Estimating restricted mean survival time and expected life-years lost in the presence of competing risks within flexible parametric survival models

Modeling Recovery Housing Retention and Program Outcomes by Justice Involvement among Residents in Virginia, USA: An Observational Study

Living in recovery housing can improve addiction recovery and desistance outcomes. This study examined whether retention in recovery housing and types of discharge outcomes (completed, "neutral," and "negative" outcomes) differed for clients with recent criminal legal system (CLS) involvement. Using data from 101 recovery residences certified by the Virginia Association of Recovery Residences based on 1,978 individuals completing the REC-CAP assessment, competing risk analyses (cumulative incidence function, restricted mean survival time, and restricted mean time lost) followed by the marginalization of effects were implemented to examine program outcomes at final discharge. Residents with recent CLS involvement were more likely to be discharged for positive reasons (successful completion of their goals) and premature/negative reasons (e.g., disciplinary releases) than for neutral reasons. Findings indicate that retention for 6-18 months is essential to establish and maintain positive discharge outcomes, and interventions should be developed to enhance retention in recovery residents with recent justice involvement.

Age of Diabetes Diagnosis and Lifetime Risk of Dementia: The Atherosclerosis Risk in Communities (ARIC) Study

OBJECTIVE

The impact of age of diabetes diagnosis on dementia risk across the life course is poorly characterized. We estimated the lifetime risk of dementia by age of diabetes diagnosis.

RESEARCH DESIGN AND METHODS

We included 13,087 participants from the Atherosclerosis Risk in Communities Study who were free from dementia at age 60 years. We categorized participants as having middle age-onset diabetes (diagnosis <60 years), older-onset diabetes (diagnosis 60-69 years), or no diabetes. Incident dementia was ascertained via adjudication and active surveillance. We used the cumulative incidence function estimator to characterize the lifetime risk of dementia by age of diabetes diagnosis while accounting for the competing risk of mortality. We used restricted mean survival time to calculate years lived without and with dementia.

RESULTS

Among 13,087 participants, there were 2,982 individuals with dementia and 4,662 deaths without dementia during a median follow-up of 24.1 (percentile 25-percentile 75, 17.4-28.3) years. Individuals with middle age-onset diabetes had a significantly higher lifetime risk of dementia than those with older-onset diabetes (36.0% vs. 31.0%). Compared with those with no diabetes, participants with middle age-onset diabetes also had a higher cumulative incidence of dementia by age 80 years (16.1% vs. 9.4%) but a lower lifetime risk (36.0% vs. 45.6%) due to shorter survival. Individuals with middle age-onset diabetes developed dementia 4 and 1 years earlier than those without diabetes and those with older-onset diabetes, respectively.

CONCLUSIONS

Preventing or delaying diabetes may be an important approach for reducing dementia risk throughout the life course.

Assessing treatment effects with adjusted restricted mean time lost in observational competing risks data

BACKGROUND

According to long-term follow-up data of malignant tumor patients, assessing treatment effects requires careful consideration of competing risks. The commonly used cause-specific hazard ratio (CHR) and sub-distribution hazard ratio (SHR) are relative indicators and may present challenges in terms of proportional hazards assumption and clinical interpretation. Recently, the restricted mean time lost (RMTL) has been recommended as a supplementary measure for better clinical interpretation. Moreover, for observational study data in epidemiological and clinical settings, due to the influence of confounding factors, covariate adjustment is crucial for determining the causal effect of treatment.

METHODS

We construct an RMTL estimator after adjusting for covariates based on the inverse probability weighting method, and derive the variance to construct interval estimates based on the large sample properties. We use simulation studies to study the statistical performance of this estimator in various scenarios. In addition, we further consider the changes in treatment effects over time, constructing a dynamic RMTL difference curve and corresponding confidence bands for the curve.

RESULTS

The simulation results demonstrate that the adjusted RMTL estimator exhibits smaller biases compared with unadjusted RMTL and provides robust interval estimates in all scenarios. This method was applied to a real-world cervical cancer patient data, revealing improvements in the prognosis of patients with small cell carcinoma of the cervix. The results showed that the protective effect of surgery was significant only in the first 20 months, but the long-term effect was not obvious. Radiotherapy significantly improved patient outcomes during the follow-up period from 17 to 57 months, while radiotherapy combined with chemotherapy significantly improved patient outcomes throughout the entire period.

