and cognitive function). It ranges from −0.022 to 1.0 where <0 is worse than dead and 1.0 is perfect health. In a general U.S. population, the distribution of PROPr is nearly a normal distribution. A score of 40 is one SD lower than the mean of the reference population, while a score of 60 is one SD higher than the mean of the reference population. Higher scores represent worse symptomatology in anxiety, depression, fatigue, sleep disturbance, and pain interference. In physical function, social participation and cognition, a higher score signals better functioning. *p < 0.05 when compared to a mean of 50.

outcomes (ie, those too sick to respond and those who died in the ICU), though nearly half of the non-respondents reported that they were back to baseline health ( Figure 1 ). PROPr, which assigns a score of zero to dead, allows us to provide a global assessment on all patients who required ICU care for COVID-19. Including those who did died along with survey respondents, the overall PROPr is 0.29. This study has limitations as well as strengths. There was thorough follow-up of patients discharged from the ICU. The measures used in this study have been extensively evaluated and assess multiple dimensions of HRQOL. Limitations include the absence of physiologic pulmonary function assessments or diagnostic imaging, which were deferred given our focus is on patient-reported outcomes. Although demographics, clinical characteristics, and treatments were similar between responders and non-responders, non-responders were sicker according to the ECI and less socially vulnerable. Importantly, we lacked baseline HRQOL data prior to ICU admission, thus limiting our ability to assess changes in function, although we attempted to capture this in retrospective self-assessment in our survey questions, albeit 3 months into the study (eTable 1). Furthermore, our follow up is limited to six-months. Lastly, the single health system source of patients limits generalizability and our COVID-19 ICU survivor cohort was less sick than others in the literature as measured by SOFA, length of ICU stay, and proportion of patients ventilated. [57] [58] [59] In conclusion, patients who survive severe COVID-19 display marked heterogeneity in outcomes six months after hospitalization, with many being in health states that are similar to that of the general US population. However, there exists a substantial minority with severely compromised health status. Mechanisms are needed to identify patients treated for COVID-19 in the ICU who are at risk for severe outcomes.

COVID-19 Map -johns hopkins coronavirus resource center

Clinical features of patients infected with 2019 novel coronavirus in wuhan

Pathophysiology, transmission, diagnosis, and treatment of coronavirus disease 2019 (COVID-19): a review

Virology, transmission, and pathogenesis of SARS-CoV-2

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in wuhan, China: a descriptive study

COVID-19 in critically ill patients in the Seattle region -case series

Post-Discharge health Status and symptoms in patients with severe COVID-19

6-month Consequences of COVID-19 in patients discharged from hospital: a cohort study

Long-term consequences in critically ill COVID-19 patients: a prospective cohort study

Medium-term outcome of severe to critically ill patients with SARS-CoV-2 infection

Continuing cardiopulmonary symptoms, disability, and financial toxicity 1 month after hospitalization for third-wave COVID-19: early results from a US nationwide cohort

1-year Outcomes in hospital survivors with COVID-19: a longitudinal cohort study

Depression, posttraumatic stress disorder, and functional disability in survivors of critical illness in the BRAIN-ICU study: a longitudinal cohort study

Anxiety, depression and post-traumatic stress disorder management after critical illness: a UK multi-centre prospective cohort study

Co-Occurrence of post-intensive care syndrome problems among 406 survivors of critical illness

Quality of life in the five years after intensive care: a cohort study

The association of intensive care with utilization and costs of outpatient healthcare services and quality of life

Association of job characteristics and functional impairments on return to work after acute respiratory distress syndrome

Return to work after critical illness: a systematic review and meta-analysis

COVID-19: ICU delirium management during SARS-CoV-2 pandemic

Prolonged grief disorder following the coronavirus (COVID-19) pandemic

COVID-19 and unfinished mourning

Bereavement support on the frontline of COVID-19: recommendations for hospital clinicians

