key: cord-0863391-w2wj55t8
authors: Okawara, M.; Ishimaru, T.; Tateishi, S.; Hino, A.; Tsuji, M.; Ogami, A.; Nagata, T.; Matsuda, S.; Fujino, Y.
title: Association between interruption to medical care and sickness presenteeism during the COVID-19 pandemic: a cross-sectional study in Japan
date: 2021-08-18
journal: nan
DOI: 10.1101/2021.08.14.21261996
sha: fa71e31ffe016106b23079ce9d1eabfc43e8baa6
doc_id: 863391
cord_uid: w2wj55t8

Objective: This study examined the relationship between interruption to routine medical care during the coronavirus disease 2019 pandemic and sickness presenteeism in Japan. Methods: An internet monitor questionnaire was conducted. Data from 27,036 people were analyzed. Interruption to medical care was defined based on the response "I have not been able to go to the hospital or receive treatment as scheduled." The number of sickness presenteeism days in the past 30 days was employed as the primary outcome. A zero-inflated negative binomial model was used for analysis. Results: The incidence rate ratio was significantly higher among workers who experienced interrupted medical care (2.26; 95% confidence interval: 2.03-2.52) than those who did not require routine medical care. Conclusions: This study suggests the importance of continuing necessary treatment during a pandemic to prevent presenteeism.

Sickness presenteeism is an increasingly important issue in occupational health. Aronsson 52 defined sickness presenteeism as "people, despite complaints and ill health that should prompt rest 53 and absence from work, still turning up at their jobs" 1 . Sickness presenteeism is the result of a choice 54 made by a worker with ill-health, disease, or capacity loss between sickness presenteeism and 55 sickness absence 2 . This decision is influenced by the individual's personality, values, economic 56 status, workplace "demands for presence" and support for adaptation, and national culture and 57 employment customs 2 . Evidence suggests that sickness presenteeism can lead to sickness absence 58 and future worsening of physical and mental health conditions 3-9 . In addition, the impact of working 59 while ill on productivity is also gaining attention, especially in the US 10,11 . A variety of diseases and 60 health conditions have been found to be associated with sickness presenteeism, suggesting the 61 importance of managing disease and maintaining good condition 4, 12 . 62

Under the coronavirus disease 2019 (COVID-19) pandemic, there is concern that both 63 organizational and individual factors will increase sickness presenteeism above that observed under 64 normal conditions [13] [14] [15] . Organizational factors that may lead to more sickness presenteeism include 65 increased workload and working hours per person, increased work pressure due to a manpower 66 shortage in the organization and changes in work demands on short notice during the pandemic. This 67 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint increased workload on workers can lead to a negative work culture around taking sick leave, such as 68

where workers who choose to work while ill are valued for their loyalty to the company and 69 motivation to work, thus promoting sickness presenteeism 16 . Examples of individual factors that may 70 increase sickness presenteeism include worsening economic situations and job insecurity; increased 71 telecommuting, which can make it easier for workers to work even while sick; the impact of the 72 pandemic on anxiety and mental health; worsening of health conditions and diseases due to lifestyle 73 changes; and worsening of chronic diseases due to the inability to access medical resources. All of 74 these factors are expected to lead to an increase in sickness presenteeism. 75

Interruption to medical care is an important problem in the COVID-19 pandemic. Access to 76 necessary routine medical care and medical resources is reportedly being affected in many countries 77 around the world 17-19 . In Japan, there is data showing that the number of prescriptions issued has 78 decreased 20 . There are multiple reasons for such interruptions to medical care, including fear of 79 being infected with COVID-19 when leaving the house to visit a hospital, worsening personal 80 economic situations, and shortages in medical personnel, all of which affect the treatment schedule 81 for chronic diseases 21 . Interruption to medical care can adversely affect management of chronic 82 diseases and delay the detection and treatment of new diseases 22 . In fact, excess deaths unrelated to 83 COVID-19 have been reported 23 . Thus, interruption to medical care during the COVID-19 pandemic 84 may lead to worsening of non-COVID-19 diseases and health conditions. 85 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

We performed a cross-sectional study based on baseline survey data obtained in the 96

