Sickness absence diagnoses among abstainers, low-risk drinkers and at-risk drinkers: consideration of the U-shaped association between alcohol use and sickness absence in four cohort studies

Associations between commute mode use and self-rated health and work ability among Finnish public sector employees

AIM

To determine the extent to which level of active commute mode use is associated with self-rated health and work ability.

METHODS

The data were sourced from the Finnish Public Sector Study survey in 2020 (n = 38,223). The associations between active commuting - assessed with the frequency of using active commute modes - and self-rated health and work ability were examined with negative binomial regression analyses. Passive commuting and low-to-moderate levels of active commuting were compared with active commuting, and the models were adjusted for sociodemographic factors, working time mode, and lifestyle risk factors. We also assessed separate associations between walking and cycling as a mode of commuting by additionally considering the commuting distance and the outcomes.

RESULTS

After adjustment, when using active commuters as a reference, passive commuters had a 1.23-fold (95% confidence intervals (CI) 1.19 to 1.29) risk of suboptimal self-rated health and a 1.18-fold (95% CI 1.13 to 1.22) risk of suboptimal work ability. More frequent and/or longer distance by foot and especially by bicycle, was positively associated with health and work ability. Never commuting by bicycle was associated with a 1.65-fold (95% CI 1.55 to 1.74) risk of suboptimal health and a 1.27-fold (95% CI 1.21 to 1.34) risk of suboptimal work ability when using high-dose bicycle commuting as a reference.

CONCLUSIONS

Passive commuting was associated with suboptimal self-rated health and suboptimal work ability. Our results suggest that using active commute modes, particularly cycling, may be beneficial for employee health and work ability.

Impact of health risk factors on healthcare resource utilization, work-related outcomes and health-related quality of life of Australians: a population-based longitudinal data analysis

Background

Health risk factors, including smoking, excessive alcohol consumption, overweight, obesity, and insufficient physical activity, are major contributors to many poor health conditions. This study aimed to assess the impact of health risk factors on healthcare resource utilization, work-related outcomes and health-related quality of life (HRQoL) in Australia.

Methods

We used two waves of the nationally representative Household, Income, and Labor Dynamics in Australia (HILDA) Survey from 2013 and 2017 for the analysis. Healthcare resource utilization included outpatient visits, hospitalisations, and prescribed medication use. Work-related outcomes were assessed through employment status and sick leave. HRQoL was assessed using the SF-6D scores. Generalized estimating equation (GEE) with logit or log link function and random-effects regression models were used to analyse the longitudinal data on the relationship between health risk factors and the outcomes. The models were adjusted for age, sex, marital status, education background, employment status, equilibrium household income, residential area, country of birth, indigenous status, and socio-economic status.

Results

After adjusting for all other health risk factors covariates, physical inactivity had the greatest impact on healthcare resource utilization, work-related outcomes, and HRQoL. Physical inactivity increased the likelihood of outpatient visits (AOR = 1.60, 95% CI = 1.45, 1.76 p < 0.001), hospitalization (AOR = 1.83, 95% CI = 1.66-2.01, p < 0.001), and the probability of taking sick leave (AOR = 1.31, 95% CI = 1.21-1.41, p < 0.001), and decreased the odds of having an above population median HRQoL (AOR = 0.48, 95% CI = 0.45-0.51, p < 0.001) after adjusting for all other health risk factors and covariates. Obesity had the greatest impact on medication use (AOR = 2.02, 95% CI = 1.97-2.29, p < 0.001) after adjusting for all other health risk factors and covariates.

Conclusion

Our study contributed to the growing body of literature on the relative impact of health risk factors for healthcare resource utilization, work-related outcomes and HRQoL. Our results suggested that public health interventions aim at improving these risk factors, particularly physical inactivity and obesity, can offer substantial benefits, not only for healthcare resource utilization but also for productivity.

Dose-response relationship between alcohol consumption and workplace absenteeism in Australia

INTRODUCTION

Workplace absenteeism is a burden in Australia. The estimated productivity losses due to alcohol were around $4.0 billion in 2017, with absenteeism driving 90% of these costs. We aim to determine the dose-response relationship between average daily alcohol consumption and heavy episodic drinking (HED) frequency and workplace absenteeism amongst Australian workers.

