Discrete-time Proportional Hazards Models Research Articles

Application of High-Dimensional Propensity Score Methods to the National Health and Aging Trends Study.

High-dimensional propensity scoring (HDPS) is a method for empirically identifying potential confounders within large healthcare databases such as administrative claims data. However, this method has not yet been applied to large national health surveys such as the National Health and Aging Trends Study (NHATS), an ongoing nationally representative survey of older adults in the United States and important resource in gerontology research. In this Research Practice article, we present an overview of HDPS and describe the specific data transformation steps and analytic considerations needed to apply it to national health surveys. We applied HDPS within NHATS to investigate the association between self-reported visual difficulty and incident dementia, comparing HDPS to conventional confounder selection methods. Among 7 207 dementia-free NHATS Wave 1 respondents, 528 (7.3%) had self-reported visual difficulty. In an unadjusted discrete time proportional hazards model accounting for the complex survey design of NHATS, self-reported visual difficulty was strongly associated with incident dementia (odds ratio [OR] 2.34, 95% confidence interval [CI]: 1.95-2.81). After adjustment for standard investigator-selected covariates via inverse probability weighting, the magnitude of this association decreased, but evidence of an association remained (OR 1.44, 95% CI: 1.11-1.85). Adding 75 HDPS-prioritized variables to the investigator-selected propensity score model resulted in further attenuation of the association between visual impairment and dementia (OR 0.94, 95% CI: 0.70-1.23). HDPS can be successfully applied to national health surveys such as NHATS and may improve confounder adjustment. We hope developing this framework will encourage future consideration of HDPS in this setting.

Interaction between visual impairment and genetic risk of dementia and psychosis in older adults.

Visual impairment (VI) is associated with dementia and other neuropsychiatric outcomes, but previous studies have not considered genetic sources of confounding or effect modification. We analysed data from the Health and Retirement Study, an ongoing nationally representative survey of older US adults, a subset of whom underwent genetic testing from 2006 to 2012 (n = 13465). Using discrete time proportional hazards models and generalised estimating equations, we measured the association between VI and dementia, depression and hallucinations adjusting for demographics and comorbidities, ancestry-specific principal components and polygenic risk scores (PRS) for Alzheimer's disease, major depressive disorder or schizophrenia. Effect modification was assessed using VI-PRS interaction terms and stratified analyses. VI was associated with dementia, depression and hallucinations after adjusting polygenic risk and other confounders. There was no VI-PRS interaction for dementia or depression. However, the association between VI and hallucinations varied by genetic risk of schizophrenia. Within the bottom four quintiles of schizophrenia PRS, VI was not associated with hallucinations among White (OR 1.16, 95% CI: 0.87-1.55) or Black participants (OR 0.96, 95% CI: 0.49-1.89). In contrast, VI was strongly associated with hallucinations among White (OR 2.08, 95% CI: 1.17-3.71) and Black (OR 10.63, 95% CI: 1.74-65.03) participants in the top quintile of schizophrenia PRS. The association between VI and neuropsychiatric outcomes is not explained by shared genetic risk factors, and there is a significant interaction between VI and polygenic risk of hallucinations in older adults.

Self-reported Visual Difficulty, Age-related Eye Disease, and Neuropsychiatric Outcomes in Older Adults

