A competing risks analysis should report results on all cause-specific hazards and cumulative incidence functions

Sample size determination for jointly testing a cause-specific hazard and the all-cause hazard in the presence of competing risks

This article considers sample size determination for jointly testing a cause-specific hazard and the all-cause hazard for competing risks data. The cause-specific hazard and the all-cause hazard jointly characterize important study end points such as the disease-specific survival and overall survival, which are commonly used as coprimary end points in clinical trials. Specifically, we derive sample size calculation methods for 2-group comparisons based on an asymptotic chi-square joint test and a maximum joint test of the aforementioned quantities, taking into account censoring due to lost to follow-up as well as staggered entry and administrative censoring. We illustrate the application of the proposed methods using the Die Deutsche Diabetes Dialyse Studies clinical trial. An R package "powerCompRisk" has been developed and made available at the CRAN R library.

Different competing risks models for different questions may give similar results in arthroplasty registers in the presence of few events

Background and purpose - In arthroplasty registry studies, the analysis of time to revision is complicated by the competing risk of death. There are no clear guidelines for the choice between the 2 main adjusted analysis methods, cause-specific Cox and Fine-Gray regression, for orthopedic data. We investigated whether there are benefits, such as insight into different aspects of progression to revision, to using either 1 or both regression methods in arthroplasty registry studies in general, and specifically when the length of follow-up is short relative to the expected survival of the implants. Patients and methods - Cause-specific Cox regression and Fine-Gray regression were performed on total hip (138,234 hips, 124,560 patients) and knee (139,070 knees, 125,213 patients) replacement data from the Dutch Arthroplasty Register (median follow-up 3.1 years, maximum 8 years), with sex, age, ASA score, diagnosis, and type of fixation as explanatory variables. The similarity of the resulting hazard ratios and confidence intervals was assessed visually and by computing the relative differences of the resulting subdistribution and cause-specific hazard ratios. Results - The outcomes of the cause-specific Cox and Fine-Gray regressions were numerically very close. The largest relative difference between the hazard ratios was 3.5%. Interpretation - The most likely explanation for the similarity is that there are relatively few events (revisions and deaths), due to the short follow-up compared with the expected failure-free survival of the hip and knee prostheses. Despite the similarity, we recommend always performing both cause-specific Cox and Fine-Gray regression. In this way, both etiology and prediction can be investigated.

Lung Dysfunction and Mortality in Patients with Chronic Kidney Disease

BACKGROUND/AIMS

Lung dysfunction associates with increased mortality but the impact of chronic kidney disease (CKD) is less clear. We evaluated lung function and its association with mortality among individuals with normal to severely reduced glomerular filtration rate (GFR).

METHODS

404 individuals representing GFR category G1 (n=31; GFR >90 mL/min/1.73 m2), G2 (n=46), G3 (n=33), G4 (n=49) and G5 (n=245; GFR< 15 mL/min/1.73 m2) underwent spirometry yielding lung function indices forced vital capacity (FVC), forced expiratory volume in the first second (FEV1) and peak expiratory flow (PEF). Associations of lung function indices expressed as percentages of predicted values (%FEV1, %FVC and %PEF) with 5-year mortality were analyzed by competing-risk regression models.

RESULTS

The prevalence of obstructive (6% in G1 and 11% in G5) and especially restrictive (9% in G1 to 36% in G5) lung dysfunction increased with declining GFR and with higher comorbidity burden. In patients (n=22) with protein-energy wasting, inflammation and cardiovascular disease, the prevalence of restrictive lung function was 64%. The highest tertiles of % FEV1 and %FVC associated with lower sub-hazard ratios (sHR) for all-cause mortality, 0.49 (95% CI, 0.27-0.88)) and 0.56 (95% CI, 0.32-0.98), and that of %FEV1 also with lower cardiovascular mortality risk (sHR 0.16; 95%CI 0.04-0.69) after adjusting for multiple confounders. Restrictive lung dysfunction (FEV1/FVC ≥ 0.70, and %FVC < 80) associated with increased mortality risk (sHR 1.80, 95%CI, 1.04-3.13) while the association with obstructive lung impairment was not statistically significant.

CONCLUSION

Lung dysfunction and in particular restrictive lung dysfunction associates with degree of renal function impairment and presence of comorbidities, and is an independent predictor of increased mortality in CKD patients.

Missingness in the Setting of Competing Risks: from missing values to missing potential outcomes

Purpose of review

The setting of competing risks in which there is an event that precludes the event of interest from occurring is prevalent in epidemiological research. Unless studying all-cause mortality, any study following up individuals is subject to having a competing risk should individuals die during time period that the study covers. While there are prior papers discussing the need for competing risk methods in epidemiologic research, we are not aware of any review that discusses issues of missing data in a competing risk setting.

Recent Findings

We provide an overview of causal inference in competing risks as potential outcomes are missing, provide some strategies in dealing with missing (or misclassified) event type, and missing covariate data in competing risks. The strategies presented are specifically focused on those that may easily be implemented in standard statistical packages. There is ongoing work in terms of causal analyses, dealing with missing event type information, and missing covariate values specific to competing risk analyses.

Summary

Competing events are common in epidemiologic research. While there has been a focus on why one should conduct a proper competing risk analysis, a perhaps unrecognized issue is in terms of missingness. Strategies exist to minimize the impact of missingness in analyses of competing risks.

Competing Risk of Death With End-Stage Renal Disease in Diabetic Kidney Disease

The concept of competing risks is particularly relevant to survival analyses of diabetic ESRD given the high likelihood of death prior to ESRD. Approaches such as Kaplan-Meier curves and Cox regression models operate on the assumption that there are no competing risks for the event of interest, yielding uninterpretable and generally biased estimates in the presence of competing risks. The cumulative incidence function and Fine-Gray regression are more appropriate methodologies for survival analysis when competing risks are present. We present an example taken from the Action to Control Cardiovascular Risk in Diabetes, a randomized trial of people with type 2 diabetes at high risk for cardiovascular disease. Participants were stratified according to baseline markers of kidney disease: (1) no kidney disease; (2) low estimated glomerular filtration rate; (3) microalbuminuria alone; and (4) macroalbuminuria. The macroalbuminuria group had the highest risk for ESRD and demonstrated the most marked difference between the Kaplan-Meier and cumulative incidence estimator. Cox and Fine-Gray regression models yielded similar risk estimates for baseline characteristics, with the exception of diabetes duration, which was significant in the Cox but not Fine-Gray model. We underscore the importance of competing risk methods, particularly when the competing risk is common, as is the case in diabetic kidney disease.

Bias Due to Confounders for the Exposure-Competing Risk Relationship

BACKGROUND

Epidemiologic studies that aim to estimate a causal effect of an exposure on a particular event of interest may be complicated by the existence of competing events that preclude the occurrence of the primary event. Recently, many articles have been published in the epidemiologic literature demonstrating the need for appropriate models to accommodate competing risks when they are present. However, there has been little attention to variable selection for confounder control in competing risk analyses.

METHODS

We employ simulation to demonstrate the bias in two variable selection strategies include covariates that are associated with the exposure and (1) which change the cause-specific hazard of any of the outcomes; or (2) which change the cause-specific hazard of the specific event of interest.

RESULTS

We demonstrated minimal to no bias in estimators adjusted for confounders of exposure and either the event of interest or the competing event, but bias of varying magnitude in almost all estimators adjusted only for confounders of exposure and the primary outcome.

DISCUSSION

When estimating causal effects for which there are competing risks, the analysis should control for confounders of both the exposure-primary outcome effect and of the exposure-competing outcome effect.