METHODS

We used the 2019 National Drug Strategy Household Survey of Australian employed workers aged ≥20 years to 69 years old. Respondents' average daily alcohol consumption was categorised into four: abstainers, light to moderate (1-20 g of alcohol/day), risky (>20-40 g of alcohol/day) and high-risk (>40 g of alcohol/day). HED was classified into four frequency measures (never, less than monthly, monthly, weekly). The outcome variables came from dichotomised measures of: (i) absence due to alcohol consumption; and (ii) broader sickness absence-absence due to illness or injury in the previous 3 months.

RESULTS

Risky (adjusted odds ratio 4.74 [95% CI 2.93-7.64]) and high-risk drinking (adjusted odds ratio 6.61 [95% CI 4.10-10.68]) were linked to increased odds of alcohol-related absence. Higher HED frequency was significantly associated with alcohol-related and broader sickness absenteeism. No significant associations exist between regular alcohol consumption and broader sickness absence in fully adjusted models.

DISCUSSION AND CONCLUSIONS

Findings suggest that only HED is linked to broader sickness absence. However, there is a strong dose-response association between alcohol consumption and alcohol-related absences for both consumption measures amongst Australian workers. Population-level policies that reduce alcohol consumption to moderate level and less frequent HED might address workplace absenteeism.

Trends in Psychological Distress Among Adults in England, 2020-2022

Importance

In the last 3 years, people in England have lived through a pandemic and cost-of-living and health care crises, all of which may have contributed to worsening mental health in the population.

Objective

To estimate trends in psychological distress among adults over this period and to examine differences by key potential moderators.

Design, Setting, and Participants

A monthly cross-sectional, nationally representative household survey of adults aged 18 years or older was conducted in England between April 2020 and December 2022.

Main Outcomes and Measures

Past-month distress was assessed with the Kessler Psychological Distress Scale. Time trends in any distress (moderate to severe, scores ≥5) and severe distress (scores ≥13) were modeled, and interactions with age, gender, occupational social grade, children in the household, smoking status, and drinking risk status were tested.

Results

Data were collected from 51 861 adults (weighted mean [SD] age, 48.6 [18.5] years; 26 609 women [51.3%]). There was little overall change in the proportion of respondents reporting any distress (from 34.5% to 32.0%; prevalence ratio [PR], 0.93; 95% CI, 0.87-0.99), but the proportion reporting severe distress increased by 46%, from 5.7% to 8.3% (PR, 1.46; 95% CI, 1.21-1.76). Although trends differed by sociodemographic characteristics, smoking, and drinking, the increase in severe distress was observed across all subgroups (with PR estimates ranging from 1.17 to 2.16), with the exception of those aged 65 years and older (PR, 0.79; 95% CI, 0.43-1.38); the increase was particularly pronounced since late 2021 among those younger than 25 years (increasing from 13.6% in December 2021 to 20.2% in December 2022).

Conclusions and Relevance

In this survey study of adults in England, the proportion reporting any psychological distress was similar in December 2022 to that in April 2020 (an extremely difficult and uncertain moment of the COVID-19 pandemic), but the proportion reporting severe distress was 46% higher. These findings provide evidence of a growing mental health crisis in England and underscore an urgent need to address its cause and to adequately fund mental health services.

Prediction of bullying at work: A data-driven analysis of the Finnish public sector cohort study

AIM

To determine the extent to which change in (i.e., start and end of) workplace bullying can be predicted by employee responses to standard workplace surveys.

METHODS

Responses to an 87-item survey from 48,537 Finnish public sector employees at T1 (2017-2018) and T2 (2019-2020) were analyzed with least-absolute-shrinkage-and-selection-operator (LASSO) regression. The predictors were modelled both at the individual- and the work unit level. Outcomes included both the start and the end of bullying. Predictive performance was evaluated with C-indices and density plots.

RESULTS

The model with best predictive ability predicted the start of bullying with individual-level predictors, had a C-index of 0.68 and included 25 variables, of which 6 remained in a more parsimonious model: discrimination at work unit, unreasonably high workload, threat that some work tasks will be terminated, working in a work unit where everyone did not feel they are understood and accepted, having a supervisor who was not highly trusted, and a shorter time in current position. Other models performed even worse, either from the point of view of predictive performance, or practical useability.

