On hazard ratio estimators by proportional hazards models in matched-pair cohort studies

Consideration of stratification in confirmatory trials with time-to-event endpoint

Stratification in randomization and analysis are widely employed to balance treatment groups in clinical trials. However, the potential power loss due to under-stratification or over-stratification has not been thoroughly evaluated in the typical setting of confirmatory clinical trials. In cases where there are too many strata and some have small sample sizes or a small number of events, it is common practice to combine these small strata during analysis. However, there is a lack of guidance on how those small strata should be combined. This paper presents extensive simulation studies to evaluate the impact of under-stratification or over-stratification on the power of survival analysis and the estimate of hazard ratio using stratified log-rank test and Cox PH model, respectively. The difference in power between stratified and unstratified log-rank tests is also investigated under different scenarios. Our results suggest that failing to consider prognostic stratification factors with strong effects, and/or accounting for non-prognostic factors such as noise and predictive factors, may reduce the power of the stratified log-rank test. Additionally, methods of combining small strata are explored and compared.

Survival of Patients With Cancer With DPYD Variant Alleles and Dose-Individualized Fluoropyrimidine Therapy-A Matched-Pair Analysis

PURPOSE

DPYD-guided fluoropyrimidine dosing improves patient safety in carriers of DPYD variant alleles. However, the impact on treatment outcome in these patients is largely unknown. Therefore, progression-free survival (PFS) and overall survival (OS) were compared between DPYD variant carriers treated with a reduced dose and DPYD wild-type controls receiving a full fluoropyrimidine dose in a retrospective matched-pair survival analysis.

METHODS

Data from a prospective multicenter study (ClinicalTrials.gov identifier: NCT02324452) in which DPYD variant carriers received a 25% (c.1236G>A and c.2846A>T) or 50% (DPYD*2A and c.1679T>G) reduced dose and data from DPYD variant carriers treated with a similarly reduced dose of fluoropyrimidines identified during routine clinical care were obtained. Each DPYD variant carrier was matched to three DPYD wild-type controls treated with a standard dose. Survival analyses were performed using Kaplan-Meier estimates and Cox regression.

RESULTS

In total, 156 DPYD variant carriers and 775 DPYD wild-type controls were available for analysis. Sixty-one c.1236G>A, 25 DPYD*2A, 13 c.2846A>T, and-when pooled-93 DPYD variant carriers could each be matched to three unique DPYD wild-type controls. For pooled DPYD variant carriers, PFS (hazard ratio [HR], 1.23; 95% CI, 1.00 to 1.51; P = .053) and OS (HR, 0.95; 95% CI, 0.75 to 1.51; P = .698) were not negatively affected by DPYD-guided dose individualization. In the subgroup analyses, a shorter PFS (HR, 1.43; 95% CI, 1.10 to 1.86; P = .007) was found in c.1236G>A variant carriers, whereas no differences were found for DPYD*2A and c.2846A>T carriers.

CONCLUSION

In this exploratory analysis, DPYD-guided fluoropyrimidine dosing does not negatively affect PFS and OS in pooled DPYD variant carriers. Close monitoring with early dose modifications based on toxicity is recommended, especially for c.1236G>A carriers receiving a reduced starting dose.

Age at diagnosis modifies associations of type 2 diabetes with cancer incidence and mortality: a retrospective matched-cohort study

AIMS/HYPOTHESIS

The age-specific associations between type 2 diabetes and cancer risk are not fully understood. The aim of this study was to assess how age at diagnosis modifies the associations between type 2 diabetes and cancer risk.

METHODS

We used data from the Yinzhou Health Information System, and included 42,279 individuals who were newly diagnosed with type 2 diabetes between 2010 and 2014, as well as 166,010 age- and sex-matched control individuals without diabetes who were selected randomly from the electronic health records of the entire population. Patients were divided into four age groups according to age at diagnosis: <50, 50-59, 60-69 and ≥70 years. Stratified Cox proportional hazards regression models, with age as the time scale, were used to estimate the HRs and 95% CIs for the associations of type 2 diabetes with the risks of overall and site-specific cancers. Population-attributable fractions were also calculated for outcomes associated with type 2 diabetes.