Direct likelihood inference on the cause-specific cumulative incidence function: A flexible parametric regression modelling approach

In a competing risks analysis, interest lies in the cause-specific cumulative incidence function (CIF) that can be calculated by either (1) transforming on the cause-specific hazard or (2) through its direct relationship with the subdistribution hazard. We expand on current competing risks methodology from within the flexible parametric survival modelling framework (FPM) and focus on approach (2). This models all cause-specific CIFs simultaneously and is more useful when we look to questions on prognosis. We also extend cure models using a similar approach described by Andersson et al for flexible parametric relative survival models. Using SEER public use colorectal data, we compare and contrast our approach with standard methods such as the Fine & Gray model and show that many useful out-of-sample predictions can be made after modelling the cause-specific CIFs using an FPM approach. Alternative link functions may also be incorporated such as the logit link. Models can also be easily extended for time-dependent effects.

Pregnancy outcomes in women on metformin for diabetes or other indications among those seeking teratology information services

AIMS

Metformin is used to treat type 2 diabetes, polycystic ovary syndrome associated infertility, and gestational diabetes. This study aims to evaluate the safety of metformin in early pregnancy.

METHOD

We evaluated the risk of major birth defects and pregnancy losses in a cohort of pregnant women exposed to metformin during the first trimester for different indications relative to a matched unexposed reference group.

RESULTS

The risk of major birth defects was 5.1% (20/392) in pregnancies exposed to metformin during the first trimester and 2.1% (9/431) in the reference group [adjusted odds ratio (OR) 1.70; 95% CI 0.70-4.38]. Among metformin users, this risk was 7.8% (17/219) in patients with pre-gestational diabetes and 1.7% (3/173) in those without this diagnosis. Compared to the unexposed reference, the OR for metformin user with diabetes was 3.95 (95% CI 1.77-9.41) and for metformin with other indications it was 0.83 (95% CI 0.18-2.81). The risk of pregnancy losses (spontaneous abortions and stillbirths) was 20.8% in women on metformin during the first trimester and 10.8% in the reference group [adjusted hazard ratio (HR) 1.57; 95% CI 0.90-2.74]. The risks for women on metformin with and without pre-gestational diabetes were 24.0% and 16.8% respectively, with adjusted HR of 2.51 (95% CI 1.44-4.36) and 1.38 (95% CI 0.74-2.59) when compared to the reference.

CONCLUSION

Pregnant women with pre-gestational diabetes on metformin are at a higher risk for adverse pregnancy outcomes than the general population. This appears to be due to the underlying diabetes since women on metformin for other indications do not present meaningfully increased risks.

An examination of the prospective association between religious service attendance and suicide: Explanatory factors and period effects

BACKGROUND

We addressed two unanswered questions from prior research, demonstrating a prospective association between frequent religious service attendance and decreased risk for suicide. First, we assessed whether religious service attendance conferred protection from suicide even after accounting for strength of religious affiliation. Second, we evaluated whether the relationship between religious service attendance and suicide was subject to period effects.

METHODS

Data were drawn from the 1978-2010 General Social Survey, a nationally representative study of 30,650 non-institutionalized, English-speaking American residents age 18 or older. Data were linked with the National Death Index through the end of 2014. We analyzed these data using moderated Cox proportional hazard analyses.

RESULTS

Religious affiliation had no relationship with suicide. Religious service attendance only had a protective effect against suicide death among those in later (2000-2010) rather than earlier (1998 and earlier) data collection periods.

LIMITATIONS

Secondary analysis of data limited the types of variables that were available.

CONCLUSIONS

The protective nature of religion is due more to participating in religious activities, such as attending religious services, than to having a strong religious affiliation, and this effect exists primarily in more recent data collection periods.

Mortality of hospitalised internal medicine patients bedspaced to non-internal medicine inpatient units: retrospective cohort study

OBJECTIVE

To compare inhospital mortality of general internal medicine (GIM) patients bedspaced to off-service wards with GIM inpatients admitted to assigned GIM wards.

METHOD

A retrospective cohort study of consecutive GIM admissions between 1 January 2015 and 1 January 2016 was conducted at a large tertiary care hospital in Canada.Inhospital mortality was compared between patients admitted to off-service wards (bedspaced) and assigned GIM wards using a Cox proportional hazards model and a competing risk model. Sensitivity analyses included propensity score and pair matching based on GIM service team, workload, demographics, time of admission, reasons for admission and comorbidities.

RESULTS

Among 3243 consecutive GIM admissions, more than a third (1125, 35%) were bedspaced to off-service wards with the rest (2118, 65%) admitted to assigned GIM wards. In hospital, 176 (5%) patients died: 88/1125 (8%) bedspaced patients and 88/2118 (4%) assigned GIM ward patients. Compared with assigned GIM wards patients, bedspaced patients had an HR of 3.42 (95% CI 2.23 to 5.26; P<0.0001) for inhospital mortality at admission, which then decreased by HR of 0.97 (95% CI 0.94 to 0.99; P=0.0133) per day in hospital. Competing risk models and sensitivity analyses using propensity scores and pair matching yielded similar results.

CONCLUSIONS

Bedspaced patients had significantly higher inhospital mortality than patients admitted to assigned GIM wards. The risk was highest at admission and subsequently declined. The results of this single centre study may not be generalisable to other hospitals and may be influenced by residual confounding. Despite these limitations, the relationship between bedspacing and patient outcomes requires investigation at other institutions to determine if this common practice represents a modifiable patient safety indicator.

Análise de sobrevivência de mulheres com câncer de mama: modelos de riscos competitivos

Resumo O objetivo deste estudo foi estimar os efeitos de fatores prognósticos na sobrevida do câncer de mama, como idade, estadiamento e extensão do tumor, utilizando modelos de riscos proporcionais de Cox e de riscos competitivos de Fine-Gray. É um estudo de coorte retrospectiva de base-populacional referente a 524 mulheres diagnosticadas com câncer de mama no período de 1993 a 1995, acompanhadas até 2011, residentes no município de Campinas, São Paulo, Brasil. O ponto de corte (cutoff) da variável idade foi definido utilizando-se modelos simples de Cox. Nos ajustes de modelos simples e múltiplo de Fine-Gray, a idade não foi significativa na presença de riscos competitivos e nem nos modelos de Cox, considerando-se, para ambas as modelagens, óbito por câncer de mama como desfecho de interesse. As curvas de sobrevidas estimadas por Kaplan-Meier evidenciaram diferenças expressivas para óbitos por câncer de mama e por riscos competitivos. As curvas de sobrevida por câncer de mama não apresentaram diferenças significativas quando comparados os grupos de idades, segundo teste de log rank. Os modelos de Cox e de Fine-Gray identificaram os mesmos fatores prognósticos que influenciavam na sobrevida do câncer de mama.

Circulating tumor cell count and thrombosis in metastatic breast cancer

Essentials Tumor cells circulating in blood (CTC) may favor thrombotic events in cancer patients. We assessed the impact of CTC on the risk of thrombosis in metastatic breast cancer. Baseline CTC detection was the only independent factor associated with the risk of thrombosis. CTC detection under therapy may be the hidden link between tumor progression & thrombosis.

SUMMARY

Background Circulating tumor cell (CTC) count is a major prognostic factor in metastatic breast cancer (MBC) and has been reported to be associated with thrombosis in short-term studies on MBC patients. Objective To assess whether CTC detection (CellSearch® ) before first-line chemotherapy impacts the risk of thrombosis throughout the course of MBC. Patients/Methods Among patients included before first-line chemotherapy for MBC in the prospective IC2006-04 CTC detection study (NCT00898014), the electronic medical files of those patients treated at Institut Curie (Paris, France) were searched in silico and manually checked for incident venous or arterial thrombotic events (TE) in the course of MBC. Univariate and multivariate analyses were performed using Cox and Fine-Gray models, adjusted for age and Khorana score. Results/Conclusions With a median follow-up of 64 months (25-81 months), among the 142 patients included, 34 (24%) experienced a TE (incidence rate, 8 TE/100 patient-years). The TE incidence rate was 13 TE/100 patient-years for the 80 patients with ≥ 1 CTC/7.5 mL of blood before initiating first-line chemotherapy, vs. only 4 TE/100 patient-years for the 62 CTC-negative patients. Fine-Gray multivariate analysis (with death as competing event) included age, Khorana score and baseline lactate dehydrogenase and CTC levels: detectable CTC was the only factor significantly associated with an increased risk of TE (sub-distribution hazard ratio [SHR] for patients with [1-4] CTC = 3.1, 95% CI [1.1; 8.6], SHR for patients with ≥ 5 CTC = 1.4, 95% CI [0.5; 4.6]). This study shows that CTC detection before starting first-line chemotherapy is an independent risk factor for TE in MBC patients.