Association of social isolation with disability burden and 1-year mortality among older adults with critical illness

Long-term cognitive impairment after critical illness

Neurologic features in severe SARS-CoV-2 infection

The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies

Acute physiology and chronic health evaluation II score as a predictor of hospital mortality in patients of coronavirus disease 2019

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

A new elixhauser-based comorbidity summary measure to predict in-hospital mortality

CDC/ATSDR social vulnerability Index

Measuring community vulnerability to natural and anthropogenic hazards: the centers for disease control and prevention's social vulnerability Index

The patient-reported outcomes measurement information system (PROMIS): progress of an NIH roadmap cooperative group during its first two years

R)) adult health profiles: efficient short-form measures of seven health domains

Self-reported cognitive concerns and abilities: two sides of one coin?

Development of physical and mental health summary scores from the patient-reported outcomes measurement information system (PROMIS) global items

The patient-reported outcomes measurement information system (PROMIS) developed and tested its first wave of adult self-reported health outcome item banks

Relative to the general US population, chronic diseases are associated with poorer health-related quality of life as measured by the patient-reported outcomes measurement information system (PROMIS)

PROMIS Measures of pain, fatigue, negative affect, physical function, and social function demonstrated clinical validity across a range of chronic conditions

Healthmeasures scoring service powered by assessment center

PROMIS((R))-29 v2.0 profile physical and mental health summary scores

Estimation of a preferencebased summary score for the patient-reported outcomes measurement information system: the PROMIS((R))-preference (PROPr) scoring system

Computing PROPr utility scores for PROMIS(R) profile instruments. Value Health

Cross-sectional validation of the PROMIS-preference scoring system

The posttraumatic stress disorder checklist for DSM-5 (PCL-5): development and initial psychometric evaluation

Developing an optimal short-form of the PTSD checklist for DSM-5 (PCL-5)

A short scale for measuring loneliness in large surveys: results from two population-based studies

A short-form measure of loneliness

Dexamethasone in hospitalized patients with COVID-19

Cross-sectional validation of the PROMIS-preference scoring system by its association with social determinants of health

Evaluation of PROMIS preference scoring system (PROPr) in patients undergoing hemodialysis or kidney transplant

Confronting our next national health disaster -long-haul COVID

Understanding long COVID: a modern medical challenge

A bridge between life and death: most COVID-19 patients put on ventilators will not survive. USA Today

Elderly COVID-19 patients on ventilators usually do not survive

Ventilators Are No Panacea For Critically Ill COVID-19 Patients: NPR

Post-intensive care syndrome after a critical COVID-19: cohort study from a Belgian follow-up clinic

Functional status of mechanically ventilated COVID-19 survivors at ICU and hospital discharge

Quality of life of COVID-19 critically ill survivors after ICU discharge: 90 days follow-up

This work was supported by a donation from the Porath Family Trust. This research was also supported by the National Institutes of Health National Center for Advancing Translational Science (NCATS) UCLA CTSI Grant Number UL1TR001881. Funders played no role in the design and conduct of the study, analysis and interpretation of the data, or preparation, review, or approval of the manuscript. We are indebted to Dr Robert H. Brook for the study idea and comments on the report.

Study concept and design: Neville, Hays, Wenger. Data collection: Gonzalez, Hong, Yamamoto, Santoso, Kung, Neville. Analysis and interpretation of data: Chen, Tseng, Neville, Hays, Wenger. Statistical Analysis: Chen, Neville. Drafting of manuscript: Neville. Critical revision of manuscript: Neville, Hays, Gonzalez, Hong, Wang, Chang, Qadir, Schwab, Yamamoto, Kung, Santoso, Chen, Tseng, Wenger. Neville and Wenger had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. The Authors declare that there is no conflict of interest.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Not applicable, because this article does not contain any studies with human or animal subjects.

The author(s) received no financial support for the research, authorship, and/or publication of this article. 

Supplemental material for this article is available online.