RESULTS

During median follow-up periods of 9.20 and 9.32 years, we identified 15,729 incident cancer cases and 5383 cancer deaths, respectively. Patients diagnosed with type 2 diabetes before 50 years of age had the highest relative risks of cancer incidence and mortality, with HRs (95% CI) of 1.35 (1.20, 1.52) for overall cancer incidence, 1.39 (1.11, 1.73) for gastrointestinal cancer incidence, 2.02 (1.50, 2.71) for overall cancer mortality, and 2.82 (1.91, 4.18) for gastrointestinal cancer mortality. Risk estimates decreased gradually with each decade increase in diagnostic age. The population-attributable fractions for overall cancer and gastrointestinal cancer mortality also decreased with increasing age.

CONCLUSIONS/INTERPRETATION

The associations of type 2 diabetes with cancer incidence and mortality varied by age at diagnosis, with a higher relative risk among patients who were diagnosed at a younger age.

Cardiovascular outcomes and mortality after incident ischaemic stroke in patients with a recent cancer history

BACKGROUND

Up to 10% of patients with ischaemic stroke have comorbid cancer and stroke in these patients is thought to have a poor short-term prognosis. There is little known about the long-term cardiovascular morbidity and mortality outcomes after incident ischaemic stroke in patients with recent cancer history.

OBJECTIVE

To assess the risk of subsequent cardiovascular morbidity and mortality outcomes in patients with an incident ischaemic stroke and recent cancer history.

METHODS

Patients aged ≥18 years with an incident ischaemic stroke between 1998 and 2017, with any diagnosis of cancer within 12 months before the stroke event, and no prior history of serious vascular event were identified from UK Clinical Practice Research Datalink (CPRD GOLD) linked to Hospital Episode Statistics (HES) data. To minimize selection bias, these patients were propensity-score matched with patients with incident ischaemic stroke and no history of cancer. Propensity-score matching was done using covariates such as demographic data, vascular risk factors, comorbid conditions, and prescribed medication. Multivariable models (Competing risks and Cox regression) were used to determine the risk of subsequent major adverse cardiovascular event (MACE) outcomes and all-cause mortality.

RESULTS

Our cohort included 22,460 patients with a median age of 75 (IQR 64-83) years and a follow-up of 12.3 (IQR 7.2-16.7) years. Recent cancer was identified in 1,149 patients (5.1%) at the time of incident ischaemic stroke. The patients with recent cancer history had a lower risk of composite MACE (sub-distribution hazard ratio (SHR) 0.83 [95% CI: 0.75-0.92]) and recurrent stroke (SHR 0.85 95% CI:0.75-0.96]) and a higher risk of all-cause mortality (hazard ratio 1.67 [95% CI:1.47-1.91]). The risk of coronary heart disease, peripheral vascular disease, heart failure, and CVD-related death outcomes did not differ significantly between the groups.

CONCLUSIONS

After incident ischaemic stroke, patients with recent cancer history have a lower risk of composite MACE and recurrent stroke outcomes but a higher risk of all-cause mortality when compared with patients without a prior history of cancer.

The Risk of Endometrial Cancer and Uterine Sarcoma Following Endometriosis or Pelvic Inflammatory Disease