ABSTRACT Purpose Self-reported visual difficulty is consistently associated with dementia and other neuropsychiatric outcomes, but studies of specific age-related eye diseases have yielded conflicting results. Methods We conducted a retrospective cohort study using data from The National Health and Aging Trends Study, an ongoing nationally representative survey of older U.S. adults (n = 10,089). All subjects are screened for self-reported visual difficulty annually. Using linked Medicare claims data, we identified subjects with age-related macular degeneration (AMD), primary open-angle glaucoma (POAG), diabetic retinopathy, and cataract. For each condition, controls with complete Medicare eligibility and at least one eye care encounter were selected. We used semiparametric discrete time proportional hazards models to measure associations with incident dementia, and generalized estimating equations to examine longitudinal associations with depression, anxiety, and hallucinations, adjusting for baseline demographics and time-varying comorbidities. Results Self-reported visual difficulty was associated with dementia (HR 1.16, 95% CI: 1.00–1.34), depression (OR 1.14, 95% CI: 1.04–1.26), anxiety (OR 1.17, 95% CI: 1.06–1.29), and hallucinations (OR 1.54, 95% CI: 1.29–1.84). Diabetic retinopathy was associated with depression (OR 1.31, 95% CI: 1.05–1.64), and cataracts were associated with a lower risk of depression (OR 0.84, 95% CI: 0.74–0.95) and anxiety (OR 0.86, 95% CI: 0.75–0.99). There were no other associations between age-related eye disease and neuropsychiatric outcomes. Conclusion Self-reported visual difficulty is associated with dementia and other neuropsychiatric outcomes to a greater degree than age-related eye disease. These findings highlight the distinction between self-reported vision and clinically diagnosed eye disease with regard to health outcomes in older adults.

287 What is the Velocity of Return-to-work After Circumferential Minimally-Invasive Surgery Versus Open Correction for Adult Spinal Deformity?

INTRODUCTION: While faster recovery is a touted benefit of minimally-invasive spinal operations, these differences have been poorly described for return-to-work after larger fusion constructs for adult spinal deformity. METHODS: Adult spinal deformity (ASD) patients =18 yr were queried from a multicenter, prospective database across the USA. To ensure equal cohorts for comparing the impact of MIS vs. open correction technique, propensity score matched gender, age, BMI, pelvic incidence-lumbar lordosis mismatch, and sagittal vertical axis. Of 190 patients with baseline surveys, 173 respondents (91.0%) had both baseline employment and 2-year follow-up. The open and cMIS cohorts were compared with a chi-squared test. The rate of return-to-work was calculated via discrete-time proportional hazards model—reporting an adjusted odds ratio (ORadj)—at three follow-up intervals: 6 weeks, 1-year, and 2-years. RESULTS: Of 173 patients, baseline employment status was nearly equal between open (n = 85) vs cMIS (n = 88) cohorts: employed (20.0% vs 21.5%), unemployed (3.5% vs 4.5%), disabled (4.7% vs 6.8%), and retired (71.7% vs 67.0%) (p = 0.888). The number of employed patients between the open vs cMIS cohort statistically significantly differed at 1-year (10.5% vs 21.5%, p = 0.049), but not at 6 weeks (14.1% vs 15.9%, p = 0.741) or 2-years (14.1% vs 19.3%, p = 0.277). The velocity of return-to-work did not differ (OR = 1.2, p = 0.549). In a sub-analysis of 36 patients who were employed at baseline, the number of employed patients between the open vs cMIS cohort statistically significantly differed at 1-year (47.0% vs 89.4%, p = 0.006), but not at 6 weeks (70.5% vs 73.6%, p = 0.636) or 2-years (58.8% vs 78.9%, p = 0.191). The velocity of return-to-work was also similar (ORadj = 1.57, p = 0.481). CONCLUSIONS: Despite an equivalent baseline employment status, more patients in the cMIS cohort were employed at 1-year; however, there was no difference at 2-years. Similar trends were appreciated in a sub-analysis of patients who were employed at baseline.

Cataract surgery utilization in Parkinson's disease: the english longitudinal survey on ageing.