Practical recommendations for reporting Fine-Gray model analyses for competing risk data

In survival analysis, a competing risk is an event whose occurrence precludes the occurrence of the primary event of interest. Outcomes in medical research are frequently subject to competing risks. In survival analysis, there are 2 key questions that can be addressed using competing risk regression models: first, which covariates affect the rate at which events occur, and second, which covariates affect the probability of an event occurring over time. The cause‐specific hazard model estimates the effect of covariates on the rate at which events occur in subjects who are currently event‐free. Subdistribution hazard ratios obtained from the Fine‐Gray model describe the relative effect of covariates on the subdistribution hazard function. Hence, the covariates in this model can also be interpreted as having an effect on the cumulative incidence function or on the probability of events occurring over time. We conducted a review of the use and interpretation of the Fine‐Gray subdistribution hazard model in articles published in the medical literature in 2015. We found that many authors provided an unclear or incorrect interpretation of the regression coefficients associated with this model. An incorrect and inconsistent interpretation of regression coefficients may lead to confusion when comparing results across different studies. Furthermore, an incorrect interpretation of estimated regression coefficients can result in an incorrect understanding about the magnitude of the association between exposure and the incidence of the outcome. The objective of this article is to clarify how these regression coefficients should be reported and to propose suggestions for interpreting these coefficients.

Risk of Alzheimer's Disease Among Senior Medicare Beneficiaries Treated With Androgen Deprivation Therapy for Prostate Cancer

Purpose To assess the relative risk of Alzheimer's disease (AD) among patients with prostate cancer who received androgen deprivation therapy (ADT), after adjustment for other cancer therapies. Methods Data from demographics, survival, diagnoses codes, procedure codes, and other information about beneficiaries age 67 years or older in the Medicare claims database was assessed to determine the unadjusted and adjusted risks of AD and of dementia from ADT. The prespecified survival analysis method was competing risk regression. Results Of the 1.2 million fee-for-service Medicare beneficiaries who developed prostate cancer in 2001 to 2014, 35% received ADT. Of these, 109,815 (8.9%) and 223,765 (18.8%) developed AD and dementia, respectively, and 26% to 33% died without either outcome. Unadjusted rates of AD and all-cause mortality per 1,000 patient-years were higher among ADT recipients; the unadjusted rates of AD were 17.0 and 15.5 per 1,000 person-years in recipients and nonrecipients, respectively, and the unadjusted rates of all-cause mortality were 73.0 and 51.6 per 1,000 person-years, respectively. The unadjusted rates for dementia in ADT recipients versus nonrecipients were 38.5 and 32.9, respectively, and the unadjusted rates of mortality were 60.2 versus 40.4, respectively. However, after analysis was adjusted for other cancer therapies and other covariates, patients with ADT treatment had no increased risk of AD (subdistribution hazard ratio [SHR], 0.98; 95% CI, 0.97 to 0.99) and had only a miniscule (1%) risk of dementia (SHR, 1.01; 95% CI, 1.01 to 1.02); patients treated with ADT were more likely to die before progression to AD (SHR, 1.24; 95% CI, 1.23 to 1.24) or dementia (SHR, 1.26; 95% CI, 1.25 to 1.26). The risks of AD and dementia were not associated with duration of ADT (ie, no dose effect). Other secondary analyses confirmed these results. Conclusion These data suggest that ADT treatment has no hazard for AD and no meaningful hazard for dementia among men age 67 years or older who are enrolled in Medicare.

Outcome of revascularisation in stable coronary artery disease without ischaemia: a Danish registry-based follow-up study

OBJECTIVES

In stable coronary artery disease (CAD), coronary revascularisation may reduce mortality of patients with a certain amount of left ventricular myocardial ischaemia. However, revascularisation does not always follow the guidance suggested by ischaemia testing. We compared outcomes in patients without ischaemia who had either revascularisation or medical treatment.

DESIGN AND POPULATION

Based on registries, 1327 consecutive patients with normal myocardial perfusion scintigraphy (MPS) and 278 with fixed perfusion defects were followed for a median of 6.1 years. Most patients received medical therapy alone (Med), but 26 (2%) with a normal MPS and 15 (5%) with fixed perfusion defects underwent revascularisation (Revasc).

OUTCOME MEASURES

Incidence rates of all-cause death (ACD) and rates of cardiac death/myocardial infarction (CD/MI).

RESULTS

With a normal MPS, the ACD rate was 6.2%/year in the Revasc group versus 1.9%/year in the Med group (p=0.01); the CD/MI rates were 6.9%/year and 0.6%/year, respectively (p<0.00001). Results persisted after adjustment for predictors of revascularisation, in particular angina score, and in comparisons of matched Revasc and Med patients. With fixed defects, the ACD rate was 9.1%/year in the Revasc group and 6.7%/year in the Med group (p=0.44); the CD/MI rate was 5.0%/year versus 4.2%/year, respectively (p=0.69). If adjusted for angiographic variables or analysed in matched subsets, differences remained insignificant.

CONCLUSIONS

With normal MPS, revascularisation conferred a higher risk, even after adjustment for predictors of revascularisation. With fixed defects, the Revascversus Med difference was close to equipoise. Hence, in patients with stable CAD without ischaemia, we could not find evidence to justify exceptional revascularisation.

Access to Heart Transplantation: A Proper Analysis of the Competing Risks of Death and Transplantation Is Required to Optimize Graft Allocation

Supplemental digital content is available in the text.

Background

Heart allocation systems are usually urgency-based, offering grafts to candidates at high risk of waitlist mortality. In the context of a revision of the heart allocation rules, we determined observed predictors of 1-year waitlist mortality in France, considering the competing risk of transplantation, to determine which candidate subgroups are favored or disadvantaged by the current allocation system.

Methods

Patients registered on the French heart waitlist between 2010 and 2013 were included. Cox cause-specific hazards and Fine and Gray subdistribution hazards were used to determine candidate characteristics associated with waitlist mortality and access to transplantation.

Results

Of the 2053 candidates, 7 variables were associated with 1-year waitlist mortality by the Fine and Gray method including 4 candidate characteristics related to heart failure severity (hospitalization at listing, serum natriuretic peptide level, systolic pulmonary artery pressure, and glomerular filtration rate) and 3 characteristics not associated with heart failure severity but with lower access to transplantation (blood type, age, and body mass index). Observed waitlist mortality for candidates on mechanical circulatory support was like that of others.

Conclusions

The heart allocation system strongly modifies the risk of pretransplant mortality related to heart failure severity. An in-depth competing risk analysis is therefore a more appropriate method to evaluate graft allocation systems. This knowledge should help to prioritize candidates in the context of a limited donor pool.

Single umbilical cord blood with or without CD34+ cells from a third-party donor in adults with leukemia

We retrospectively compared the clinical outcomes of adults with acute leukemia who received single-unit umbilical cord blood (UCB) transplantation (sUCBT) (n = 135) or stem cell transplant using coinfusion of a UCB graft with CD34⁺ cells from a third-party donor (Haplo-Cord) (n = 72) at different institutions within the Grupo Español de Trasplante Hematopoyético. In multivariable analysis, patients in the Haplo-Cord group showed more rapid neutrophil (hazard ratio [HR], 2.3; 95% confidence interval [CI], 1.5-3.3; P < .001) and platelet recovery (HR, 1.6; 95% CI, 1.2-2.3; P = .015) and lower incidence of chronic graft-versus-host disease (GVHD) (relative risk, 0.5; 95% CI, 0.3-0.8; P = .01). Nonrelapse mortality, relapse, disease-free survival (DFS), and GVHD/relapse-free survival were similar in the 2 groups. Regarding disease-specific outcomes, DFS in both acute myeloid leukemia (AML) and acute lymphoblastic leukemia patients was not significantly different; however, a significantly higher relapse rate was found in patients with AML treated with Haplo-Cord (HR, 2.3; 95% CI, 1-5.4; P = .04). Our study confirms that Haplo-Cord was an effective strategy to accelerate neutrophil and platelet recovery and shows that, in the context of specific treatment platforms, sUCBT and Haplo-Cord offer similar long-term outcomes.