The relationship between uterine corpus cancer and endometriosis was conflicting. We aimed to determine the risk of uterine cancer in patients with endometriosis or pelvic inflammatory disease (PID). In this population-based cohort study, a total of 135,236 females with endometriosis (n = 20,510) or PID (n = 114,726), as well as 135,236 age-matched controls, were included. Cox regression models estimated the risk of uterine cancer in each group. Sub-outcomes of risk for uterine corpus cancer included endometrial cancer and uterine sarcoma were analyzed. An age subgroup analysis was performed to determine the moderator effect of age. A landmark analysis depicted the time varying effect of endometriosis and PID. A propensity score matching analysis was conducted to validate the findings. Patients with endometriosis had significantly higher risk of endometrial cancer (adjusted hazard ratio, aHR = 2.92; 95% CI = 2.12-4.03) and uterine sarcoma (aHR = 5.83; 95% CI = 2.02-16.89), while PID was not associated with the risk of uterine cancer. The increased risk of uterine cancer in patients with endometriosis persisted after propensity score matching (aHR = 2.83, 95%CI = 1.70-4.71). The greatest risk of endometrial cancer occurred in patients who had endometriosis for 37 to 60 months (adjusted relative risk, aRR = 9.15, 95% CI = 4.40-19.02). Females aged 12 to 35 years were at the greatest risk of endometriosis-associated uterine cancer (RR = 6.97, 95% CI = 3.41-14.26). In conclusion, patients with endometriosis were at great risk of uterine cancer, including endometrial cancer and uterine sarcoma, compared with propensity score-matched populations and compared with patients of PID. Younger females with endometriosis and patients who had endometriosis for three to five years were at the greatest risk of endometriosis-associated uterine cancer.

Cardiovascular disease and mortality sequelae of COVID-19 in the UK Biobank

OBJECTIVE

To examine association of COVID-19 with incident cardiovascular events in 17 871 UK Biobank cases between March 2020 and 2021.

METHODS

COVID-19 cases were defined using health record linkage. Each case was propensity score-matched to two uninfected controls on age, sex, deprivation, body mass index, ethnicity, diabetes, prevalent ischaemic heart disease (IHD), smoking, hypertension and high cholesterol. We included the following incident outcomes: myocardial infarction, stroke, heart failure, atrial fibrillation, venous thromboembolism (VTE), pericarditis, all-cause death, cardiovascular death, IHD death. Cox proportional hazards regression was used to estimate associations of COVID-19 with each outcome over an average of 141 days (range 32-395) of prospective follow-up.

RESULTS

Non-hospitalised cases (n=14 304) had increased risk of incident VTE (HR 2.74 (95% CI 1.38 to 5.45), p=0.004) and death (HR 10.23 (95% CI 7.63 to 13.70), p<0.0001). Individuals with primary COVID-19 hospitalisation (n=2701) had increased risk of all outcomes considered. The largest effect sizes were with VTE (HR 27.6 (95% CI 14.5 to 52.3); p<0.0001), heart failure (HR 21.6 (95% CI 10.9 to 42.9); p<0.0001) and stroke (HR 17.5 (95% CI 5.26 to 57.9); p<0.0001). Those hospitalised with COVID-19 as a secondary diagnosis (n=866) had similarly increased cardiovascular risk. The associated risks were greatest in the first 30 days after infection but remained higher than controls even after this period.

CONCLUSIONS

Individuals hospitalised with COVID-19 have increased risk of incident cardiovascular events across a range of disease and mortality outcomes. The risk of most events is highest in the early postinfection period. Individuals not requiring hospitalisation have increased risk of VTE, but not of other cardiovascular-specific outcomes.

Gout as a Risk Factor for Age-Related Macular Degeneration in Taiwanese Adults-A Population-Based Study in Taiwan

The relationship between gout and age-related macular degeneration (AMD) was suggested in previous literature but has yet to be accepted fully among physicians. This study aimed to explore the effect of gout on the development of age-related macular degeneration in Taiwan. A retrospective cohort study was conducted using Taiwan's National Health Insurance Database that includes a 2-million-persons dataset. The crude hazard ratio, Kaplan-Meier plot, and separate cox proportional hazard ratio were utilized to demonstrate the effect of gout on the development of age-related macular degeneration. The crude hazard ratio for gout patients developing AMD was 1.55 and the adjusted hazard ratio 1.20. In conclusion, gout is a risk factor for developing AMD, and achieving good disease management is therefore essential for preventing AMD from occurring.

Comparison of risk of serious cardiovascular events after haemorrhagic versus ischaemic stroke: a population-based study

BACKGROUND

Patients with ischaemic stroke are considered a very-high risk population for subsequent cardiovascular events and guidelines recommend intensive preventive strategies. In contrast, there is no clear recommendation that patients with haemorrhagic stroke should also be regarded as a very-high cardiovascular risk population.