Older adults with Alzheimer's disease are less likely to be offered cataract surgery than the general population, but these disparities have not been examined in the context of other neurodegenerative disorders such as Parkinson's disease (PD). Using data from the English Longitudinal Study on Ageing (ELSA), an ongoing, longitudinal population-based survey of older adults in England, we examined the relationship between PD and cataract surgery among adults 50 and older. We used logistic regression to measure the association between PD and history of cataract surgery at baseline. In longitudinal analyses of subjects with no history of cataract surgery at time of enrollment, we used semiparametric, discrete-time proportional hazards models to model the incidence of cataract surgery as a function of PD and other time-dependent covariates. Models were adjusted for demographic variables, self-reported comorbidities, and measures of daily activity limitation. We included data from 19,241 eligible ELSA respondents, of whom 231 (1.2%) reported a diagnosis of PD. PD was positively associated with a history of self-reported cataract surgery at baseline (OR 3.66, 95% CI: 2.55-5.26), but this did not remain significant after adjusting for confounders (OR 1.22, 95% CI: 0.75-1.98). Among subjects with no history of cataract surgery at baseline, PD was also not associated with incident cataract surgery (adjusted HR 1.32, 95% CI: 0.86-2.02). Unlike Alzheimer's disease, people with PD were no less likely to receive cataract surgery compared to those without PD.

Ability of seedlings to survive heat and drought portends future demographic challenges for five southwestern US conifers.

Climate change and disturbance are altering forests and the rates and locations of tree regeneration. In semi-arid forests of the southwestern USA, limitations imposed by hot and dry conditions are likely to influence seedling survival. We examined how the survival of 1-year seedlings of five southwestern US conifer species whose southwestern distributions range from warmer and drier woodlands and forests (Pinus edulis Engelm., Pinus ponderosa Douglas ex C. Lawson) to cooler and wetter subalpine forests (Pseudotsuga menziesii (Mirb.) Franco, Abies concolor (Gord. & Glend.) Lindl. Ex Hildebr. and Picea engelmannii Parry ex Engelm.) changed in response to low moisture availability, high temperatures and high vapor pressure deficit in incubators. We used a Bayesian framework to construct discrete-time proportional hazard models that explained 55-75% of the species-specific survival variability. We applied these to the recent climate (1980-2019) of the southwestern USA as well as 1980-2099 CMIP5 climate projections with the RCP8.5 emissions pathway. We found that the more mesic species (i.e., P. menziesii, A. concolor and P. engelmannii) were more susceptible to the effects of hot and dry periods. However, their existing ranges are not projected to experience the conditions we tested as early in the 21st century as the more xeric P. edulis and P. ponderosa, leading to lower percentages of their existing ranges predicted to experience seedling-killing conditions. By late-century, extensive areas of each species southwestern range could experience climate conditions that increase the likelihood of seedling mortality. These results demonstrate that empirically derived physiological limitations can be used to inform where species composition or vegetation type change are likely to occur in the southwestern USA.

Night and shift work patterns and incidence of type 2 diabetes and hypertension in a prospective cohort study of healthcare employees.

This study aimed to evaluate effects of night and shift work patterns on type 2 diabetes (T2D) and hypertension in a longitudinal study, with detailed information on working hours. The cohort comprised about 28 000 nurses and nursing assistants employed for more than one year 2008-2016 in Stockholm, Sweden. The employee register held detailed individual information on daily working hours. Information on diagnoses came from national and regional registers. Hazard ratios (HR) and confidence intervals (CI) were estimated by discrete-time proportional hazard models, adjusting for sex, age, country of birth, and profession. During follow-up in 2013-2017, we identified 232 cases of T2D and 875 of hypertension. We observed an increased risk of T2D, but not hypertension, among employees who worked only night shifts the previous year (HR 1.59, 95% CI 1.02-2.43) and those with intensive shift work (>120 afternoon and/or night shifts the previous year: HR 1.67, 95% CI 1.11-2.48) compared to only day work. There was a non-significantly increased risk of T2D related to mixed day and afternoon shifts (HR 1.34, 95% CI 0.97-1.88). We observed tendencies in increased risk of T2D related to frequent spells of ≥3 consecutive night shifts and with number of years with exclusive (but not mixed) night work. Permanent night work and frequent afternoon and/or night shifts were associated with an increased risk of T2D the following year, but not hypertension. The T2D risk was, to some extent, affected by frequent spells of several night shifts in a row and by cumulative years with permanent night work.

