Inequalities in access to paid sick leave among workers in England and Wales

Abstract Background It is poorly understood which workers lack access to sick pay in England and Wales. This evidence gap has been of particular interest in the context of the Covid‐19 pandemic given the relationship between presenteeism and infectious disease transmission. Method This cross‐sectional analysis (n = 8874) was nested within a large community cohort study based across England and Wales (Virus Watch). An online survey in February 2021 asked participants in work if they had access to paid sick leave. We used logistic regression to examine sociodemographic factors associated with lacking access to sick pay. Results Only 66% (n = 5864) of participants reported access to sick pay. South Asian workers (adjusted odds ratio [OR] 1.40, 95% confidence interval [CI] 1.06–1.83) and those from Other minority ethnic backgrounds (OR 2.93, 95% CI 1.54–5.59) were more likely to lack access to sick pay compared to White British workers. Older workers (OR range 1.72 [1.53–1.93]–5.26 [4.42–6.26]), workers in low‐income households (OR 2.53, 95% CI 2.15–2.98) and those in transport, trade, and service occupations (OR range 2.03 [1.58–2.61]–5.29 [3.67–7.72]) were also more likely to lack access to sick pay compared respectively to workers aged 25–44, those in high income households and managerial occupations. Discussion Unwarranted age and ethnic inequalities in sick pay access are suggestive of labour market discrimination. Occupational differences are also cause for concern. Policymakers should consider expanding access to sick pay to mitigate transmission of Covid‐19 and other endemic respiratory infections in the community, and in the context of pandemic preparation.


| INTRODUCTION
An estimated two million employees in the UK do not earn enough to be eligible for statutory sick pay. 1,2For those who do, mandatory paid sick leave as a proportion of previous earnings is among the lowest of the countries constituting the Organisation for Economic and Co-operation and Development (OECD). 3 Employees on casual or flexible contracts (including those on zero hour contracts) have a legal right to statutory sick pay if they are able prove their average earnings are above the eligibility threshold, which can be challenging.Statutory sick pay in the UK does not extend to those who are self-employed.
[4][5] From a living standards perspective, limited access to paid sick leave risked income loss and destitution for low income households during periods of Covid-19 illness or self-isolation.From a public health perspective, the risk of income loss may encourage presenteeism and drive transmission of Covid-19 and other acute respiratory pathogens in the community.For example, fear of income loss was a common reason given for non-adherence with self-isolation among workers during the 2003 SARS outbreak in Toronto. 6Furthermore, the introduction of mandatory sick pay policies across US states has been associated with lower disease rates in the community and mitigation of flu epidemics. 7,8Understanding factors associated with access to sick pay is consequently important to address both endemic Covid-19 and respiratory viral transmission and for pandemic planning.
Official UK government estimates on sick pay coverage, which come from a 2014 survey of 2030 employees in Great Britain eligible for sick pay, found that 26% receive the statutory minimum rate, 57% receive sick pay above this minimum and 17% do not know. 9However, this survey does not capture features of workers lacking access to paid sick leave.The characteristics of that group can only indirectly be inferred by examining workers earning below the income threshold to access statutory sick pay using data from the Office of National Statistics' Labour Force Survey. 10This assumes all workers earnings above the eligibility threshold will automatically have access to paid sick leave.It is therefore limited in its ability to infer unwarranted labour market inequalities in access to paid sick leave.
Inequalities in access to sick pay among UK workers are poorly understood.In that context, this short communication exploits data from the Virus Watch community cohort study to examine factors associated with access to sick pay.

K E Y W O R D S
COVID-19, health equity, health policy, public health

| Study design and procedure
Data were collected as part of the Virus Watch study, a prospective household cohort study of Covid-19 transmission in England and Wales.The full study design and methodology has been described elsewhere. 11Participants were recruited using several methods including post, social media, SMS, and letters from General Practices.Eligibility criteria were consent to participate from all household members, household size up to 6 people (due to limitations of the online survey infrastructure), access to the internet and email, and at least one household member able to complete English-language surveys.
After enrolling in the study, an initial baseline survey collected demographic, occupational, social and medical history data from participants.Monthly surveys collected detailed information on social, clinical and behavioural factors relevant to the phase of the pandemic as the time of collection.A February 2021 survey focussed on financial and work-related determinants of COVID-19 illness included an item relating to sick pay access.The analysis in this communication is limited to adults (>16) who responded to this February 2021 survey and reported being employed or self-employed.

| Exposures
The exposures of interest were demographic and social variables potentially associated with sick pay access among working adults.These were age; sex; region; ethnicity; household income; and occupation.Data on exposure variables were collected through the baseline survey completed on entry into the Virus Watch study.

| Outcome
The outcome of interest was self-reported access to sick pay.Participants were able to report 'Yes', 'No', 'Unsure' and 'Not applicable' when asked if they had access to paid sick leave if required.Those who responded 'Not applicable' were excluded from this study.Outcome data was collected between 17 and 28 February 2021.