DISCUSSION

While many bivariate associations between socioeconomic characteristics, work characteristics, leadership, team climate, and job satisfaction were observed, reliable individualized detection of individuals at risk of becoming bullied at workplace was not successful. The predictive performance of the developed risk scores was suboptimal, and we do not recommend their use as an individual-level risk prediction tool. However, they might be useful tool to inform decision-making when planning the contents of interventions to prevent bullying at an organizational level.

Concurrent trajectories of self-rated health and working hour patterns in health care shift workers: A longitudinal analysis with 8-year follow-up

Background

The association between health and working hours is hypothesized to be reciprocal, but few longitudinal studies have examined changes in both health and working hour patterns over time. We examined combined trajectories of self-related health and two working hour patterns (working <35 h/week and working night shifts) and the extent to which these trajectories were predicted by employees' lifestyle and mental health.

Methods

Participants of this cohort study with a 8-year follow-up were 5,947 health care shift workers. We linked self-reports of health from three repeated surveys with objective pay-roll based data on working hours. Using group-based multi-trajectory analysis we identified concurrent trajectories for self-rated health and working hour patterns. We examined their associations with baseline lifestyle-related factors (smoking, at-risk alcohol use, obesity, and physical inactivity) and mental health (sleep problems and psychological distress) using multinomial regression analysis.

Results

Three combined trajectories of self-rated health and working <35 h/week and four combined trajectories of self-rated health and night work were identified. Unhealthy lifestyle and poor mental health were associated with trajectories of moderate and declining health. Sleep problems were linked with working <35 h/week. Younger age and good mental health were associated with a combined trajectory of good health and continued night shift work.

Conclusion

Trajectories of suboptimal and declining health are associated with trajectories of reducing working hours and leaving night work, and are more common in employees with unhealthy lifestyle, sleep problems, and psychological distress.

Cannabis Use Increases the Risk of Sickness Absence: Longitudinal Analyses From the CONSTANCES Cohort

Aims

To examine the longitudinal associations between cannabis use and risks of short (<7 days), medium (7-28 days), and long (>28 days) sickness absences at one-year follow-up.

Methods

87,273 participants aged 18-65 years from the French CONSTANCES cohort reported their frequency of cannabis use at inclusion between 2012 and 2018. Sickness absences occurring during one year of follow-up were collected from national medico-administrative registries. Multivariable generalized linear regressions were used to compute the Odds Ratios (OR) with their 95% Confidence Intervals (CI) of having at least one sickness absence at follow-up compared to no sickness absence, while controlling for sociodemographic factors, chronic conditions and occupational factors.

Results

Cannabis use more than once a month was associated with an increased risk of short (OR, [95% CI]: 1.56 [1.32–1.83]) and medium (1.29 [1.07–1.54]) sickness absences at one-year follow-up, with dose-dependent relationships for short sickness absences (1.13 [1.08–1.18], p-for-trend <0.001). In stratified analyses, cannabis use was associated with an increased risk of sickness absences in older individuals, men, participants with good self-rated health, living or having lived as a couple, and having an open-ended contract.

Conclusions

Cannabis use prospectively increased the risk of short and medium sickness absences, even from once a month and with a dose-dependent relationship for short sickness absences. These findings should be considered in information and prevention public health campaigns to alert the general population and workers to this increased risk.

Association of changes in work due to COVID-19 pandemic with psychosocial work environment and employee health: a cohort study of 24 299 Finnish public sector employees

OBJECTIVES

To examine the associations of COVID-19-related changes in work with perceptions of psychosocial work environment and employee health.

METHODS

In a cohort of 24 299 Finnish public sector employees, psychosocial work environment and employee well-being were assessed twice before (2016 and 2018=reference period) and once during (2020) the COVID-19 pandemic. Those who reported a change (='Exposed') in work due to the pandemic (working from home, new tasks or team reorganisation) were compared with those who did not report such change (='Non-exposed').