Short report on navigating access to care for Medicaid-enrolled autistic youth and young adults: Examining accrual of intellectual disability diagnoses in adolescence.

What is known? In most states, Medicaid waivers provide individuals with an intellectual disability diagnosis generous healthcare coverage throughout adulthood. By comparison, fewer Medicaid programs are available for autistic individuals, and they are more likely to experience disruptions, or gaps, in Medicaid coverage and subsequently not re-enroll.What this paper adds? One in five autistic individuals with Medicaid coverage between ages 8 and 25 accrued a new intellectual disability diagnosis. The probability of a new intellectual disability diagnosis was higher among those who had previous disruptions in Medicaid coverage.Implications for research and policy. Expanding Medicaid to cover autistic people of all ages could decrease the need for intellectual disability diagnosis accrual. Input from autistic individuals and their families regarding their health insurance access and healthcare experiences is critically important to understanding next steps for research.

Night work and breast cancer risk in a cohort of female healthcare employees in Stockholm, Sweden

ObjectivesNight work has been classified as probably carcinogenic to humans by the International Agency for Research on Cancer, but epidemiological evidence was considered limited due to variability in findings and...

Dynamics of multidimensional poverty and its determinants among the middle-aged and older adults in China

The aim of this study was to understand the dynamics of multidimensional poverty and its determinants among mid-aged and older adults. We used 2011, 2013, 2015, and 2018 data from the China Health and Retirement Longitudinal Study. We utilised basic descriptive statistics, the poverty transition matrix, Kaplan–Meier estimates, and the discrete-time proportional hazards model for data analysis. From 2011 to 2018, the incidence of multidimensional poverty among mid-aged and older people basically decreased, but the average poverty intensity remained stable. Most mid-aged and older adults had transient multidimensional poverty. The longer an individual remained multidimensionally poor, the smaller the probability of exit from poverty; The longer an individual remained nonpoor after escaping from poverty, the smaller the probability of returning to poverty. As to other factors, individual characteristics, family structure, living arrangements, social capital, and living areas significantly affected the risks of multidimensional poverty exit and reentry. Based on these results, the government should implement targeted interventions for frail older adults with the identified characteristics to prevent them from persistent multidimensional poverty or return to poverty.

Modifiable Risk Factor Score and Fecundability in a Preconception Cohort in Singapore

Although multiple modifiable risk factors have been identified for reduced fecundability (defined as lower probability of conception within a menstrual cycle), no scoring system has been established to systematically evaluate fecundability among females who are attempting to conceive. To examine the association of a risk score based on 6 modifiable factors with fecundability, and to estimate the percentage reduction in incidence of nonconception if all study participants achieved a minimal risk score level. This population-based cohort study obtained data from the S-PRESTO (Singapore Preconception Study of Long-Term Maternal and Child Outcomes) prospective cohort study. Females of reproductive age who were trying to conceive were enrolled from February 2015 to October 2017 and followed for 1 year, ending in November 2018. Data were analyzed from March to May 2022. A reduced fecundability risk score was derived by giving participants 1 point for each of the following factors: unhealthy body mass index, unhealthy diet, smoking, alcohol intake, folic acid supplement nonuser, and older maternal age. Total scores ranged from 0 to 6 and were classified into 5 levels: level 1 (score of 0 or 1), level 2 (score of 2), level 3 (score of 3), level 4 (score of 4), and level 5 (score of 5 or 6). Fecundability, measured by time to conception in cycles, was analyzed using discrete-time proportional hazards models with confounder adjustment. A total of 937 females (mean [SD] age, 30.8 [3.8] years) were included, among whom 401 (42.8%) spontaneously conceived within 1 year of attempting conception; the median (IQR) number of cycles before conception was 4 (2-7). Compared with participants with a level 1 risk score, those with level 2, 3, 4, and 5 risk scores had reductions in fecundability of 31% (adjusted fecundability ratio [FR], 0.69; 95% CI, 0.54-0.88), 41% (FR, 0.59; 95% CI, 0.45-0.78), 54% (FR, 0.46; 95% CI, 0.31-0.69) and 77% (FR, 0.23; 95% CI, 0.07-0.73), respectively. Assessment of the population attributable fraction showed that all participants achieving a minimal (level 1) risk level would be associated with a reduction of 34% (95% CI, 30%-39%) in nonconception within a year. Results of this study revealed the co-occurrence of multiple modifiable risk factors for lowered fecundability and a substantially higher conception rate among participants with no or minimal risk factors. The risk assessment scoring system proposed is a simple and potentially useful public health tool for mitigating risks and guiding those who are trying to conceive.