Flexible parametric modelling of the cause-specific cumulative incidence function

Competing risks arise with time-to-event data when individuals are at risk of more than one type of event and the occurrence of one event precludes the occurrence of all other events. A useful measure with competing risks is the cause-specific cumulative incidence function (CIF), which gives the probability of experiencing a particular event as a function of follow-up time, accounting for the fact that some individuals may have a competing event. When modelling the cause-specific CIF, the most common model is a semi-parametric proportional subhazards model. In this paper, we propose the use of flexible parametric survival models to directly model the cause-specific CIF where the effect of follow-up time is modelled using restricted cubic splines. The models provide smooth estimates of the cause-specific CIF with the important advantage that the approach is easily extended to model time-dependent effects. The models can be fitted using standard survival analysis tools by a combination of data expansion and introducing time-dependent weights. Various link functions are available that allow modelling on different scales and have proportional subhazards, proportional odds and relative absolute risks as particular cases. We conduct a simulation study to evaluate how well the spline functions approximate subhazard functions with complex shapes. The methods are illustrated using data from the European Blood and Marrow Transplantation Registry showing excellent agreement between parametric estimates of the cause-specific CIF and those obtained from a semi-parametric model. We also fit models relaxing the proportional subhazards assumption using alternative link functions and/or including time-dependent effects. Copyright © 2016 John Wiley & Sons, Ltd.

Epidemiology of low-proteinuric chronic kidney disease in renal clinics

CKD patients with low-grade proteinuria (LP) are common in nephrology clinics. However, prevalence, characteristics, and the competing risks of ESRD and death as the specific determinants, are still unknown. We analyzed epidemiological features of LP status in a prospective cohort of 2,340 patients with CKD stage III-V referred from ≥6 months in 40 nephrology clinics in Italy. LP status was defined as proteinuria <0.5 g/24h according to current KDIGO guidelines. Patients with higher proteinuria constituted the control group (CON). LP patients were 54.5% of the whole cohort. As compared to CON, LP were older (70.0±12.1 vs 65.4±14.1 y), and less likely to be male (55.8 vs 62.0%) and diabetic (27.6 vs 34.1%), and had hypertension as the most common cause of CKD (39.8%). They had higher eGFR (34.8±13.5 vs 26.8±13.2 mL/min/1.73m²) and hemoglobin (12.7±1.7 vs 12.3±1.7 g/dL), while systolic blood pressure (137±18 vs 140±18 mmHg) and serum phosphorus (3.7±0.8 vs 3.9±0.8 mg/dL) were lower [P<0.001 for all comparisons]. Over a median follow-up of 48 months, an inverse relative risk of ESRD and death was observed in LP (death>>ESRD; P = 0.002) versus CON (ESRD>>death; P<0.0001). Modifiable risk factors were also different in LP, with smoking, lower hemoglobin, and proteinuria being associated with higher mortality risk while lower BMI and higher phosphorus predicting ESRD at multivariable Cox analyses [P<0.05 for all]. Therefore, in nephrology clinics, LP patients are the majority and show distinctive basal features. More important, they are more exposed to death than ESRD and do present specific modifiable determinants of either outcome; indeed, in LP, while smoking plays a role for mortality, lower BMI and higher phosphorus levels -even if in the normal range- are predictors of ESRD. These data support the need to further study the low proteinuric CKD population to guide management.

Low renal replacement therapy incidence among slowly progressing elderly chronic kidney disease patients referred to nephrology care: an observational study

Background

Elderly patients with advanced chronic kidney disease (CKD) have a high risk of death before reaching end-stage kidney disease. In order to allocate resources, such as advanced care nephrology where it is most needed, it is essential to know which patients have the highest absolute risk of advancing to renal replacement therapy (RRT).

Methods

We included all nephrology-referred CKD stage 3b-5 patients in Sweden 2005–2011 included in the Swedish renal registry (SRR-CKD) who had at least two serum creatinine measurements one year apart (+/− 6 months). We followed these patients to either initiation of RRT, death, or September 30, 2013. Decline in estimated glomerular filtration rate (eGFR) (%) was estimated during the one-year baseline period. The patients in the highest tertile of progression (>18.7% decline in eGFR) during the initial year of follow-up were classified as “fast progressors”. We estimated the cumulative incidence of RRT and death before RRT by age, eGFR and progression status using competing risk models.

Results

There were 2119 RRT initiations (24.2%) and 2060 deaths (23.5%) before RRT started. The median progression rate estimated during the initial year was −8.8% (Interquartile range [IQR] - 24.5–6.5%). A fast initial progression rate was associated with a higher risk of RRT initiation (Sub Hazard Ratio [SHR] 2.24 (95% confidence interval [CI] 2.00–2.51) and also a higher risk of death before RRT initiation (SHR 1.27 (95% CI 1.13–1.43). The five year probability of RRT was highest in younger patients (<65 years) with fast initial progression rate (51% in CKD stage 4 and 76% in stage 5), low overall in patients >75 years with a slow progression rate (7, 13, and 25% for CKD stages 3b, 4 and 5 respectively), and slightly higher in elderly patients with a fast initial progression rate (28% in CKD stage 4 and 47% in CKD stage 5) or with diabetic kidney disease.

Conclusions

The 5-year probability of RRT was low among referred slowly progressing CKD patients >75 years of age because of the competing risk of death.

Electronic supplementary material

The online version of this article (doi:10.1186/s12882-017-0473-1) contains supplementary material, which is available to authorized users.

Survival biases lead to flawed conclusions in observational treatment studies of influenza patients

BACKGROUND AND OBJECTIVE

Several observational studies reported that Oseltamivir (Tamiflu) reduced mortality in infected and hospitalized patients. Because of the restriction of observation to hospital stay and time-dependent treatment assignment, such findings were prone to common types of survival bias (length, time-dependent and competing risk bias).

METHODS

British hospital data from the Influenza Clinical Information Network (FLU-CIN) study group were used which included 1,391 patients with confirmed pandemic influenza A/H1N1 2009 infection. We used a multistate model approach with following states: hospital admission, Oseltamivir treatment, discharge, and death. Time origin is influenza onset. We displayed individual data, risk sets, hazards, and probabilities from multistate models to study the impact of these three common survival biases.

RESULTS

The correct hazard ratio of Oseltamivir for death was 1.03 (95% confidence interval [CI]: 0.64-1.66) and for discharge 1.89 (95% CI: 1.65-2.16). Length bias increased both hazard ratios (HRs): HR (death) = 1.82 (95% CI: 1.12-2.98) and HR (discharge) = 4.44 (95% CI: 3.90-5.05), whereas the time-dependent bias reduced them: HR (death) = 0.62 (95% CI: 0.39-1.00) and HR (discharge) = 0.85 (95% CI: 0.75-0.97). Length and time-dependent bias were less pronounced in terms of probabilities. Ignoring discharge as a competing event for hospital death led to a remarkable overestimation of hospital mortality and failed to detect the reducing effect of Oseltamivir on hospital stay.

CONCLUSIONS

The impact of each of the three survival biases was remarkable, and it can make neuraminidase inhibitors appear more effective or even harmful. Incorrect and misclassified risk sets were the primary sources of biased hazard rates.