OBJECTIVE

To compare the risk of subsequent cardiovascular morbidity/mortality between patients with incident haemorrhagic versus ischaemic stroke.

METHODS

Patients aged >18 years with incident haemorrhagic or ischaemic stroke between 1998-2017, and no prior history of serious vascular event were identified from UK Clinical Practice Research Datalink (CPRD-GOLD) linked to Hospital Episode Statistics (HES) data.

RESULTS

The cohort included 32,091 patients with an overall follow-up of 381,237 person-years (median: 11.8 years). After adjusting for potential confounders, patients with incident haemorrhagic stroke had no significantly different risk of subsequent cardiovascular morbidity compared with patients with incident ischaemic stroke - CHD [HR:0.86, 95%CI:0.56-1.32], recurrent stroke [HR:0.92, 95%CI:0.83-1.02], PVD [HR:1.15, 95%CI:0.56-2.38], or heart failure [HR:1.03, 95%CI:0.61-1.74]. Patients with incident haemorrhagic stroke had significantly higher risk of subsequent CVD-related mortality [HR:2.35, 95%CI:2.04-2.72] and all-cause mortality [HR:2.16, 95%CI: 1.94-2.41]. Propensity-score matched analysis of 1,039 patients with haemorrhagic stroke and 1,039 with ischaemic stroke showed similar risk in subsequent cardiovascular morbidity - CHD [stratified hazard ratio (sHR):0.92, 95%CI:0.55-1.54], recurrent stroke [sHR:0.93, 95%CI:0.82-1.02)], PVD [sHR:1.04 95%CI:0.45-2.41], or heart failure [HR:0.71, 95%CI:0.39-1.27].

CONCLUSIONS

The risk of subsequent cardiovascular events is similar between patients with incident haemorrhagic or ischaemic stroke. Patients with previous haemorrhagic stroke should be regarded as a population at very-high risk of subsequent CVD.

Introduction to Matching in Case-Control and Cohort Studies

Matching is a technique through which patients with and without an outcome of interest (in case-control studies) or patients with and without an exposure of interest (in cohort studies) are sampled from an underlying cohort to have the same or similar distributions of some characteristics. This technique is used to increase the statistical efficiency and cost efficiency of studies. In case-control studies, besides time in risk set sampling, controls are often matched for each case with respect to important confounding factors, such as age and sex, and covariates with a large number of values or levels, such as area of residence (e.g., post code) and clinics/hospitals. In the statistical analysis of matched case-control studies, fixed-effect models such as the Mantel-Haenszel odds ratio estimator and conditional logistic regression model are needed to stratify matched case-control sets and remove selection bias artificially introduced by sampling controls. In cohort studies, exact matching is used to increase study efficiency and remove or reduce confounding effects of matching factors. Propensity score matching is another matching method whereby patients with and without exposure are matched based on estimated propensity scores to receive exposure. If appropriately used, matching can improve study efficiency without introducing bias and could also present results that are more intuitive for clinicians.

Respiratory support after delayed cord clamping: a prospective cohort study of at-risk births at ≥35+0 weeks gestation

OBJECTIVE

To identify risk factors associated with delivery room respiratory support in at-risk infants who are initially vigorous and received delayed cord clamping (DCC).

DESIGN

Prospective cohort study.

SETTING

Two perinatal centres in Melbourne, Australia.

PATIENTS

At-risk infants born at ≥35⁺⁰ weeks gestation with a paediatric doctor in attendance who were initially vigorous and received DCC for >60 s.

MAIN OUTCOME MEASURES

Delivery room respiratory support defined as facemask positive pressure ventilation, continuous positive airway pressure and/or supplemental oxygen within 10 min of birth.

RESULTS

Two hundred and ninety-eight infants born at a median (IQR) gestational age of 39⁺³ (38⁺²-40⁺²) weeks were included. Cord clamping occurred at a median (IQR) of 128 (123-145) s. Forty-four (15%) infants received respiratory support at a median of 214 (IQR 156-326) s after birth. Neonatal unit admission for respiratory distress occurred in 32% of infants receiving delivery room respiratory support vs 1% of infants who did not receive delivery room respiratory support (p<0.001). Risk factors independently associated with delivery room respiratory support were average heart rate (HR) at 90-120 s after birth (determined using three-lead ECG), mode of birth and time to establish regular cries. Decision tree analysis identified that infants at highest risk had an average HR of <165 beats per minute at 90-120 s after birth following caesarean section (risk of 39%). Infants with an average HR of ≥165 beats per minute at 90-120 s after birth were at low risk (5%).