CONCLUSIONS

We propose the approach that is easy to interpret and implement for assessing treatment effects in observational competing risk data.

More Frequent On-Site Dialysis May Hasten Return to Home for Nursing Home Patients with End-Stage Kidney Disease

Key Points

Prior efficacy study—established that more frequent dialysis achieved better outcomes than CONVENTIONAL dialysis in outpatients. We undertook an effectiveness observational on-site nursing home study (N =195) comparing on-site more frequent dialysis with CONVENTIONAL dialysis. More frequent dialysis patients, despite being sicker at baseline, returned home faster than CONVENTIONALLY dialyzed patients without worsened death or hospitalization.

Background

A direct outcome comparison between skilled nursing facility (SNF) patients receiving on-site more frequent dialysis (MFD) targeting 14 hours of treatment over five sessions weekly compared with on-site CONVENTIONAL dialysis for death, hospitalization, and speed of returning home has not been reported.

Methods

From January 1, 2022, to July 1, 2023, in a retrospective prospective observational design, using an intention-to-treat and competing risk strategy, all new admissions for an on-site SNF dialysis service done to nursing homes with on-site MFD were compared with admissions to nursing homes providing on-site CONVENTIONAL dialysis for the outcome goal of 90-day cumulative incidence of discharge to home, while monitoring safety issues represented by the competing risks of hospitalization and death.

Results

In total, 10,246 MFD dialytic episodes and 3451 CONVENTIONAL dialytic episodes were studied in 195 nursing homes in 12 states. At baseline, the MFD population was consistently sicker than CONVENTIONAL dialysis population with a first systolic BP of <100 mm Hg in 13% versus 7.6% (P < 0.001), lower mean hemoglobin (9.3 versus 10.4 g/dl; P < 0.001), lower iron saturation (25.7% versus 26.6%; P = 0.02), higher Charlson score (3.5 versus 3.0; P < 0.001), higher mean age (67.6 versus 66.7; P < 0.001), more complicated diabetes (31% versus 24%; P < 0.001), cerebrovascular disease (12.6% versus 6.8%; P <0.001), and congestive heart failure (24% versus 18%). At 42 days, discharge to home was 25% greater in the MFD than CONVENTIONAL dialysis group (17.5% versus 14%) without worsened hospitalization or death.

Conclusions

Despite a handicap of sicker patients at baseline, real-world application of MFD appears to hasten return to home from SNFs compared with CONVENTIONAL dialysis. The findings suggest that MFD allows for SNF acceptance of sicker patients, presumably permitting earlier discharge from hospital, without safety compromise as measured by death or rehospitalization, benefitting hospitals, patients, and payers.

Early Prostate Cancer Deaths Among Men With Higher vs Lower Genetic Risk

Importance

Prostate cancer, a leading cause of cancer death among men, urgently requires new prevention strategies, which may involve targeting men with an underlying genetic susceptibility.

Objective

To explore differences in risk of early prostate cancer death among men with higher vs lower genetic risk to inform prevention efforts.

Design, Setting, and Participants

This cohort study used a combined analysis of genotyped men without prostate cancer at inclusion and with lifestyle data in 2 prospective cohort studies in Sweden and the US, the Malmö Diet and Cancer Study (MDCS) and the Health Professionals Follow-Up Study (HPFS), followed up from 1991 to 2019. Data were analyzed between April 2023 and April 2024.

Exposures

Men were categorized according to modifiable lifestyle behaviors and genetic risk. A polygenic risk score above the median or a family history of cancer defined men at higher genetic risk (67% of the study population); the remaining men were categorized as being at lower genetic risk.

Main Outcomes and Measures

Prostate cancer death analyzed using time-to-event analysis estimating hazard ratios (HR), absolute risks, and preventable deaths by age.

Results

Among the 19 607 men included for analysis, the median (IQR) age at inclusion was 59.0 (53.0-64.7) years (MDCS) and 65.1 (58.0-71.8) years (HPFS). During follow-up, 107 early (by age 75 years) and 337 late (after age 75 years) prostate cancer deaths were observed. Compared with men at lower genetic risk, men at higher genetic risk had increased rates of both early (HR, 3.26; 95% CI, 1.82-5.84) and late (HR, 2.26; 95% CI, 1.70-3.01) prostate cancer death, and higher lifetime risks of prostate cancer death (3.1% vs 1.3% [MDCS] and 2.3% vs 0.6% [HPFS]). Men at higher genetic risk accounted for 94 of 107 early prostate cancer deaths (88%), of which 36% (95% CI, 12%-60%) were estimated to be preventable through adherence to behaviors associated with a healthy lifestyle (not smoking, healthy weight, high physical activity, and a healthy diet).