Collaborative Online Research on the Novel-Coronavirus and Work (CORoNaWork) project, a 97 prospective cohort study that performed a questionnaire-based survey of Internet monitors to 98 determine the effect of the COVID-19 pandemic on workers' health. Before completing the online 99 survey, participants read a description of the survey's aims and details about the handling of their 100 information. Only participants who agreed with the contents of the description were allowed to 101 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint participate. Participation was anonymized. This study was approved by the Ethics Committee of the 102 University of Occupational and Environmental Health, Japan (Approval No. R2-079 and R3-006). 103

The baseline survey was conducted from December 22 to 26, 2020. A total of 33,302 104 participants aged from 20 and 65 years who indicated they were working when completing the 105 survey were included. Participants were selected such that sex and occupation (office and non-office (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint following question: "Do you have a disease that requires regular visits to the hospital or treatment?" 119

Responses were "I do not have any such disease"; "I am able to go to the hospital or receive 120 treatment as scheduled"; "I have not been able to go to the hospital or receive treatment as 121 scheduled." 122

Those who answered "I do not have any such disease" were defined as workers who did not 123 require routine medical care, and thus did not have any disease that requires hospital visits or 124 treatment. Those who answered "I am able to go to the hospital or receive treatment as scheduled" 125 were defined as workers who used medical care. Those who answered "I have not been able to go to 126 the hospital or receive treatment as scheduled" were defined as workers who experienced interrupted 127 medical care. 128 129

Respondents' number of sickness presenteeism days was ascertained based on the following 131 question and used as the primary outcome: "In the last 30 days, how many days have you worked 132 (including work from home) despite feeling that you really should have taken sick leave due to your 133 state of health?" 134 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The cumulative infection rate of COVID-19 in the province of residence was employed as a 141 community-level variable. 142

To control for potential confounders, we also asked participants to indicate their main 143 symptoms using the following question: "Which of the following conditions or body parts give you 144 the most trouble during your work?" The options were "No problem"; "pain"; "movement"; 145 "tightness, loss of energy, appetite, fever, dizziness, or feeling poor"; "toileting or elimination"; 146 "mental health"; "skin, hair, or beauty"; "sleep"; "eyes"; "nose"; "ears"; and "other." 147 148 Statistical analysis 149 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint with median and interquartile range (IQR). Other covariates were expressed as categorical variables 151 using percentages. 152

We compared the results of linear regression, Poisson regression, Zero-inflated Poisson 153 regression (ZIP), negative binomial regression, and Zero-inflated Negative Binomial regression 154 (ZINB) as statistical models, as they treat the number of sickness presenteeism days as continuous 155 count data. Negative binomial regression can handle over-dispersed data, where the variance is much 156 higher than the mean, which cannot be assumed in Poisson distribution. Further, to handle data with 157 excess zeros, which indicates a population at low risk of sickness presenteeism, we used a 158 zero-inflated model. In addition to dealing with the excess zeros that often occur in count data, a 159 zero-inflated model has also been proposed as a way to handle the difficulty of defining sickness 160 presenteeism cutoffs 24,25 . As a measure of model fitness, we compared the Akaike's Information 161 Criterion (AIC), and ultimately adopted the ZINB model. 162

ZINB regression analysis was conducted with the number of continuous sickness 163 presenteeism days as the dependent variable, the respondents' category of treatment status as an 164 independent variable, and the number of days worked per week as an offset variable. 165 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint household income, education, company size, cumulative infection rate by prefecture, and main 167

In further analysis, we estimated the margins of sickness presenteeism days for each 169 treatment status and symptom. First, we used the same statistical model as that in the main analysis. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint the survey company and 6,051 satisfied the exclusion criteria during data cleaning), leaving a total of 182 27,036 responses for analysis. Because all responses were mandatory, there were no missing data in 183 this study. 184

The demographic and sociological characteristics of the analyzed population are shown in 185 Table 1 . A total of 13,814 (51%) were men, with a median age of 48 years (IQR: 39-55). Of the total 186 population, 17,526 (65%) were workers who did not require routine medical care, 8,451 (31%) were 187 using medical care as scheduled, and 1,059 (4%) experienced interrupted medical care. The overall 188 median number of sickness presenteeism days was 0.0 (IQR: 0-2). The distribution of sickness 189 presenteeism is shown for the three treatment statuses in a histogram in Figure 1 . 190 The association between the number of sickness presenteeism days and treatment status is 191 shown in Table 2 p<0.001) compared to workers who did not require routine medical care. 197 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint

The association between the number of sickness presenteeism days and participants' main 198 symptoms is shown in Table 3 . There were significant associations between the number of sickness 199 presenteeism days and some symptoms using the model presented in Table 2 showed that there were significant differences between workers with the same symptoms who did 212 and did not require routine medical care, and between workers with the same symptoms who 213 experienced interruption to medical care and who did not require routine medical care. For example, 214 the number of sickness presenteeism days significantly differed between those with mental health 215 symptoms who used medical care and those with mental health symptoms who did not require 216 routine medical care (p<0.001). 217

This study demonstrated an association between treatment interruption and sickness 220 presenteeism among Japanese workers in the COVID-19 pandemic. Compared to workers who did 221 not require routine medical care, workers who had diseases that required routine medical care 222 reported more days of sickness presenteeism, and those who experienced interrupted medical care 223

reported even more such days. Furthermore, our findings reveal differences in the occurrence of 224 sickness presenteeism depending on workers' symptoms. 225

We found that workers who experienced interrupted medical care had increased sickness 226 presenteeism. This is because appropriate treatment can improve work function and productivity by 227 improving workers' health and subjective symptoms 10,27 . This is supported by the fact that workers 228 who used medical care reported fewer sickness presenteeism days than those who experienced 229 interrupted medical care, although workers who used medical care had a higher incidence of 230 sickness presenteeism than workers who did not require routine medical care. Employees who 231 experience interrupted treatment for chronic diseases may be forced to return to work due to fear of 232 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint being laid off, depending on the financial situation of their workplace during the pandemic. It is thus 233 important to continue regular treatment during the COVID-19 pandemic to manage disease and 234 maintain good condition 28 . 235

We found that the occurrence of sickness presenteeism depends on the type of symptoms 236 experienced by workers. Workers with symptoms related to mental health problems, loss of 237 energy/fever, and body movements reported more sickness presenteeism than those who reported 238 having "no problem." In contrast, workers with symptoms related to sleep, pain, and elimination 239 reported comparable or fewer days of sickness presenteeism to those who reported having "no 240 problem." Sickness presenteeism is the result of a worker's choice to be absent from work or to 241 attend work despite being unwell. Many previous studies have evaluated sickness presenteeism 242 based on whether or not workers "worked one or more days in a certain period of time with a health 243 condition for which they think they really should be absent" 2,29 . However, workers experiencing 244 symptoms not typically associated with sickness presenteeism may not consider their symptoms 245 suitable for an absence from work, and thus may not have indicated that they experienced sickness 246 presenteeism. For example, symptoms related to beauty are not directly related to an individual's 247 ability to work; sleep may be considered something that the individual simply needs to get more of 248 on holidays; and chronic pain may be considered an instruction to move rather than rest. However, 249 some symptoms that can pose a health risk are also unlikely to be recognized as contributing to 250 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint sickness presenteeism. These include symptoms that can lead to delayed detection or worsening of a 251 disease if left untreated when rest or treatment is in fact required. For individuals with insomnia or 252 elimination symptoms, for example, resting or visiting a hospital when feeling unwell can lead to 253 prevention or early diagnosis and treatment of mental health problems or inflammatory bowel 254 disease, respectively. While forcing oneself to work with such symptoms can pose a health risk, 255 workers may not consider this sickness presenteeism due to differences in interpretation of "health 256 conditions that require absence from work." This is an important point when evaluating sickness 257

We also found that the impact of continuing treatment on the prevention of sickness 259 presenteeism varied by symptom. Sickness presenteeism was more frequent in workers who 260 experienced interrupted medical care with symptoms related to tightness and loss of energy, toileting 261 and elimination, sleep, and eyes than those who did not require routine medical care. In contrast, no 262 difference in sickness presenteeism was observed between workers who used medical care and those 263 who did not require routine medical care, suggesting the importance of continuing necessary routine 264 medical care for preventing sickness presenteeism due to these symptoms. Workers who experienced 265 interrupted medical care with these symptoms may be able to reduce the incidence of sickness 266 presenteeism by continuing appropriate treatment to maintain and improve their health condition. 267