Survival analysis in the presence of competing risks

Survival analysis in the presence of competing risks imposes additional challenges for clinical investigators in that hazard function (the rate) has no one-to-one link to the cumulative incidence function (CIF, the risk). CIF is of particular interest and can be estimated non-parametrically with the use cuminc() function. This function also allows for group comparison and visualization of estimated CIF. The effect of covariates on cause-specific hazard can be explored using conventional Cox proportional hazard model by treating competing events as censoring. However, the effect on hazard cannot be directly linked to the effect on CIF because there is no one-to-one correspondence between hazard and cumulative incidence. Fine-Gray model directly models the covariate effect on CIF and it reports subdistribution hazard ratio (SHR). However, SHR only provide information on the ordering of CIF curves at different levels of covariates, it has no practical interpretation as HR in the absence of competing risks. Fine-Gray model can be fit with crr() function shipped with the cmprsk package. Time-varying covariates are allowed in the crr() function, which is specified by cov2 and tf arguments. Predictions and visualization of CIF for subjects with given covariate values are allowed for crr object. Alternatively, competing risk models can be fit with riskRegression package by employing different link functions between covariates and outcomes. The assumption of proportionality can be checked by testing statistical significance of interaction terms involving failure time. Schoenfeld residuals provide another way to check model assumption.

Accounting for competing risks in randomized controlled trials: a review and recommendations for improvement

In studies with survival or time‐to‐event outcomes, a competing risk is an event whose occurrence precludes the occurrence of the primary event of interest. Specialized statistical methods must be used to analyze survival data in the presence of competing risks. We conducted a review of randomized controlled trials with survival outcomes that were published in high‐impact general medical journals. Of 40 studies that we identified, 31 (77.5%) were potentially susceptible to competing risks. However, in the majority of these studies, the potential presence of competing risks was not accounted for in the statistical analyses that were described. Of the 31 studies potentially susceptible to competing risks, 24 (77.4%) reported the results of a Kaplan–Meier survival analysis, while only five (16.1%) reported using cumulative incidence functions to estimate the incidence of the outcome over time in the presence of competing risks. The former approach will tend to result in an overestimate of the incidence of the outcome over time, while the latter approach will result in unbiased estimation of the incidence of the primary outcome over time. We provide recommendations on the analysis and reporting of randomized controlled trials with survival outcomes in the presence of competing risks. © 2017 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.

Antithrombotic agents intake prior to injury does not affect outcome after a traumatic brain injury in hospitalized elderly patients

BACKGROUND

The purpose of this study is to investigate the effect of risk factors including International Normalized Ratio (INR) as well as the Partial Thromboplastin Time (PTT) scores on several outcomes, including hospital length of stay (LOS) and The Extended Glasgow Outcome Scale (GOSE) following TBI in the elderly population.

METHODS

Data were retrospectively collected on patients (n=982) aged 65 and above who were admitted post TBI to the McGill University Health Centre-Montreal General Hospital from 2000 to 2011. Age, Injury Severity Score (ISS), Glasgow Coma Scale score (GCS), type of trauma (isolated TBI vs polytrauma including TBI), initial CT scan results according to the Marshall Classification and the INR and PTT scores and prescriptions of antiplatelet or anticoagulant agents (AP/AC) were collected.

RESULTS

Results also indicated that age, ISS and GSC score have an effect on the GOSE score. We also found that taking AC/AP has an effect on GOSE outcome, but that this effects depends on PTT, with lower odds of a worse outcome for those taking AC/AP agents as the PTT value goes up. However, this effect only becomes significant as the PTT value reaches 60 and above.

CONCLUSION

Age and injury severity rather than antithrombotic agent intake are associated with adverse acute outcome such as GOSE in hospitalized elderly TBI patients.

The number of primary events per variable affects estimation of the subdistribution hazard competing risks model

OBJECTIVES

To examine the effect of the number of events per variable (EPV) on the accuracy of estimated regression coefficients, standard errors, empirical coverage rates of estimated confidence intervals, and empirical estimates of statistical power when using the Fine-Gray subdistribution hazard regression model to assess the effect of covariates on the incidence of events that occur over time in the presence of competing risks.

STUDY DESIGN AND SETTING

Monte Carlo simulations were used. We considered two different definitions of the number of EPV. One included events of any type that occurred (both primary events and competing events), whereas the other included only the number of primary events that occurred.

RESULTS

The definition of EPV that included only the number of primary events was preferable to the alternative definition, as the number of competing events had minimal impact on estimation. In general, 40-50 EPV were necessary to ensure accurate estimation of regression coefficients and associated quantities. However, if all of the covariates are continuous or are binary with moderate prevalence, then 10 EPV are sufficient to ensure accurate estimation.

CONCLUSION

Analysts must base the number of EPV on the number of primary events that occurred.

Choosing a Prescription Isodose in Stereotactic Radiosurgery for Brain Metastases: Implications for Local Control

OBJECTIVE

Stereotactic radiosurgery (SRS) achieves excellent local control (LC) with limited toxicity for most brain metastases. SRS dose prescription variables influence LC; therefore, we evaluated the impact of prescription isodose line (IDL) on LC after SRS.

METHODS

A retrospective analysis of patients with brain metastases treated on a Gamma Knife platform from 2004 to 2014 was conducted. Clinical, toxicity, radiographic, and dosimetric data were collected. Cox proportional hazards regression was used to determine progression-free survival (PFS) and competing risks analysis was used to determine predictive factors for LC.

RESULTS

A total of 134 patients with 374 brain metastases were identified with a median survival of 8.7 months (range, 0.2-64.8). The median tumor maximum dimension was 8 mm (range, 2-62 mm), median margin dose was 20 Gy (range, 5-24 Gy), and 12-month LC rate was 88.7%. On multivariate analysis, PFS improved with increasing IDL (P = 0.003) and decreased with non-non-small-cell lung cancer histology (P = 0.001). Margin dose, tumor size, conformality, and previous whole-brain irradiation failed to independently affect PFS. When adjusting for death as a competing risk, the cumulative likelihood of LC improved with higher IDL (P = 0.04). The rate of SRS-induced radiographic and clinical toxicity was low (16.6% and 1.5%, respectively), and neither was affected by IDL.

CONCLUSIONS

Our results confirm that SRS for brain metastases results in favorable LC, particularly for patients with smaller tumors. We noted that dose delivery to a higher prescription IDL is associated with small but measurable improvements in LC. This finding could be related to higher dose just beyond the radiographically apparent tumor.

Competing risk of death and end-stage renal disease in incident chronic kidney disease (stages 3 to 5): the EPIRAN community-based study

BACKGROUND

Although chronic kidney disease (CKD) affects a growing number of people, epidemiologic data on incident CKD in the general population are scarce. Screening strategies to increase early CKD detection have been developed.

METHODS

From a community-based sample of 4,409 individuals residing in a well-defined geographical area, we determined the number of patients having a first serum creatinine value ≥1.7 mg/dL and present for at least 3 months that allowed us to calculate an annual incidence rate of CKD (stages 3 to 5). CKD (stages 3 to 5) was defined by estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2. We also described the primary care, outcomes and risk factors associated with outcomes using competing risks analyses for these CKD patients.

RESULTS

A total of 631 incident CKD patients (stages 3 to 5) were followed-up until the occurrence of death and dialysis initiation for more than 3 years. The annual incidence rate of CKD (stages 3 to 5) was estimated at 977.7 per million inhabitants. Analyses were performed on 514 patients with available medical data. During the study, 155 patients (30.2 %) were referred to a nephrologist, 193 (37.5 %) died and 58 (11.3 %) reached end-stage renal disease and initiated dialysis. A total of 139 patients (27.6 %) had a fast decline of their renal function, 92 (18.3 %) a moderate decline and the 272 remaining patients had a physiological decline (21.1 %) or a small improvement of their renal function (33.0 %). Predictors of death found in both Cox and Fine-Gray multivariable regression models included age at diagnosis, anemia, active neoplasia and chronic heart failure, but not a low glomerular filtration rate (GFR). Age at diagnosis, anemia and a low GFR were independently associated with dialysis initiation in Cox model, but anemia was not found to be a risk factor for dialysis initiation in Fine-Gray model.

CONCLUSIONS

This large cohort study provided useful epidemiological data on incident CKD (stages 3 to 5) and stressed the need to improve the hands-on implementation of clinical practice guidelines for the evaluation and the management of CKD in primary care.