Conclusions and Relevance

In this 20-year follow-up study, men with a genetic predisposition accounted for the vast majority of early prostate cancer deaths, of which one-third were estimated to be preventable. This suggests that men at increased genetic risk should be targeted in prostate cancer prevention strategies.

Evaluation of MADIT-II Risk Stratification Score Among Nationwide Registry of Heart Failure Patients With Primary Prevention Implantable Cardiac Defibrillators or Resynchronization Therapy Devices

The current guidelines advocate prophylactic implantable cardioverter-defibrillator (ICD) for all patients with symptomatic heart failure (HF) with low left ventricular ejection fraction. Because many patients will never use their device, a score delineating subgroups with differential ICD benefit is crucial. We aimed to evaluate the MADIT-II-based Risk Stratification Score (MRSS) feasibility to delineate the ICD survival benefit in a nationwide registry of patients with HF with prophylactic ICDs. Accordingly, all Israeli patients with HF with prophylactic ICD/cardiac resynchronization therapy defibrillators were categorized into MRSS-based risk subgroups. The study end points included overall mortality, sustained ventricular arrhythmia (VA), and a competing risk of VA (potential preventable arrhythmic death, where ICD could benefit survival) versus nonarrhythmic death. Potential ICD survival benefit was estimated by the area between these cumulative incidence curves. In 2,177 patients with HF implanted prophylactic device, 189 patients (8.7%) had VA and 316 (14.5%) died during a median follow-up of 2.9 years. The MRSS risk subgroups were significantly associated with overall mortality (p <0.001) and weakly with VA (p = 0.3). The competing risk analysis of VA versus nonarrhythmic death revealed a significantly shorter duration (p <0.001) and smaller magnitude of ICD survival benefit with increased risk subgroups, yielding an estimated 76, 60, 38, and 0 life days gained from prophylactic ICD implant during a 5-year follow-up for the MRSS low-, intermediate-, high-, and very high-risk subgroups, respectively (p for trend <0.05). In conclusion, MRSS use in a nationwide registry of patients with ischemic and nonischemic cardiomyopathy, revealed subgroups with differing ICD survival benefit, suggesting it could help evaluate prophylactic ICD survival benefit.

Analysis of survival data with nonproportional hazards: A comparison of propensity-score-weighted methods

One of the most common ways researchers compare cancer survival outcomes across treatments from observational data is using Cox regression. This model depends on its underlying assumption of proportional hazards, but in some real-world cases, such as when comparing different classes of cancer therapies, substantial violations may occur. In this situation, researchers have several alternative methods to choose from, including Cox models with time-varying hazard ratios; parametric accelerated failure time models; Kaplan-Meier curves; and pseudo-observations. It is unclear which of these models are likely to perform best in practice. To fill this gap in the literature, we perform a neutral comparison study of candidate approaches. We examine clinically meaningful outcome measures that can be computed and directly compared across each method, namely, survival probability at time T, median survival, and restricted mean survival. To adjust for differences between treatment groups, we use inverse probability of treatment weighting based on the propensity score. We conduct simulation studies under a range of scenarios, and determine the biases, coverages, and standard errors of the average treatment effects for each method. We then demonstrate the use of these approaches using two published observational studies of survival after cancer treatment. The first examines chemotherapy in sarcoma, which has a late treatment effect (i.e., similar survival initially, but after 2 years the chemotherapy group shows a benefit). The other study is a comparison of surgical techniques for kidney cancer, where survival differences are attenuated over time.

Communicating and understanding statistical measures when quantifying the between-group difference in competing risks

Competing risks issues are common in clinical trials and epidemiological studies for patients in follow-up who may experience a variety of possible outcomes. Under such competing risks, two hazard-based statistical methods, cause-specific hazard (CSH) and subdistribution hazard (SDH), are frequently used to assess treatment effects among groups. However, the outcomes of the CSH-based and SDH-based methods have a close connection with the proportional hazards (CSH or SDH) assumption and may have an non-intuitive interpretation. Recently, restricted mean time lost (RMTL) has been used as an alternative summary measure for analysing competing risks, due to its clinical interpretability and robustness to the proportional hazards assumption. Considering the above approaches, we summarize the differences between hazard-based and RMTL-based methods from the aspects of practical interpretation, proportional hazards model assumption and the selection of restricted time points, and propose corresponding suggestions for the analysis of between-group differences under competing risks. Moreover, an R package 'cRMTL' and corresponding step-by-step guidance are available to help users for applying these approaches.