Occupational exposure to noise and myocardial infarction risk one year later in Sweden

Objective: To explore the association between occupational noise exposure and myocardial infarction (MI) one year later. Methods: Data came from the SNOW cohort, comprised of all individuals born between 1930 and 1990 in Sweden, with demographic, occupational, and outcome data available from 1960 until 2017. In this study, we included working individuals with at least one occupational code between 1985 and 2013. These were matched to a job exposure matrix (JEM) in five categories (LAeq8h): <70, 70-74, 75-80, 80-85, >85 dB(A). MI status in the year following exposure was ascertained using the patient register. To account for time-varying occupational data, we utilized a discrete-time proportional hazards model adjusted for individual confounders and other occupational exposures. Results: Preliminary results show that exposure to over 75 dB(A) of occupational is associated with a 13-21% increased risk for MI one year later. After adjusting for psychosocial work exposures, physical workload, exposure to vibrations, and chemical and particle exposure, noise exposure was no longer associated with an increased risk for MI. Conclusion: Exposure to noise was not associated with an increased risk for MI one year later after adjusting for other work exposures among this younger, working population. Additional in-depth analyses are ongoing.

Determinants of Adoption and Dis-Adoption of Integrated Pest Management Practices in the Suppression of Mango Fruit Fly Infestation: Evidence from Embu County, Kenya

This study evaluates the drivers of the adoption and dis-adoption of Integrated Pest Management (IPM) practices in the suppression of mango fruit-fly infestation in Embu County, Kenya. It employs a Correlated Random Effects Probit Model and a Discrete-time Proportional Hazard Model on two-wave panel data of 149 mango farmers selected using a cluster sampling technique. The descriptive results show that 59% and 17% of the respondents were adopters and dis-adopters of mango fruit fly IPM practices, respectively. Empirical findings reveal that the cost of IPM and training on IPM positively and significantly influenced adoption, while the unavailability of the technology had a negative and significant effect on adoption. For dis-adoption, the results indicate that farm size and the quality of IPM positively influenced the hazard of exit from IPM use, and hence, enhanced the sustained adoption of IPM. The study recommends capacity building for mango farmers through training and increased access to extension services to enhance the adoption of this technology and prevent dis-adoption.

The Association Between Self-Reported Hearing Loss and Loss of Usual Source of Health Care Among Older Medicare Beneficiaries: Evidence From the National Health and Aging Trends Study.

The purpose of the study is to investigate the association of hearing loss (HL) with maintaining a usual source of care (USOC). In this study we implemented a time-to-event analysis using data from the National Health and Aging Trends Study (NHATS), a nationally representative study of older Medicare beneficiaries in the United States. The study sample included 2 114 older adults, aged 65+ years, 58.9% female, 20.4% Black, who reported having a USOC during the baseline round of NHATS and who remained community-dwelling during the 2011-2018 study period. Based on self-report measures at baseline, individuals' hearing status was classified into 3 categories: no HL, treated HL (hearing aids users), and untreated HL (nonhearing aid users who reported having hearing difficulties). Time-to-event was computed as the time elapsed between baseline and the study round in which the respondent first reported no longer having a USOC. Discrete-time proportional hazard models were estimated. In fully adjusted models, untreated HL at baseline was associated with a hazard ratio (HR) for losing one's USOC 1.60 (95% confidence interval: 1.01, 2.56) times higher than that of participants with no HL. We found no HR differences between the treated- and no-HL group. Untreated HL at baseline was associated with a higher probability of losing one's USOC over time. Noninvasive interventions such as hearing aids may be beneficial for maintaining a USOC.