Risk of Cause-Specific Death in Individuals With Diabetes: A Competing Risks Analysis

OBJECTIVE

Diabetes is a common cause of shortened life expectancy. We aimed to assess the association between diabetes and cause-specific death.

RESEARCH DESIGN AND METHODS

We used the pooled analysis of individual data from 12 Spanish population cohorts with 10-year follow-up. Participants had no previous history of cardiovascular diseases and were 35-79 years old. Diabetes status was self-reported or defined as glycemia >125 mg/dL at baseline. Vital status and causes of death were ascertained by medical records review and linkage with the official death registry. The hazard ratios and cumulative mortality function were assessed with two approaches, with and without competing risks: proportional subdistribution hazard (PSH) and cause-specific hazard (CSH), respectively. Multivariate analyses were fitted for cardiovascular, cancer, and noncardiovascular noncancer deaths.

RESULTS

We included 55,292 individuals (15.6% with diabetes and overall mortality of 9.1%). The adjusted hazard ratios showed that diabetes increased mortality risk: 1) cardiovascular death, CSH = 2.03 (95% CI 1.63-2.52) and PSH = 1.99 (1.60-2.49) in men; and CSH = 2.28 (1.75-2.97) and PSH = 2.23 (1.70-2.91) in women; 2) cancer death, CSH = 1.37 (1.13-1.67) and PSH = 1.35 (1.10-1.65) in men; and CSH = 1.68 (1.29-2.20) and PSH = 1.66 (1.25-2.19) in women; and 3) noncardiovascular noncancer death, CSH = 1.53 (1.23-1.91) and PSH = 1.50 (1.20-1.89) in men; and CSH = 1.89 (1.43-2.48) and PSH = 1.84 (1.39-2.45) in women. In all instances, the cumulative mortality function was significantly higher in individuals with diabetes.

CONCLUSIONS

Diabetes is associated with premature death from cardiovascular disease, cancer, and noncardiovascular noncancer causes. The use of CSH and PSH provides a comprehensive view of mortality dynamics in a population with diabetes.

Competing risks survival of older patients with metastatic cutaneous melanoma: a SEER population-based study

Mortality from metastatic cutaneous melanoma is substantially heterogeneous as reflected in three distant metastatic (M1) subtypes with metastasis to skin, subcutaneous tissue, or distant lymph nodes (M1a), conferring nearly half the risk of death compared with distant visceral metastasis (M1c). It remains unknown whether older patients experience the survival benefit from the M1a subtype given a higher overall mortality risk. Surveillance, Epidemiology, and End Result data were retrieved from 1878 metastatic melanoma patients, from 2005 to 2009, with follow-up through 2011. Hazard ratios (HRs) for 2-year overall survival were estimated for M1 subtypes among older (≥65) and younger (<65) patients. Proportional subdistribution hazard ratios (SHRs) were calculated for melanoma-specific and competing risk mortality. For both older and younger patients, worse overall survival was observed for the M1c compared to the M1a subtype [HR: 2.65, 95% confidence interval (CI): 2.02-3.49; and, SHR: 3.36, 95% CI: 2.56-4.41; respectively]. For competing mortality, older compared to younger patients had increased risk in the M1a and M1b subtypes (SHR: 6.07, 95% CI: 1.94-19.0, and SHR: 2.34, 95% CI: 1.08-5.05, respectively). Conversely, when examining melanoma-specific mortality, older patients had decreased risk in M1a and M1b subtypes (SHR: 0.28, 95% CI: 0.14-0.53, and SHR: 0.60, 95% CI: 0.38-0.94, respectively) compared to those under 65 years. The persistent prognostic advantage of M1a among older patients should be considered when calculating the risk-benefit ratio for treatment. Prior reports of a protective effect of older age on melanoma-specific mortality, when based on traditional competing risks analyses, might be explained as an artifact of increased competing mortality risk.

Effect of alcohol consumption on all-cause and liver-related mortality among HIV-infected individuals

OBJECTIVES

The aim of the study was to examine the association between levels of past and current alcohol consumption and all-cause and liver-related mortality among people living with HIV (PLWH).

METHODS

A prospective cohort study of 1855 PLWH in Baltimore, MD was carried out from 2000 to 2013. We ascertained alcohol use by (1) self-report (SR) through a computer-assisted self interview, and (2) medical record abstraction of provider-documented (PD) alcohol use. SR alcohol consumption was categorized as heavy (men: > 4 drinks/day or > 14 drinks/week; women: > 3 drinks/day or > 7 drinks/week), moderate (any alcohol consumption less than heavy), and none. We calculated the cumulative incidence of liver-related mortality and fitted adjusted cause-specific regression models to account for competing risks.

RESULTS

All-cause and liver-related mortality rates (MRs) were 43.0 and 7.2 per 1000 person-years (PY), respectively. All-cause mortality was highest among SR nondrinkers with PD recent (< 6 months) heavy drinking (MR = 85.4 deaths/1000 PY) and lowest among SR moderate drinkers with no PD history of heavy drinking (MR = 23.0 deaths/1000 PY). Compared with SR moderate drinkers with no PD history of heavy drinking, SR nondrinkers and moderate drinkers with PD recent heavy drinking had higher liver-related mortality [hazard ratio (HR) = 7.28 and 3.52, respectively]. However, SR nondrinkers and moderate drinkers with a PD drinking history of > 6 months ago showed similar rates of liver-related mortality (HR = 1.06 and 2.00, respectively).

CONCLUSIONS

Any heavy alcohol consumption was associated with all-cause mortality among HIV-infected individuals, while only recent heavy consumption was associated with liver-related mortality. Because mortality risk among nondrinkers varies substantially by drinking history, current consumption alone is insufficient to assess risk.

CHK1 as a therapeutic target to bypass chemoresistance in AML

The nucleoside analog cytarabine, an inhibitor of DNA replication fork progression that results in DNA damage, is currently used in the treatment of acute myeloid leukemia (AML). We explored the prognostic value of the expression of 72 genes involved in various aspects of DNA replication in a set of 198 AML patients treated by cytarabine-based chemotherapy. We unveiled that high expression of the DNA replication checkpoint gene CHEK1 is a prognostic marker associated with shorter overall, event-free, and relapse-free survivals and determined that the expression of CHEK1 can predict more frequent and earlier postremission relapse. CHEK1 encodes checkpoint kinase 1 (CHK1), which is activated by the kinase ATR when DNA replication is impaired by DNA damage. High abundance of CHK1 in AML patient cells correlated with higher clonogenic ability and more efficient DNA replication fork progression upon cytarabine treatment. Exposing the patient cells with the high abundance of CHK1 to SCH900776, an inhibitor of the kinase activity of CHK1, reduced clonogenic ability and progression of DNA replication in the presence of cytarabine. These results indicated that some AML cells rely on an efficient CHK1-mediated replication stress response for viability and that therapeutic strategies that inhibit CHK1 could extend current cytarabine-based treatments and overcome drug resistance. Furthermore, monitoring CHEK1 expression could be used both as a predictor of outcome and as a marker to select AML patients for CHK1 inhibitor treatments.

Methodologic Issues When Estimating Risks in Pharmacoepidemiology

Risk is an important parameter to describe the occurrence of health outcomes over time. However, many outcomes of interest in healthcare settings, such as disease incidence, treatment initiation, and cause-specific mortality, may be precluded from occurring by other events, often referred to as competing events. Here, we review straightforward approaches to estimate risk in the presence of competing events. We illustrate the application of these methods using timely examples in pharmacoepidemiologic research and compare results to those obtained using analytic simplifications commonly used to handle competing events. These examples demonstrate how the analytic methods used to account for competing events affect the interpretation of results from pharmacoepidemiologic studies.

Cardiovascular Disease Mortality Among Breast Cancer Survivors

BACKGROUND

Cardiovascular disease (CVD) is of increasing concern among breast cancer survivors. However, the burden of this comorbidity in this group relative to the general population, and its temporal pattern, remains unknown.