CONCLUSIONS

We present a clinical decision pathway for at-risk infants who may benefit from close observation following DCC. Our findings provide a novel perspective of HR beyond the traditional threshold of 100 beats per minute.

Development of a dual-factor measure of adolescent mental health: an analysis of cross-sectional data from the 2014 Canadian Health Behaviour in School-aged Children (HBSC) study

INTRODUCTION

Studies of adolescent mental health require valid measures that are supported by evidence-based theories. An established theory is the dual-factor model, which argues that mental health status is only fully understood by incorporating information on both subjective well-being and psychopathology.

OBJECTIVES

To develop a novel measure of adolescent mental health based on the dual-factor model and test its construct validity.

DESIGN

Cross-sectional analysis of national health survey data.

SETTING AND PARTICIPANTS

Nationally weighted sample of 21 993 grade 6-10 students; average age: 14.0 (SD 1.4) years from the 2014 Canadian Health Behaviour in School-aged Children study.

MEASURES

Self-report indicators of subjective well-being (life satisfaction, positive and negative affect), and psychopathology (psychological symptoms and overt risk-taking behaviour) were incorporated into the dual-factor measure. Characteristics of adolescents families, specific mental health indicators and measures of academic and social functioning were used in the assessment of construct validity.

RESULTS

Proportions of students categorised to the four mental health groups indicated by the dual-factor measure were 67.6% 'mentally healthy', 17.5% 'symptomatic yet content', 5.5% 'asymptomatic yet discontent' and 9.4% 'mentally unhealthy'. Being mentally healthy was associated with the highest functioning (greater social support and academic functioning) and being mentally unhealthy was associated with the worst. A one-unit increase (ranges=0-10) in peer support (OR 1.19; 95% CI 1.15 to 1.22), family support (OR 1.32; 95% CI 1.28 to 1.36), student support (OR 1.20; 95% CI 1.17 to 1.24) and average school marks (OR 1.18; 95% CI 1.10 to 1.27) increased the odds of being symptomatic yet content versus mentally unhealthy. Mentally healthy youth were the most likely to live with both parents (77% vs ≤65%) and report their family as well-off (62% vs ≤53%).

CONCLUSIONS

We developed a novel, construct valid dual-factor measure of adolescent mental health. This potentially provides a nuanced and comprehensive approach to the assessment of adolescent mental health that is direly needed.

The Structural and Intercultural Competence for Epidemiological Studies (SICES) guidelines: a 22-item checklist

Structural and intercultural competence approaches have been widely applied to fields such as medical training, healthcare practice, healthcare policies and health promotion. Nevertheless, their systematic implementation in epidemiological research is absent. Based on a scoping review and a qualitative analysis, in this article we propose a checklist to assess cultural and structural competence in epidemiological research: the Structural and Intercultural Competence for Epidemiological Studies guidelines. These guidelines are organised as a checklist of 22 items and consider four dimensions of competence (awareness and reflexivity, cultural and structural validation, cultural and structural sensitivity, and cultural and structural representativeness), which are applied to the different stages of epidemiological research: (1) research team building and research questions; (2) study design, participant recruitment, data collection and data analysis; and (3) dissemination. These are the first guidelines addressing structural and cultural competence in epidemiological inquiry.

Statin use and risk of joint replacement due to osteoarthritis and rheumatoid arthritis: a propensity-score matched longitudinal cohort study

OBJECTIVE

Statins are reported to have a potential benefit on progression of OA and on disease activity in RA, but existing evidence is conflicting. Our objective was to examine whether statins associate with reduction in the risk for joint replacement due to OA and RA.