Symptoms that led to more sickness presenteeism in both workers who used medical care and 268 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint workers who experienced interrupted medical care compared to workers who did not require routine 269 medical care were pain and mental health-related symptoms. For these symptoms, sickness 270 presenteeism remained high even with continued treatment, indicating the need to identify 271 appropriate treatment and manage one's daily health condition in addition to continuing treatment. In 272 contrast, symptoms that led to comparable sickness presenteeism in workers who used medical care 273 and workers who experienced interrupted medical care compared to workers who did not require 274 routine medical care were related to movement and mobility; skin, hair, or beauty; nose; ears; and 275 other symptoms, for which treatment is ineffective to prevent sickness presenteeism. The lack of a 276 difference in sickness presenteeism for these symptoms may be due to the fact that individuals do 277 not consider these symptoms sufficiently adverse to require an absence from work. Alternatively, 278 some individuals, such as those with physical movement symptoms, may experience chronic 279 symptoms for which support and adaptive behaviors have already been put into place; thus, whether 280 or not these individuals experience sickness presenteeism may be unrelated to their treatment status. 281

Thus, the impact of continuing treatment on sickness presenteeism may be related to whether an 282 individual considers their symptoms to be sufficiently adverse to require an absence from work, or 283 whether or not the symptoms can be improved with treatment. 284

There are several limitations to this study. First, we did not obtain detailed information 285 related to treatment interruptions, including the type of disease, duration, and reasons for interruption. 286 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint patient-related reasons (e.g., economic situation and anxiety) or hospital-related reasons (e.g., 288

schedule adjustment). Second, interruptions to treatment may be the result of better disease control 289 and improved health. It is unclear how these factors would affect the occurrence of sickness 290

presenteeism. Finally, we did not consider all possible confounders affecting sickness presenteeism 291 because we did not obtain information on some confounders, such as job insecurity, annual leave 292 rights, and the culture around employment and sick leave in each company. 293 294

Interruption to medical care during the COVID-19 pandemic was associated with the 296 occurrence of sickness presenteeism. This study demonstrates the importance of maintaining one's 297 health condition and continuing necessary treatment even during an infectious disease pandemic. 298 299 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 18, 2021. ; https://doi.org/10.1101/2021.08.14.21261996 doi: medRxiv preprint 1 

Sick but yet at work. An empirical study of sickness 301 presenteeism

Sickness presenteeism: prevalence, attendance-pressure factors, and 303 an outline of a model for research

The consequences of sickness presenteeism on health and wellbeing 305 over time: A systematic review

Presenteeism in the workplace: A review and research agenda

Working while ill 309 as a risk factor for serious coronary events: the Whitehall II study

The relation between 312 presenteeism and different types of future sickness absence

Sickness presenteeism today, 314 sickness absenteeism tomorrow? A prospective study on sickness presenteeism and future 315 sickness absenteeism

Associations between health and combinations of sickness presence 317 and absence

Consequences of sickness presence and sickness absence on health 319 and work ability: a Swedish prospective cohort study

The cost and impact of health conditions on 322 presenteeism to employers: a review of the literature

Working while sick in context of regional 329 unemployment: a Europe-wide cross-sectional study

Socioeconomic Factors Associated With an Intention to Work 332

While Sick From COVID-19

Presenteeism during the COVID-19 pandemic: risks and solutions

Sickness presenteeism at work: prevalence, costs and management

Reduced access to care among older American adults during

CoVID-19 pandemic: results from a prospective cohort study

Delay or 341 avoidance of medical care because of COVID-19

No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted August 18, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted August 18, 2021. ; All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Figure 1 . Sickness presenteeism days under each treatment status 1 2 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Figure 2 . Predictive margins of each treatment status and symptom 1 2 *others include loss of energy, appetite, fever, dizziness, or feeling poor 3 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint this version posted August 18, 2021.; https://doi.org/10.1101/2021.08.14.21261996doi: medRxiv preprint