International migration and its short‐term effect on partnership stability

International migration and its short‐term effect on partnership stability

Economic integration and survival of Ghana’s manufactured and agriculture raw material exports

The paper investigates the effects of economic integration agreements on the survival of Ghana's manufactured and agriculture raw material exports. Using annual trade data for Ghanaian manufactured and agriculture raw material product exports from 1988 to 2018 and a discrete-time proportional hazard model, we found that economic integration agreements enhance the survival of manufactured products and agriculture raw material exports. The Non-Reciprocal Preferential Trade Agreement is the most survival-enhancing. In addition, economic integration agreements strengthen the survival of existing active export relations but decrease the survival of export relations initiated after the agreement was signed. We also found that the average duration of manufactured products and agriculture raw material exports is 9.041 and 8.817 years, respectively. The study recommends that the Ministry of Trade and Industry should continue negotiating for Non-Reciprocal Trade Agreements.

Peak Expiratory Flow Predicts Incident Dementia in a Representative Sample of U.S. Older Adults: The National Health and Aging Trends Study (NHATS).

Dementia is an increasingly important public health problem with various risk factors. Respiratory function, measured via peak expiratory flow (PEF), may be a modifiable dementia risk factor. We investigated the association between PEF and incident dementia in 5 935 older adults from the National Health and Aging Trends Study (NHATS) from 2011 to 2014. Baseline PEF, expressed as a standardized residual (SR) percentile, was analyzed as a predictor of incident dementia using discrete-time proportional hazards models, while controlling for several health and sociodemographic covariates. After 14 332 person-years of follow-up, 9.0% (N = 536) had incident cases of dementia. Compared to the lowest PEF category (SR-percentile < 10%), the highest PEF category (SR-percentile ≥ 80%) had 49% lower risk of incident dementia (hazard ratio [HR] = 0.51; 95% confidence interval [CI; 0.37, 0.71]), and the second highest PEF category (SR-percentile 50%-80%) had 25% lower risk of incident dementia (HR = 0.75; 95% CI [0.56, 1.00]). A sensitivity analysis using multiple imputation to account for missing PEF measurements yielded similar associations with incident dementia. These associations suggest a dose-dependent relationship such that higher PEF categories were more protective against incident dementia. PEF may be considered as an easily administered, low-cost measure of respiratory function and a potentially modifiable dementia risk factor. Improving PEF may reduce dementia risk through vascular mechanisms, such as increased brain oxygenation. Future research should explore potential causal pathways between PEF and dementia.

Analysis of Severe Illness After Postvaccination COVID-19 Breakthrough Among Adults With and Without HIV in the US