METHODS

This was a propensity score-matched cohort study. Electronic health records from the UK Clinical Practice Research Datalink were used. We selected people prescribed statins and people never prescribed statins. Each statin user was matched to a non-user by age, gender, practice and propensity score for statin prescription. The main outcome measures were knee or hip joint replacement overall, and specifically because of OA or RA. The association between statins and risk of joint replacement was assessed using Cox proportional hazard regression. Statin exposure was categorized according to the potency of reducing low-density lipoprotein as low (21-28%), medium (32-38%) or high (42-55%) intensity.

RESULTS

A total of 178 467 statin users were matched with 178 467 non-users by age, gender, practice and propensity score. Overall, statin was not associated with reduced risk of knee or hip replacement (hazard ratio 0.99, 95% CI: 0.97, 1.03), unless prescribed at high strength (0.86, 0.75-0.98). The reduced risk was only observed for joint replacement due to RA (0.77, 0.63-0.94) but not OA (0.97, 0.94-1.01).

CONCLUSION

Statins at high intensity may reduce the risk of hip or knee replacement. This effect may be RA specific. Further studies to investigate mechanisms of risk reduction and the impact in people with RA are warranted.

Ambulatory blood pressure profile and stroke recurrence

OBJECTIVES

To establish a new ambulatory blood pressure (ABP) parameter (24-hour ABP profile) and evaluated its performance on stroke outcome in ischaemic stroke (IS) or transient ischaemic attack (TIA) patients.

METHODS

The prospective cohort consisted of 1996 IS/TIA patients enrolled for ABP monitoring and a 3-month follow-up for stroke recurrence as outcome. Profile groups of systolic blood pressure (SBP) were identified via an advanced functional clustering method, and the associations of the profile groups and conventional ABP parameters with stroke recurrence were examined in a Cox proportional hazards model.

RESULTS

Three discrete profile groups (n=604, 781 and 611 in profiles 1, 2 and 3, respectively) in 24-hour ambulatory SBP were identified. Profile 1 resembled most to the normal diurnal blood pressure pattern; profile 2 also dropped at night, but climbed earlier and with higher morning surge; while profile 3 had sustained higher nocturnal SBP without significant nocturnal SBP decline. The incidence of stroke recurrence was 2.9%, 3.9% and 5.5% in profiles 1, 2 and 3, respectively. After adjustment for covariates, profile 3 was significantly associated with higher risk of stroke recurrence with profile 1 as reference (HR 1.76, 95% CI: 1.00 to 3.09), while no significant difference was observed between profiles 2 and 1 (HR 1.22, 95% CI: 0.66 to 2.25). None of conventional ABP parameters showed significant associations with the outcome.

CONCLUSIONS

Ambulatory 24-hour SBP profile is associated with short-term stroke recurrence. Profiles of ABP may help improve identification of stroke recurrence by capturing the additive effects of individual ABP parameters.

Risk of recurrent corneal erosion in patients with diabetes mellitus in Taiwan: a population-based cohort study

OBJECTIVE

To investigate the risk of recurrent corneal erosion (RCE) in patients with diabetes mellitus (DM).

DESIGN, SETTING AND PARTICIPANTS

This retrospective, nationwide, matched cohort study included 239 854 patients with DM recruited between 2003 and 2005 from the Longitudinal Cohort of Diabetes Patients database. The control group included the same number of age-matched and sex-matched patients without DM selected from the Taiwan Longitudinal Health Insurance Database, 2000. Data for each patient were collected from the index date until December 2013.

MAIN OUTCOMES AND MEASURES

The incidence and risk of RCE were compared between the two groups. Cox proportional hazards regression was used to calculate the HR for RCE after adjustment for potential confounders. The cumulative RCE incidence rate was calculated using Kaplan-Meier analysis.

RESULTS

In total, 1236 patients with DM and 884 controls developed RCE during the follow-up period, resulting in an incidence rate of RCE in patients with DM (5.87/10 000 person-years (PY)) higher than that in the controls (4.23/10 000 PY). After adjustment for potential confounders, including hypertension, hyperlipidaemia, chronic renal disease and keratoconjunctivitis sicca, patients with DM were 1.35 times (95% CI, 1.24 to 1.48) more likely to develop RCE than the total sample cohort.