RESULTS

After adjusting for sex, age, socioeconomic status and lifestyle risk score, working from home (44%) was associated with greater increase in worktime control (standardised mean difference (SMD)Exposed=0.078, 95% CI 0.066 to 0.090; SMDNon-exposed=0.025, 95% CI 0.014 to 0.036), procedural justice (SMDExposed=0.101, 95% CI 0.084 to 0.118; SMDNon-exposed=0.053, 95% CI 0.038 to 0.068), workplace social capital (SMDExposed=0.094, 95% CI 0.077 to 0.110; SMDNon-exposed=0.034, 95% CI 0.019 to 0.048), less decline in self-rated health (SMDExposed=-0.038, 95% CI -0.054 to -0.022; SMDNon-exposed=-0.081, 95% CI -0.095 to -0.067), perceived work ability (SMDExposed=-0.091, 95% CI -0.108 to -0.074; SMDNon-exposed=-0.151, 95% CI -0.167 to -0.136) and less increase in psychological distress (risk ratio (RR)Exposed=1.06, 95% CI 1.02 to 1.09; RRNon-exposed=1.16, 95% CI 1.13 to 1.20). New tasks (6%) were associated with greater increase in psychological distress (RRExposed=1.28, 95% CI 1.19 to 1.39; RRNon-exposed=1.10, 95% CI 1.07 to 1.12) and team reorganisation (5%) with slightly steeper decline in perceived work ability (SMDExposed=-0.151 95% CI -0.203 to -0.098; SMDNon-exposed=-0.124, 95% CI -0.136 to -0.112).

CONCLUSION

Employees who worked from home during the pandemic had more favourable psychosocial work environment and health, whereas those who were exposed to work task changes and team reorganisations experienced more adverse changes.

Distribution of sickness absence risk across different levels and patterns of drinking: findings from the Stockholm Public Health Cohort

Aim:

This study estimated (i) the risk function between different indicators of alcohol use and long-term sickness absence, adjusting for possible confounding factors, (ii) whether the risk function between average volume of consumption and sickness absence is modified by heavy episodic drinking (HED), and (iii) to what extent the risk for sickness absence among abstainers is due to health selection bias.

Data and methods:

The study was based on data from the Stockholm Public Health Cohort 2006, with an analytical sample of 16,477 respondents aged 18–64 years. The outcome included register-based long-term (> 14 days) sickness absence. Negative binominal regression was used to estimate the association between sickness absence and average weekly volume of consumption, frequency of HED, and both in interaction.

Results:

Abstainers, chronic heavy drinkers and respondents with the highest frequency of HED had approximately two-fold higher rates of sickness absence relative to the reference groups, i.e., moderate drinkers and those with HED one to 6 times per year. Adjustment for confounding factors did not materially affect the shape of the risk function. After exclusion of abstainers with alcohol-related problems, or poor health, the estimates for abstainers became non-significant. Moderate drinkers with HED did not have significantly higher rates of sickness absence than moderate drinkers without HED.

Conclusions:

Our results suggest a significant association between alcohol use and sickness absence. There were indications that the U-shaped risk function may largely be due to health selection bias among abstainers. We found no indication of effect modification of HED on moderate drinking.

A systematic review and meta-analysis uncovering the relationship between alcohol consumption and sickness absence. When type of design, data, and sickness absence make a difference

AIM

Earlier research has revealed a strong relationship between alcohol use and sickness absence. The aim of this review was to explore and uncover this relationship by looking at differences in type of design (cross-sectional vs. longitudinal), type of data (self-reported vs. registered data), and type of sickness absence (long-term vs. short term).

METHOD

Six databases were searched through June 2020. Observational and experimental studies from 1980 to 2020, in English or Scandinavian languages reporting the results of the association between alcohol consumption and sickness absence among working population were included. Quality assessment, and statistical analysis focusing on differences in the likelihood of sickness absence on subgroup levels were performed on each association, not on each study. Differences in the likelihood of sickness absence were analyzed by means of meta-analysis. PROSPERO registration number: CRD42018112078.

RESULTS

Fifty-nine studies (58% longitudinal) including 439,209 employees (min. 43, max. 77,746) from 15 countries were included. Most associations indicating positive and statistically significant results were based on longitudinal data (70%) and confirmed the strong/causal relationship between alcohol use and sickness absence. The meta-analysis included eight studies (ten samples). The increased risk for sickness absence was likely to be found in cross-sectional studies (OR: 8.28, 95% CI: 6.33-10.81), studies using self-reported absence data (OR: 5.16, 95% CI: 3.16-8.45), and those reporting short-term sickness absence (OR: 4.84, 95% CI: 2.73-8.60).