METHODS

We compared deaths due to CVD in a population-based sample of 1,413 women with incident breast cancer diagnosed in 1996-1997, and 1,411 age-matched women without breast cancer. Date and cause of death through December 31, 2009 were assessed through the national death index and covariate data was gathered through structured interviews and medical record abstraction. Hazard ratios (HR) and 95% confidence intervals were calculated using Cox regression for overall mortality (HR) and CVD-specific death (cause-specific HR). Subdistribution HRs for CVD death were estimated from the Fine-Gray model.

RESULTS

Risk of death was greater among breast cancer survivors compared with women without breast cancer (HR: 1.8 [1.5, 2.1]). An increase in CVD-related death among breast cancer survivors was evident only 7 years after diagnosis (years 0-7, cause-specific HR: 0.80 [0.53, 1.2], subdistribution HR: 0.59 [0.40, 0.87]); years 7+, cause-specific HR: 1.8 [1.3, 2.5], subdistribution HR: 1.9 [1.4, 2.7]; P interaction: 0.001). An increase in CVD-related mortality was observed among breast cancer survivors receiving chemotherapy.

CONCLUSIONS

Breast cancer survivors are at greater risk for CVD-related mortality compared with women without breast cancer and this increase in risk is manifested approximately 7 years after diagnosis. Efforts should be made to identify risk factors and interventions that can be employed during this brief window to reduce the excess burden of CVD in this vulnerable population.

Multiple time scales in modeling the incidence of infections acquired in intensive care units

Background

When patients are admitted to an intensive care unit (ICU) their risk of getting an infection will be highly depend on the length of stay at-risk in the ICU. In addition, risk of infection is likely to vary over calendar time as a result of fluctuations in the prevalence of the pathogen on the ward. Hence risk of infection is expected to depend on two time scales (time in ICU and calendar time) as well as competing events (discharge or death) and their spatial location. The purpose of this paper is to develop and apply appropriate statistical models for the risk of ICU-acquired infection accounting for multiple time scales, competing risks and the spatial clustering of the data.

Methods

A multi-center data base from a Spanish surveillance network was used to study the occurrence of an infection due to Methicillin-resistant Staphylococcus aureus (MRSA). The analysis included 84,843 patient admissions between January 2006 and December 2011 from 81 ICUs. Stratified Cox models were used to study multiple time scales while accounting for spatial clustering of the data (patients within ICUs) and for death or discharge as competing events for MRSA infection.

Results

Both time scales, time in ICU and calendar time, are highly associated with the MRSA hazard rate and cumulative risk. When using only one basic time scale, the interpretation and magnitude of several patient-individual risk factors differed. Risk factors concerning the severity of illness were more pronounced when using only calendar time. These differences disappeared when using both time scales simultaneously.

Conclusions

The time-dependent dynamics of infections is complex and should be studied with models allowing for multiple time scales. For patient individual risk-factors we recommend stratified Cox regression models for competing events with ICU time as the basic time scale and calendar time as a covariate. The inclusion of calendar time and stratification by ICU allow to indirectly account for ICU-level effects such as local outbreaks or prevention interventions.

Electronic supplementary material

The online version of this article (doi:10.1186/s12874-016-0199-y) contains supplementary material, which is available to authorized users.

Quality of Care and Ischemic Stroke Risk After Hospitalization for Transient Ischemic Attack: Findings From Get With The Guidelines-Stroke

BACKGROUND

Patients with transient ischemic attack (TIA) are at increased risk for ischemic stroke. We derived a prediction rule for 1-year ischemic stroke risk post-TIA, examining estimated risk, receipt of inpatient quality of care measures for TIA, and the presence or absence of stroke at 1 year post discharge.

METHODS AND RESULTS

We linked 67 892 TIA Get With The Guidelines-Stroke patients >65 years (2003-2008) to Medicare inpatient claims to obtain longitudinal outcomes. Using Cox proportional hazards modeling in a split sample, we identified baseline demographics and clinical characteristics associated with ischemic stroke admission during the year post-TIA, and developed a Get With The Guidelines Ischemic Stroke after TIA Risk Score; performance was examined in the validation sample. Quality of care was estimated by a global defect-free care measure, and individual performance measures within estimated risk score quintiles. The overall hospital admission rate for ischemic stroke during the year post-TIA was 5.7%. Patients with ischemic stroke were more likely to be older, black, and have higher rates of smoking, previous stroke, diabetes mellitus, previous myocardial infarction, heart failure, and atrial fibrillation. The Risk Score showed moderate discriminative performance (c-statistic=0.606); highest quintile patients were less likely to receive statins, smoking cessation counseling, and defect-free care. Although not associated with 1-year ischemic stroke, DCF was associated with a significantly lower risk of all-cause mortality.

CONCLUSIONS

TIA patients with high estimated ischemic stroke risk are less likely to receive defect-free care than low-risk patients. Standardized risk assessment and delivery of optimal inpatient care are needed to reduce this risk-treatment mismatch.

Competing risks analysis of cause-specific mortality in patients with oral squamous cell carcinoma

BACKGROUND

Survival studies on head and neck cancers are frequently reported with inadequate account for competing causes of death. Realistic descriptions and predictions of postdiagnosis mortality should be based on proper competing risks methodology.

METHODS

Prognosis of patients with oral squamous cell carcinoma (OSCC) in terms of mortality from OSCC and from other causes, respectively, was analyzed according to recent methodological recommendations using cumulative incidence functions and models for cause-specific hazards and subdistribution hazards in 306 patients treated in a tertiary care center in Northern Finland.

RESULTS

More coherent and informative descriptions and predictions of mortality by cause were obtained with state-of-the-art statistical methods for competing risks than using the prevalent but questionable practice to graph "disease-specific survival."

CONCLUSION

From the patients' perspective, proper competing risks analysis offers more relevant prognostic scenarios than naïve analyses of "disease-specific survival"; therefore, it should be used in prognostic studies of head and neck cancers. © 2016 Wiley Periodicals, Head Neck 39: 56-62, 2017.

Stroke After Radiation Therapy for Head and Neck Cancer: What Is the Risk?

PURPOSE

A retrospective population-based cohort study was conducted to determine the risk of ischemic stroke with respect to time, associated with curative radiation therapy in head and neck squamous cell carcinomas (HNSCC).

METHODS AND MATERIALS

On the basis of data from the Ontario Cancer Registry and regional cancer treatment centers, 14,069 patients were identified with diagnoses of squamous cell carcinoma of the oral cavity, larynx, and pharynx who were treated for cure between 1990 and 2010. Hazards of stroke and time to stroke were examined, accounting for the competing risk of death. Stroke risk factors identified through diagnostic and procedural administrative codes were adjusted for in the comparison between treatment regimens, which included surgery alone versus radiation therapy alone and surgery alone versus any exposure to radiation therapy.

RESULTS

Overall, 6% of patients experienced an ischemic stroke after treatment, with 5% experiencing a stroke after surgery, 8% after radiation therapy alone, and 6% after any exposure to radiation therapy. The cause-specific hazard ratios of ischemic stroke after radiation therapy alone and after any exposure to radiation therapy compared with surgery were 1.70 (95% confidence interval [CI]: 1.41-2.05) and 1.46 (95% CI: 1.23-1.73), respectively, after adjustment for stroke risk factors, patient factors, and disease-related factors.

CONCLUSIONS

Radiation therapy was associated with an increased risk of ischemic stroke compared with surgery alone: for both radiation therapy alone and after all treatment modalities that included any radiation treatment were combined. Because of a shift toward a younger HNSCC patient population, our results speak to the need for adequate follow-up and survivorship care among patients who have been treated with radiation therapy. Advances in treatment that minimize chronic morbidity also require further evaluation.