| Statistical analysis
To model the association between covariates and access to sick pay, we conducted univariable and multivariable fixed effects logistic regression models in R 4.0.3.A sensitivity analysis controlling for self-employment status (employed vs. self-employed) was also conducted.

| Patient and public involvement
The study team worked with a community advisory group to inform equity-related aspects of recruitment, design and dissemination.This advisory group, consisting of lay members of the public, community leaders, charities and policy experts, suggested this analysis.T A B L E 1 Sociodemographic characteristics of survey respondents.

| RESULTS
(Continues) A sensitivity analysis controlling for self-employment status was consistent with these findings (Table 4).
All observed differences in sick pay access between age, ethnic and income groups persisted.Indeed, ethnic contrasts in access to sick pay heightened.After adjusting for self-employment status, 16-24-year-old workers T A B L E 2 Description of participant characteristics by self-reported access to sick pay. (Continues)

| DISCUSSION
Our findings reveal substantial inequalities in access to paid sick leave among workers in England and Wales.Older workers, certain minority ethnic groups, workers in low-income households and trade, transport and service occupations were more likely to lack access to paid sick leave.These differences were not explainable by self-employment status.
This cross-sectional analysis was nested in the larger Virus Watch prospective cohort study.By harnessing data already collected by this large cohort study, this analysis was able to fill an important policy-relevant evidence gap on inequalities in sick pay access without needing to duplicate efforts to generate research data.Individuals in the Virus Watch study are well distributed across England and Wales and the cohort is diverse in terms of age, sex, ethnicity, and socioeconomic composition. 11Given participation in the Virus Watch study is voluntary and sampling non-random, the cohort likely oversamples people concerned with COVID-19 and participants were more likely to be White British, over the age of 65 and have a higher income than the general population.Our multivariable regression analysis adjusted for age, ethnicity and income to address this sampling bias, but any residual confounding is likely to mean our findings overestimate the true magnitude of age disparities in sick pay access and underestimate ethnic and income disparities in sick pay access.
There is little public data or previous literature on sick pay coverage in the UK.A descriptive analysis of 3974 UK workers found leisure and personal services, outdoor trades, transport and mobile machinery and indoor trades occupations have the lowest rates of paid sick leave coverage, corroborating the findings of our study. 12A 2014 survey of 2030 employees by the Department of Work and Pensions was limited to employees eligible to access sick pay. 9It reported descriptive differences in the amount of sick pay workers receive, with a greater proportion of older workers and those in leisure and personal service occupations reporting access to only the minimum statutory rate.There has been greater study of disparities in sick leave access in the US, where ethnic inequalities have been documented. 13,14 is unsurprising that those in low-income households are more likely to lack access to sick pay than those in high-income households given statutory sick pay entitlement in the UK is conditional on earning above an income T A B L E 4 Sensitivity analysis multivariable logistic regression model including self-employment status as a covariate.
threshold.Inequalities in sick pay access between age and ethnic groups that cannot be explained by differences in income, occupation and employment status are suggestive of age-and race-based discrimination in the labour market.Occupations with elevated odds of lacking access to sick pay relative to managerial occupations are classifiable as manual occupations and unskilled non-manual occupations according to the widely used Goldthorpe class scheme. 15These occupations are referred to as Working Class in the Goldthorpe occupation-based class taxonomy. 16at working class occupations are most likely to lack access to sick pay is cause for concern with regard to inequalities in the labour market.
In the context of the Covid-19 pandemic, fear of income loss is likely to encourage presenteeism and SARS-CoV-2 transmission.A recent study of care homes in England found lower rates of SARS-CoV-2 transmission from staff when they had access to sick pay, compared to care homes where staff lacked access to statutory sick pay. 17Future research should investigate the relationship between access to sick pay and presenteeism in the community setting during the Covid-19 pandemic.
More broadly, paid sick leave has been judged an effective intervention to reduce transmission of SARS-CoV-2 across OECD countries. 3The UK government made some changes to statutory sick pay when the pandemic began by allowing eligible employees to receive sick pay during periods of self-isolation in addition to confirmed Covid-19 illness, and to receive sick pay from the first day of illness or self-isolation, rather than from the fourth day of illness as is the case for other illnesses.However, unlike around half of other OECD countries, the UK has not altered the wage replacement rate, nor has it modified the eligibility criteria to expand access to statutory sick pay.Given the inequalities we highlight in this paper, improving access to sick pay should be both an employment and health policy priority as Covid-19 becomes an endemic disease and in the context of planning for future public health emergencies.

Ms. Wing Lam
2.03 [1.58-2.61]-5.29 [3.67-7.72])were also more likely to lack access to sick pay compared respectively to workers aged 25-44, those in high income households and managerial occupations.Discussion: Unwarranted age and ethnic inequalities in sick pay access are suggestive of labour market discrimination.Occupational differences are also cause for concern.Policymakers should consider expanding access to sick pay to mitigate transmission of Covid-19 and other endemic respiratory infections in the community, and in the context of pandemic preparation.