CONCLUSION

This review supports, but also challenges earlier evidence on the association between alcohol use and sickness absence. Certain types of design, data, and types of sickness absence may produce large effects. Hence, to investigate the actual association between alcohol and sickness absence, research should produce and review longitudinal designed studies using registry data and do subgroup analyses that cover and explain variability of this association.

Return to Work After a Disabling Mental Disorder: Predictors From a Brazilian 1-Year Longitudinal Study

OBJECTIVE

This study aims to analyze factors that influence return to work (RTW) among workers on sickness absence due to mental disorders.

METHODS

A longitudinal study conducted between 2014 and 2017 in São Paulo, Brazil. The 385 participants answered a questionnaire including sociodemographics, habits/lifestyle, job characteristics, and clinical information. Survival analysis was performed to identify factors influencing the RTW.

RESULTS

Most of participants were females (74.5%), worked in jobs dealing with public (44.2%) and were depressed (52.4%). RTW occurred for 68.3% participants over 1-year follow-up. Mean duration of absence was 163.83 days. The risk profile for remaining absent was heavy smokers, be abstainer, obese, deal with the public, perceived great effort at work, and low self-efficacy.

CONCLUSIONS

These findings can contribute in discussion about disability prevention and interventions to assure mental health care for workers.

Systematic Review and Dose-Response Meta-Analysis on the Relationship Between Alcohol Consumption and Sickness Absence

AIMS

Alcohol consumption (AC) may cause workplace absence, but the findings of individual studies vary markedly. To date, no dose-response meta-analysis (DRMA) of the relationship between AC and sickness absence (SA) has been completed. This paper aims to estimate the dose-response relationship between AC and the risk of SA based on published observational studies.

METHODS

We used DRMA and modelling to investigate the effects of varying doses of AC (including heavy episodic drinking (HED)) onSA.

RESULTS

The meta-analysis included 21 studies (12 cohort studies and 9 cross-sectional). It showed that HED, risky (20-40 g of alcohol/day) and high-risk (>40 g of alcohol/day) drinkers had an elevated risk of SA when compared with light-to-moderate drinkers for both sexes. Those who abstained from alcohol had a higher risk of SA than those who drink moderately.

CONCLUSIONS

Our results indicate that risky, high-risk drinking and HED may increase the risk of absenteeism. The implementation of population-based strategies may be appropriate to address the burdens of alcohol-related SA. Additionally, economic evaluations of alcohol policies should incorporate their impacts on SA. However, the current literature has substantial limitations, relying on modestly designed studies from just a few settings and more studies are needed-especially those that measure abstention in more nuancedways.

Health behaviours and psychosocial working conditions as predictors of disability pension due to different diagnoses: a population-based study

BACKGROUND

To investigate whether the clustering of different health behaviours (i.e. physical activity, tobacco use and alcohol consumption) influences the associations between psychosocial working conditions and disability pension due to different diagnoses.

METHODS

A population-based sample of 24,987 Swedish twins born before 1958 were followed from national registers for disability pension until 2013. Baseline survey data in 1998-2003 were used to assess health behaviours and psychosocial Job Exposure Matrix for job control, job demands and social support. Cox proportional hazards models were used to calculate hazard ratios (HR) with 95% confidence intervals (CI).

RESULTS

During follow-up, 1252 disability pensions due to musculoskeletal disorders (5%), 601 due to mental diagnoses (2%) and 1162 due to other diagnoses (5%) occurred. In the models controlling for covariates, each one-unit increase in job demands was associated with higher (HR 1.16, 95%CI 1.01-1.33) and in job control with lower (HR 0.87, 95%CI 0.80-0.94) risk of disability pension due to musculoskeletal disorders among those with unhealthy behaviours. Among those with healthy behaviours, one-unit increase of social support was associated with a higher risk of disability pension due to mental and due to other diagnoses (HRs 1.29-1.30, 95%CI 1.04-1.63).

CONCLUSIONS

Job control and job demands were associated with the risk of disability pension due to musculoskeletal disorders only among those with unhealthy behaviours. Social support was a risk factor for disability pension due to mental or other diagnoses among those with healthy behaviours. Workplaces and occupational health care should acknowledge these simultaneous circumstances in order to prevent disability pension.