CONCLUSIONS

DM increases the risk of RCE, which is an interdisciplinary issue. Therefore, close collaboration between endocrinologists and ophthalmologists is important in managing RCE following DM.

Association of the Work Schedules of Hospitalists With Patient Outcomes of Hospitalization

IMPORTANCE

The working schedules of hospitalists vary widely. Discontinuous schedules, such as 24 hours on and 48 hours off, result in several hospitalists providing care during a patient's hospital stay. Poor continuity of care during hospitalization may be associated with poor patient outcomes.

OBJECTIVE

To determine whether admitted patients receiving care from hospitalists with more discontinuous schedules experience worse outcomes.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study used conditional models to assess Medicare claims data for 114 777 medical admissions of patients with a 3-day to 6-day length of stay from January 1, 2014, through November 30, 2016, who received all general medical care from hospitalists in 229 hospitals in Texas. Data were analyzed from November 2018 to June 2019.

EXPOSURES

For each admission, the weighted mean of schedule continuity for the treating hospitalists, assessed as the percentage of all their working days in that year that were part of a block of 7 or more consecutive working days, was calculated.

MAIN OUTCOMES AND MEASURES

The primary outcome was patient mortality in the 30 days after discharge. Secondary outcomes were readmission rates and Medicare costs in the 30 days after discharge, and discharge destination.

RESULTS

Of the 114 777 patient admissions, the mean (SD) age was 79.9 (8.3) years, and 70 047 (61.0%) were women. For admissions in the lowest quartile for continuity of hospitalist schedules, the hospitalists providing care worked 0% to 30% of their total working days as part of a block of 7 or more consecutive days vs 67% to 100% for hospitalists providing care for patients in the highest quartile for continuity. Patient characteristics were not associated with the continuity of working schedules for the hospitalist(s) providing care. In conditional logistic regression models, admitted patients cared for by hospitalists in the highest quartile of schedule continuity (vs the lowest quartile) had lower 30-day mortality after discharge (adjusted odds ratio [aOR], 0.88; 95% CI, 0.81-0.95), lower readmission rates (aOR, 0.94; 95% CI, 0.90-0.99), higher rates of discharge to the home (aOR, 1.08; 95% CI, 1.03-1.13), and lower 30-day postdischarge costs (-$223; 95% CI, -$441 to -$7). The results were similar across a range of different methods for defining continuity of hospitalist schedules and selecting the cohort.

CONCLUSIONS AND RELEVANCE

Hospitalist schedules vary widely. Admitted patients receiving care from hospitalists with schedules that promote inpatient continuity of care may experience better outcomes of hospitalization.

Hazard Ratio Estimators after Terminating Observation within Matched Pairs in Sibling and Propensity Score Matched Designs

Similar to unmatched cohort studies, matched cohort studies may suffer from the censoring of events prior to the end of follow-up. Moreover, in some matched-pair cohort studies, observation time is prematurely terminated immediately after the follow-up of his/her matched member is completed by an event or censoring. Although the follow-up termination within matched pairs may or may not change the hazard ratio estimators, when and how the change occurs has not been clarified. We study the change in the estimates of the hazard ratio conditional on matched pairs and/or covariates by considering two types of matched-pair designs in cohort studies-sibling pair matching and propensity score matching-in which termination can be naturally considered. If all possible confounders are shared within the matched pairs, after termination, a wide range of hazard ratio estimators coincides with that obtained from a stratified Cox model. If unshared confounders should be adjusted for in the analysis, however, such coincidence is not observed. Simulation studies on sibling designs with unshared confounders suggested that the pair-stratified covariate-adjusted Cox model for the hazard ratio conditional on matched pairs and covariates is generally preferred, for which termination does not deteriorate the estimation. Conversely, the comparison between stratifying or not stratifying on pair is a more subtle issue in propensity score matching which targets a marginal or covariate-conditional hazard ratio. Based on simulation studies considering Cox models after matching based on estimated propensity scores, we discourage pair-stratified analysis and termination, particularly after data collection.