Table 1
reports respondents' characteristics (n = 8874).Of respondents, 5864 (66%) reported having access to paid sick leave, 2218 (25%) reported no access to paid sick leave and 792 (8.9%) were unsure.Table2describes characteristics of those with and without access to sick pay.

Table 3
presents results of the univariable and multivariable logistic regression models.In the multivariable model, workers over the age of 65 (odds ratio [OR] 5.26, 95% CI 4.42-6.26)and between ages of 45 and 64 (OR 1.72, 95% CI 1.53-1.93)had greater odds of lacking access to sick pay in reference to workers aged 25-44.South Asian workers (OR 1.40, 95% CI 1.06-1.83)and 'Other minority ethnic' workers (OR 2.93, 95% CI 1.54-5.59)also had greater odds of lacking access to sick pay compared to White British workers.It is worth noting that although not always statistically significant, point estimates for workers from most minority ethnic backgrounds indicated elevated odds of lacking access to sick pay compared to White British workers.People in low income households were more likely to lack access to sick pay compared to those high income households, with households earning under £25,000 (OR 2.53, 95% CI 2.15-2.98)and households earning £25,000-£49,999 (OR 1.43, 95% CI 1.25-1.63)at greater odds of lacking access to sick pay if required than those in households earning above £75,000.

Access to sick pay Characteristic Yes, N = 5864 a No, N = 2218 a Unsure, N = 792 a
OR 1.73, 95% CI 1.25-2.38)alsohadgreaterodds of lacking access to sick pay compared to 24-44-year-old workers.With the exception of outdoor trade occupations, all observed occupational differences in sick pay access persisted.an(%).T A B L E 1(Continued)

Access to sick pay Characteristic Yes, N = 5864 a No, N = 2218 a Unsure, N = 792 a
Univariable and multivariable logistic regression models examining the relationship between sociodemographic characteristics and lacking access to sick pay.
a n (%).T A B L E 2 (Continued)Abbreviations: CI, confidence interval; OR, odds ratio.aOR>1indicates increased odds of lacking access to sick pay.T A B L E 3a OR >1 indicates increased odds of lacking access to sick pay.
Erica Fong is a research assistant at the UCL Institute of Health Informatics.She holds an MSc Applied Infectious Disease Epidemiology and an MSc in Health Data Science from the London School of Hygiene and Tropical Medicine.Her research interests include public health data science, health equity, and infectious disease epidemiology.Dr. Ellen Fragaszy is an honorary senior research fellow at the UCL Institute of Health Informatics and a course director at the London School of Hygiene and Tropical Medicine.She received her PhD in Epidemiology from University College London.Her research interests include infectious disease epidemiology and respiratory viral infections.Mr. Cyril Geismar is a research assistant at the UCL Institute of Health Informatics.He is also now undertaking a PhD in Infectious Disease Epidemiology at Imperial College London.His research interests include public health data science, epidemiological software and methods, and infectious disease epidemiology.Dr. Susan Hoskins was a statistician at the UCL Institute of Epidemiology and Health Care.She received her PhD from University College London.Her research interests include statistical methodology and infectious disease epidemiology.Dr. Madhumita Shrotri is a public health registrar at University College London.She received her medical degree from University College London and her Masters in Public Health from the London School of Hygiene and Tropical Medicine.Her research interests include health equity and infectious disease epidemiology.Dr. Jana Kovar is a scientific project manager at the UCL Institute of Health Informatics.She received her PhD in Anatomy and Human Biology from the University of Western Australia.Her research interests include public health data science.Dr. Anna Aryee is specialist trainee in Infectious Diseases & Microbiology, a previous NIHR Academic Clinical Fellow and a current UCL Wellcome Clinical PhD Fellow at the UCL Institute of Health Informatics.Her research interests include health equity and infectious diseases.Prof. Andrew C. Hayward is a Professor of Infectious Disease Epidemiology and Inclusion Health at the UCL Institute for Epidemiology and Health Care.He qualified in Medicine at United Medical and Dental Schools London, completed an MSc in Communicable Disease Epidemiology at the London School of Hygiene and Tropical Medicine, and a Doctorate in Medicine at University College London.He is founder and Co-Director of the UCL Collaborative Centre for Inclusion Health CCIH which launched in 2019.His research interests include health equity, inclusion health, and infectious disease epidemiology.Professor of Public Health Data Science and Wellcome Trust Clinical Research Career Development Fellow at the UCL Institute of Health Informatics.He qualified in medicine from University College London in 2007, gained an MSc in Epidemiology at The London School of Hygiene and Tropical Medicine in 2010 and completed his PhD at University College London in 2015.His research interests include health equity, inclusion health, and public health data science.
Dr. Annalan M. D. Navaratnam is an NIHR Academic Clinical Fellow Public Health and a Clinical Fellow at the UCL Institute of Health Informatics.He received his medical degree from Imperial College London and a PGCert in clinical education from King's College London.His research interests include public health data science and health equity.Prof. Robert W. Aldridge is a