Estimates of years of life lost depended on the method used: tutorial and comparative investigation

OBJECTIVES

This review aims to summarize key methods for estimating years of life lost (YLL), highlighting their differences and how they can be implemented in current software, and applies them in a real-world example.

STUDY DESIGN AND SETTING

We investigated the common YLL methods: (1) Years of potential life lost (YPLL); (2) Global Burden of Disease (GBD) approach; (3) Life tables; (4) Poisson regression; and (5) Flexible parametric Royston-Parmar regression. We used data from UK Biobank and multimorbidity as our example.

RESULTS

For the YPLL and GBD method, the analytical procedures allow only to quantify the average YLL within each group (with and without multimorbidity) and, from them, their difference; conversely, for the other methods both the remaining life expectancy within each group and the YLL could be estimated. At 65 years, the YLL in those with vs. without multimorbidity was 1.8, 1.2, and 2.7 years using the life tables approach and the Poisson, and Royston-Parmar regression, respectively; corresponding values were -0.73 and -0.05 years for YPLL and using the GBD approach.

CONCLUSION

While deciding among different methods to estimate YLL, researchers should consider the purpose of the research, the type of available data, and the flexibility of the model.

Estimating causal effects in the presence of competing events using regression standardisation with the Stata command standsurv

BACKGROUND

When interested in a time-to-event outcome, competing events that prevent the occurrence of the event of interest may be present. In the presence of competing events, various estimands have been suggested for defining the causal effect of treatment on the event of interest. Depending on the estimand, the competing events are either accommodated or eliminated, resulting in causal effects with different interpretations. The former approach captures the total effect of treatment on the event of interest while the latter approach captures the direct effect of treatment on the event of interest that is not mediated by the competing event. Separable effects have also been defined for settings where the treatment can be partitioned into two components that affect the event of interest and the competing event through different causal pathways.

METHODS

We outline various causal effects that may be of interest in the presence of competing events, including total, direct and separable effects, and describe how to obtain estimates using regression standardisation with the Stata command standsurv. Regression standardisation is applied by obtaining the average of individual estimates across all individuals in a study population after fitting a survival model.

RESULTS

With standsurv several contrasts of interest can be calculated including differences, ratios and other user-defined functions. Confidence intervals can also be obtained using the delta method. Throughout we use an example analysing a publicly available dataset on prostate cancer to allow the reader to replicate the analysis and further explore the different effects of interest.

CONCLUSIONS

Several causal effects can be defined in the presence of competing events and, under assumptions, estimates of those can be obtained using regression standardisation with the Stata command standsurv. The choice of which causal effect to define should be given careful consideration based on the research question and the audience to which the findings will be communicated.

Doubly-robust estimator of the difference in restricted mean times lost with competing risks data

In the context of competing risks data, the subdistribution hazard ratio has limited clinical interpretability to measure treatment effects. An alternative is the difference in restricted mean times lost (RMTL), which gives the mean time lost to a specific cause of failure between treatment groups. In non-randomized studies, the average causal effect is conventionally used for decision-making about treatment and public health policies. We show how the difference in RMTL can be estimated by contrasting the integrated cumulative incidence functions from a Fine-Gray model. We also show how the difference in RMTL can be estimated by using inverse probability of treatment weighting and contrasts between weighted non-parametric estimators of the area below the cumulative incidence. We use pseudo-observation approaches to estimate both component models and we integrate them into a doubly-robust estimator. We demonstrate that this estimator is consistent when either component is correctly specified. We conduct simulation studies to assess its finite-sample performance and demonstrate its inherited consistency property from its component models. We also examine the performance of this estimator under varying degrees of covariate overlap and under a model misspecification of nonlinearity. We apply the proposed method to assess biomarker-treatment interaction in subpopulations of the POPLAR and OAK randomized controlled trials of second-line therapy for advanced non-small-cell lung cancer.