Cost and consequence analysis of the Healthy Choices at Work programme to prevent non-communicable diseases in a commercial power plant, South Africa

Background

The workplace is an ideal setting for the implementation of a health promotion programmes to prevent non-communicable diseases (NCD). There are limited resources assigned to workplace health promotion programmes in low-and middle-income countries (LMIC).

Aim

This study aimed to conduct a cost and consequence analysis of the Healthy Choices at Work programme.

Setting

This study was conducted at a commercial power plant in South Africa.

Methods

Incremental costs were obtained for the activities of the Healthy Choices at Work programme over a two-year period. A total of 156 employees were evaluated in the intervention, although the effect was experienced by all employees. An annual health risk factor assessment at baseline and follow up evaluated the consequences of the programme.

Results

The total incremental costs over the two-year period accumulated to $4015 for 1743 employees. The cost per employee on an annual basis was $1.15 and was associated with a −10.2mmHg decrease in systolic blood pressure, −3.87mmHg in diastolic blood pressure, −0.45mmol/l in total cholesterol and significant improvement in harmful alcohol use, fruit and vegetable intake and physical inactivity (p < 0.001). There was no correlation between sickness absenteeism and risk factors for NCDs.

Conclusion

The cost to implement the multicomponent HCW programme was low with significant beneficial consequences in transforming the workplace environment and reducing risks factors for NCDs. Findings of this study will be useful for small, medium and large organisations, the national department of health, and similar settings in LMICs.

Lifestyle factors and risk of sickness absence from work: a multicohort study

Background

Lifestyle factors influence the risk of morbidity and mortality, but the extent to which they are associated with employees' absence from work due to illness is unclear. We examined the relative contributions of smoking, alcohol consumption, high body-mass index, and low physical activity to diagnosis-specific sickness absence.

Methods

We did a multicohort study with individual-level data of participants of four cohorts from the UK, France, and Finland. Participants' responses to a lifestyle survey were linked to records of sickness absence episodes, typically lasting longer than 9 days; for each diagnostic category, the outcome was the total number of sickness absence days per year. We estimated the associations between lifestyle factors and sickness absence by calculating rate ratios for the number of sickness absence days per year and combining cohort-specific estimates with meta-analysis. The criteria for assessing the evidence included the strength of association, consistency across cohorts, robustness to adjustments and multiple testing, and impact assessment by use of population attributable fractions (PAF), with both internal lifestyle factor prevalence estimates and those obtained from European populations (PAF_external).

Findings

For 74 296 participants, during 446 478 person-years at risk, the most common diagnoses for sickness absence were musculoskeletal diseases (70·9 days per 10 person-years), depressive disorders (26·5 days per 10 person-years), and external causes (such as injuries and poisonings; 12·8 days per 10 person-years). Being overweight (rate ratio [adjusted for age, sex, socioeconomic status, and chronic disease at baseline] 1·30, 95% CI 1·21–1·40; PAF_external 8·9%) and low physical activity (1·23, 1·14–1·34; 7·8%) were associated with absences due to musculoskeletal diseases; heavy episodic drinking (1·90, 1·41–2·56; 15·2%), smoking (1·70, 1·42–2·03; 11·8%), low physical activity (1·67, 1·42–1·96; 19·8%), and obesity (1·38, 1·11–1·71; 5·6%) were associated with absences due to depressive disorders; heavy episodic drinking (1·64, 1·33–2·03; 11·3%), obesity (1·48, 1·27–1·72; 6·6%), smoking (1·35, 1·20–1·53; 6·3%), and being overweight (1·20, 1·08–1·33; 6·2%) were associated with absences due to external causes; obesity (1·82, 1·40–2·36; 11·0%) and smoking (1·60, 1·30–1·98; 10·3%) were associated with absences due to circulatory diseases; low physical activity (1·37, 1·25–1·49; 12·0%) and smoking (1·27, 1·16–1·40; 4·9%) were associated with absences due to respiratory diseases; and obesity (1·67, 1·34–2·07; 9·7%) was associated with absences due to digestive diseases.

Interpretation

Lifestyle factors are associated with sickness absence due to several diseases, but observational data cannot determine the nature of these associations. Future studies should investigate the cost-effectiveness of lifestyle interventions aimed at reducing sickness absence and the use of information on lifestyle for identifying groups at risk.