Fetal Growth and the Risk of Spontaneous Preterm Birth in a Prospective Cohort Study of Nulliparous Women

Previous studies have suggested an association between fetal growth restriction and the risk of spontaneous preterm birth (sPTB). However, addressing this association is methodologically challenging. We conducted a prospective cohort study of nulliparous women with a singleton pregnancy in Cambridge, United Kingdom (2008-2012). Ultrasonic fetal biometry was performed at 20 weeks of gestation as per routine clinical care. Participants also had blinded research ultrasonography performed at approximately 28 weeks. Biometric measurements were expressed as gestational-age-adjusted z scores. Fetal growth velocity was quantified by change in z score between 20 weeks and 28 weeks. Risk of sPTB, defined as delivery at ≥28 weeks and <37 weeks associated with labor in the absence of induction, was analyzed using cause-specific Cox regression. Of 3,892 women, 98 (2.5%) had sPTB. When compared with the other decile groups, the lowest decile of growth velocity of the fetal femur between 20 and 28 weeks was associated with increased risk of sPTB (hazard ratio = 2.37, 95% confidence interval: 1.43, 3.93; P < 0.001). Adjustment for maternal characteristics had no material effect (hazard ratio = 2.50, 95% confidence interval: 1.50, 4.14; P < 0.001). There were no significant associations between other fetal measurements and risk of sPTB. To conclude, slow growth velocity of the fetal femur is associated with an increased risk of sPTB.

Colorectal cancer mortality reduction is associated with having at least 1 colonoscopy within the previous 10 years among a population-wide cohort of screening age

BACKGROUND AND AIMS

Colonoscopy has been demonstrated to be effective in colorectal cancer (CRC) mortality reduction, although current screening guidelines have yet to be evaluated. We assessed the protective benefit of colonoscopy within the previous 10 years and whether this effect is maintained with age.

METHODS

We used administrative data to compare risk of CRC death (CCD) across colonoscopy utilization among a population-wide cohort comprising individuals aged 60 to 80 years (N = 1,509,423). Baseline and time-dependent colonoscopy exposure models were assessed in the context of competing "other causes of death" (OCDs). Cumulative incidence of CCD and OCD across colonoscopy exposure, over follow-up, was estimated. Relative hazards were computed by age strata (60-69 years, 70-74 years, 75+ years) and proximal and distal cancer subsites.

RESULTS

At least 1 colonoscopy during 10 years before baseline was estimated to provide a 51% reduced hazard of CCD (hazard ratio [HR] 0.49; 95% confidence interval [CI], 0.45-0.54) over the following 8 years. When colonoscopy was modeled as a time-dependent covariate, the risk of CCD was further diminished (multivariable-adjusted HR 0.36; 95% CI, 0.33-0.38). Stratified analyses suggested moderately attenuated CCD risk reduction among the oldest age group; however, consideration of OCDs suggest that this is related to competing risks. CCD risk reduction related to colonoscopy was lower for proximal cancers.

CONCLUSIONS

Colonoscopy within the previous 10 years provides substantial protective benefit for average-risk individuals over 60 years. CCD risk reduction may be maintained well beyond 74 years, a common upper age limit recommended by screening guidelines.

Introduction to the Analysis of Survival Data in the Presence of Competing Risks

Supplemental Digital Content is available in the text.

Competing risks occur frequently in the analysis of survival data. A competing risk is an event whose occurrence precludes the occurrence of the primary event of interest. In a study examining time to death attributable to cardiovascular causes, death attributable to noncardiovascular causes is a competing risk. When estimating the crude incidence of outcomes, analysts should use the cumulative incidence function, rather than the complement of the Kaplan-Meier survival function. The use of the Kaplan-Meier survival function results in estimates of incidence that are biased upward, regardless of whether the competing events are independent of one another. When fitting regression models in the presence of competing risks, researchers can choose from 2 different families of models: modeling the effect of covariates on the cause-specific hazard of the outcome or modeling the effect of covariates on the cumulative incidence function. The former allows one to estimate the effect of the covariates on the rate of occurrence of the outcome in those subjects who are currently event free. The latter allows one to estimate the effect of covariates on the absolute risk of the outcome over time. The former family of models may be better suited for addressing etiologic questions, whereas the latter model may be better suited for estimating a patient’s clinical prognosis. We illustrate the application of these methods by examining cause-specific mortality in patients hospitalized with heart failure. Statistical software code in both R and SAS is provided.

Safety data from randomized controlled trials: applying models for recurrent events

Simple descriptive listings and inference statistics based on 2×2 tables are still the most common way of summarizing and reporting adverse events data from randomized controlled trials, although these methods do not account for differences in observation times between treatment groups. Using standard methods from survival analysis such as the Cox model or Kaplan-Meier estimates would overcome this problem but limit the analysis to the first safety-related event of each subject. As an alternative, we discuss two models for recurrent events data-the Andersen-Gill and Prentice-Williams-Peterson model-regarding their applicability to safety data from randomized controlled trials. We argue that these models can be used to estimate two different quantities: a direct treatment effect on the risk of an event (Prentice-Williams-Peterson) and a total treatment effect as sum of the direct effect and the treatment's indirect effect via the event history (Anderson-Gill). Using simulated data, we illustrate the difference between these treatment effects and analyze the performance of both models in different scenarios. Because both models are limited to the analysis of cause-specific hazards if competing risks are present, we suggest to incorporate estimates of the mean frequency of events in the analysis to additionally allow the comparison of treatment effects on absolute event probabilities. We demonstrate the application of both models and the mean frequency function to safety endpoints with an illustrative analysis of data from a randomized phase-III study. Copyright © 2016 John Wiley & Sons, Ltd.

Developing points-based risk-scoring systems in the presence of competing risks

Predicting the occurrence of an adverse event over time is an important issue in clinical medicine. Clinical prediction models and associated points‐based risk‐scoring systems are popular statistical methods for summarizing the relationship between a multivariable set of patient risk factors and the risk of the occurrence of an adverse event. Points‐based risk‐scoring systems are popular amongst physicians as they permit a rapid assessment of patient risk without the use of computers or other electronic devices. The use of such points‐based risk‐scoring systems facilitates evidence‐based clinical decision making. There is a growing interest in cause‐specific mortality and in non‐fatal outcomes. However, when considering these types of outcomes, one must account for competing risks whose occurrence precludes the occurrence of the event of interest. We describe how points‐based risk‐scoring systems can be developed in the presence of competing events. We illustrate the application of these methods by developing risk‐scoring systems for predicting cardiovascular mortality in patients hospitalized with acute myocardial infarction. Code in the R statistical programming language is provided for the implementation of the described methods. © 2016 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.

Trends and outcomes of patients with adult congenital heart disease and pulmonary hypertension listed for orthotopic heart transplantation in the United States

BACKGROUND

Heart transplantation is increasing in patients with adult congenital heart disease (ACHD). In this population, the association of pulmonary hypertension (PH) with post-transplant outcomes is not well-defined.

METHODS

Using data from the United Network for Organ Sharing database (1987 to 2014), we identified ACHD patients listed for heart transplantation, and examined survival between those with and without PH (pre-transplant PH defined as transpulmonary pressure gradient >12 mm Hg).

RESULTS

Among 983 ACHD patients, 216 (22%) had PH. At time of listing, PH patients had a transpulmonary pressure gradient of 17.0 mm Hg vs 6.0 mm Hg (p < 0.01) in the no-PH group. Although left ventricular assist device (LVAD) use was infrequent, 3.1% of PH patients were treated with an LVAD versus 6.8% of the no-PH patients. Days from listing to transplant, days from listing to death on the waitlist and length of post-transplant hospitalization were not significantly different between the PH and no-PH groups. However, PH was associated with higher waitlist mortality (HR 1.73, CI 1.25 to 2.41). Pre-transplant PH was not associated with post-transplant mortality at 30 days (HR 0.51, CI 0.23 to 1.13), 1 year (HR 0.68, 95% CI 0.40 to 1.18) or 5 years (HR 0.84, 95% CI 0.55 to 1.29).

CONCLUSIONS

PH is common among ACHD patients listed for transplant and is associated with increased waitlist mortality. Conversely, PH was not associated with worse survival after transplant. Bridge-to-transplant LVAD therapy was uncommon in this ACHD population.