Understanding the severity of postvaccination SARS-CoV-2 (ie, COVID-19) breakthrough illness among people with HIV (PWH) can inform vaccine guidelines and risk-reduction recommendations. To estimate the rate and risk of severe breakthrough illness among vaccinated PWH and people without HIV (PWoH) who experience a breakthrough infection. In this cohort study, the Corona-Infectious-Virus Epidemiology Team (CIVET-II) collaboration included adults (aged ≥18 years) with HIV who were receiving care and were fully vaccinated by June 30, 2021, along with PWoH matched according to date fully vaccinated, age group, race, ethnicity, and sex from 4 US integrated health systems and academic centers. Those with postvaccination COVID-19 breakthrough before December 31, 2021, were eligible. HIV infection. The main outcome was severe COVID-19 breakthrough illness, defined as hospitalization within 28 days after a breakthrough SARS-CoV-2 infection with a primary or secondary COVID-19 discharge diagnosis. Discrete time proportional hazards models estimated adjusted hazard ratios (aHRs) and 95% CIs of severe breakthrough illness within 28 days of breakthrough COVID-19 by HIV status adjusting for demographic variables, COVID-19 vaccine type, and clinical factors. The proportion of patients who received mechanical ventilation or died was compared by HIV status. Among 3649 patients with breakthrough COVID-19 (1241 PWH and 2408 PWoH), most were aged 55 years or older (2182 patients [59.8%]) and male (3244 patients [88.9%]). The cumulative incidence of severe illness in the first 28 days was low and comparable between PWoH and PWH (7.3% vs 6.7%; risk difference, -0.67%; 95% CI, -2.58% to 1.23%). The risk of severe breakthrough illness was 59% higher in PWH with CD4 cell counts less than 350 cells/μL compared with PWoH (aHR, 1.59; 95% CI, 0.99 to 2.46; P = .049). In multivariable analyses among PWH, being female, older, having a cancer diagnosis, and lower CD4 cell count were associated with increased risk of severe breakthrough illness, whereas previous COVID-19 was associated with reduced risk. Among 249 hospitalized patients, 24 (9.6%) were mechanically ventilated and 20 (8.0%) died, with no difference by HIV status. In this cohort study, the risk of severe COVID-19 breakthrough illness within 28 days of a breakthrough infection was low among vaccinated PWH and PWoH. PWH with moderate or severe immune suppression had a higher risk of severe breakthrough infection and should be included in groups prioritized for additional vaccine doses and risk-reduction strategies.

Night and shift work characteristics and incident ischemic heart disease and atrial fibrillation among healthcare employees - a prospective cohort study.

This study aimed to examine the effects of various aspects of night and shift work on the risk of incident ischemic heart disease (IHD) and atrial fibrillation (AF) using detailed and registry-based exposure data. This prospective cohort study included >30 300 healthcare employees (eg, nurses, nursing assistants) employed for at least one year in Region Stockholm 2008-2016. Information on daily working hours was obtained from a computerized administrative employee register and outcomes from national and regional registers. Using discrete-time proportional hazard models, we analyzed the outcomes as functions of working hour characteristics the preceding year, adjusted for sex, age, country of birth, education, and profession. We observed 223 cases of IHD and 281 cases of AF during follow-up 2009-2016. The risk of IHD was increased among employees who the preceding year had permanent night shifts compared to those with permanent day work [hazard ratio (HR) 1.61, 95% confidence interval (CI) 1.06-2.43] and among employees working night shifts >120 times per year compared to those who never worked night (HR 1.53, 95% CI 1.05-2.21). When restricted to non-night workers, the risk of IHD was increased for employees having frequent quick returns from afternoon shifts. No increased risks were observed for AF. Night work, especially working permanent night shifts and frequent night shifts, is associated with an increased risk of incident IHD but not AF. Moreover, frequent quick returns from afternoon shifts (among non-night workers) increased IHD risk. Organizing work schedules to minimize these exposures may reduce IHD risk.

Analysis of odds, probability, and hazard ratios: From 2 by 2 tables to two-sample survival data

Analysis of 2 by 2 tables and two-sample survival data has been widely used. Exact calculation is computational intractable for conditional likelihood inference in odds ratio models with large marginals in 2 by 2 tables, or partial likelihood inference in Cox’s proportional hazards models with considerable tied event times. Approximate methods are often employed, but their statistical properties have not been formally studied while taking into account the approximation involved. We develop new methods and theory by constructing suitable estimating functions while leveraging knowledge from conditional or partial likelihood inference. We propose a weighted Mantel–Haenszel estimator in an odds ratio model such as Cox’s discrete-time proportional hazards model. Moreover, we consider a probability ratio model, and derive as a consistent estimator the Breslow–Peto estimator, which has been regarded as an approximation to partial likelihood estimation in the odds ratio model. We study both model-based and model-robust variance estimation. For the Breslow–Peto estimator, our new model-based variance estimator is no greater than the commonly reported variance estimator. We present numerical studies which support the theoretical findings.