Funding

NordForsk, British Medical Research Council, Academy of Finland, Helsinki Institute of Life Sciences, and Economic and Social Research Council.

A novel approach to determine two optimal cut-points of a continuous predictor with a U-shaped relationship to hazard ratio in survival data: simulation and application

BACKGROUND

In clinical and epidemiological researches, continuous predictors are often discretized into categorical variables for classification of patients. When the relationship between a continuous predictor and log relative hazards is U-shaped in survival data, there is a lack of a satisfying solution to find optimal cut-points to discretize the continuous predictor. In this study, we propose a novel approach named optimal equal-HR method to discretize a continuous variable that has a U-shaped relationship with log relative hazards in survival data.

METHODS

The main idea of the optimal equal-HR method is to find two optimal cut-points that have equal log relative hazard values and result in Cox models with minimum AIC value. An R package 'CutpointsOEHR' has been developed for easy implementation of the optimal equal-HR method. A Monte Carlo simulation study was carried out to investigate the performance of the optimal equal-HR method. In the simulation process, different censoring proportions, baseline hazard functions and asymmetry levels of U-shaped relationships were chosen. To compare the optimal equal-HR method with other common approaches, the predictive performance of Cox models with variables discretized by different cut-points was assessed.

RESULTS

Simulation results showed that in asymmetric U-shape scenarios the optimal equal-HR method had better performance than the median split method, the upper and lower quantiles method, and the minimum p-value method regarding discrimination ability and overall performance of Cox models. The optimal equal-HR method was applied to a real dataset of small cell lung cancer. The real data example demonstrated that the optimal equal-HR method could provide clinical meaningful cut-points and had good predictive performance in Cox models.

CONCLUSIONS

In general, the optimal equal-HR method is recommended to discretize a continuous predictor with right-censored outcomes if the predictor has an asymmetric U-shaped relationship with log relative hazards based on Cox regression models.

Sickness absence diagnoses among abstainers, low-risk drinkers and at-risk drinkers: consideration of the U-shaped association between alcohol use and sickness absence in four cohort studies

AIMS

To estimate differences in the strength and shape of associations between alcohol use and diagnosis-specific sickness absence.

DESIGN

A multi-cohort study. Participants (n = 47 520) responded to a survey on alcohol use at two time-points, and were linked to records of sickness absence. Diagnosis-specific sickness absence was followed for 4-7 years from the latter survey.

SETTING AND PARTICIPANTS

From Finland, we had population cohort survey data from 1998 and 2003 and employee cohort survey data from 2000-02 and 2004. From France and the United Kingdom, we had employee cohort survey data from 1993 and 1997, and 1985-88 and 1991-94, respectively.

MEASUREMENTS

We used standard questionnaires to assess alcohol intake categorized into 0, 1-11 and > 11 units per week in women and 0, 1-34 and > 34 units per week in men. We identified groups with stable and changing alcohol use over time. We linked participants to records from sickness absence registers. Diagnoses of sickness absence were coded according to the International Classification of Diseases. Estimates were adjusted for sex, age, socio-economic status, smoking and body mass index.

FINDINGS

Women who reported drinking 1-11 units and men who reported drinking 1-34 units of alcohol per week in both surveys were the reference group. Compared with them, women and men who reported no alcohol use in either survey had a higher risk of sickness absence due to mental disorders [rate ratio = 1.51, 95% confidence interval (CI) = 1.22-1.88], musculoskeletal disorders (1.22, 95% CI = 1.06-1.41), diseases of the digestive system (1.35, 95% CI = 1.02-1.77) and diseases of the respiratory system (1.49, 95% CI = 1.29-1.72). Women who reported alcohol consumption of > 11 weekly units and men who reported alcohol consumption of > 34 units per week in both surveys were at increased risk of absence due to injury or poisoning (1.44, 95% CI = 1.13-1.83).

CONCLUSIONS

In Finland, France and the United Kingdom, people who report not drinking any alcohol on two occasions several years apart appear to have a higher prevalence of sickness absence from work with chronic somatic and mental illness diagnoses than those drinking below a risk threshold of 11 units per week for women and 34 units per week for men. Persistent at-risk drinking in Finland, France and the United Kingdom appears to be related to increased absence due to injury or poisoning.