Immortal time bias in contemporary CDK4/6 inhibitor studies: Evidence from target trial emulation.

Target trial emulation (TTE) is an observational, quasi-experimental research design that emulates a randomized clinical trial (RCT) structure within a large set of observational data; the "target trial" is a hypothetical RCT that would have ideally answered the research question. TTEs can address study objectives that, for ethical or logistic reasons, cannot easily be examined in RCTs. Advantages of TTEs over conventional approaches to observational data are that TTEs can reduce bias, improve the understanding of findings, and facilitate causal inference. This article explains what TTEs are, how TTEs are performed, and how TTEs differ from observational studies, quasi controlled studies, and RCTs. Prevalent user bias and immortal time bias are explained, as is how TTEs are designed to avoid these biases. Strengths and limitations of TTEs are discussed. This article also presents 2 recent studies: one, comprising 3 TTEs that examined scholastic outcomes in children gestationally exposed to benzodiazepines and z-drugs in different periods during pregnancy; and the other, a TTE that examined manic switch as an outcome in bipolar depression patients who received antidepressant treatment. The TTEs found that early, mid, or late pregnancy exposure to benzodiazepines or z-drugs was not associated with impairment in fifth-grade numeracy and literacy performance; and that, in patients with bipolar depression, antidepressant drugs (with or without concurrent mood stabilizers) did not increase the 1-year risk of hypomania, mania, or mixed episodes, nor did they reduce the risk of recurrence of bipolar depression. The TTEs that yielded these results had limitations, and so these findings are suggestive, not definitive. As a general conclusion, TTEs may be viewed as pragmatic, naturalistic, real-world emulations of RCTs, with some advantages over conventional observational studies, but they cannot drive causal inference.

Target trial emulation (TTE) has demonstrated popularity because of its ability to improve the reliability of causal inference from observational data. Nevertheless, knowledge about the current use, potential challenges, and insights of target trials in oncology is scarce. A total of 90 TTE studies in cancer areas were identified through systematic reviews in PubMed and Embase. Among the 54 applications in cancer treatment, registry databases (44.4%) and overall survival (OS, 63.0%) were predominantly used as data sources and primary endpoints, respectively. Approximately 30 (55.6%) of the included TTE cases were associated with immortal time bias, and 21 (38.9%) were associated with prevalent user bias. Among the 21 trials from 13 studies that aimed to calibrate the results from preexisting randomized controlled trials (RCTs), only 42.9% met both statistical agreement and estimate agreement. The availability of fit-for-purpose data sources and uncertainty about result concordance were the main hurdles limiting the quantity and quality of TTE in oncology areas. Promoting regulatory acceptance by initiating special projects could be crucial for the expanded application of real-world data (RWD) using TTE. Potential solutions, such as the integration of electronic medical records at the regional or country level, linkage with insurance claims databases, the modernization of eligibility criteria, the use of OS as the primary endpoint, and other best practices, were recommended for improving the feasibility and quality of oncology TTE.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40779-026-00685-9.

Unsubstantiated concerns over the safety of use of sulphonylureas and insulin for increased risk of diabetes complications (使用磺脲类药物和胰岛素增加糖尿病并发症的担心并无事实根据)

During the COVID-19 pandemic, real-world data and observational studies played an important role in assessing treatment effectiveness. Methodological challenges such as confounding, immortal time bias, and competing risks were observed. Target trial emulation provides a structured framework for evaluating treatment effectiveness using observational data while mitigating these biases. To describe common biases in observational COVID-19 research, introduce the target trial emulation framework, and discuss how these biases can be addressed in this framework. Specifically, we discuss the clone-censor-weight approach and provide real-world study examples demonstrating its application in COVID-19 research. We summarise key principles of target trial emulation and the clone-censor-weight approach using published methodological articles. Additionally, we demonstrate the practical implementation by reviewing three studies that emulated a target trial to evaluate the effects of treatments in patients with COVID-19. These studies were selected without a predefined search strategy. We define and discuss confounding, immortal time bias, and competing risks in studies using observational data. To facilitate the understanding of these biases, we use a hypothetical example evaluating the effects of hydroxychloroquine in hospitalised patients with COVID-19. We provide an overview of the target trial emulation framework and its core elements, explaining how it can mitigate these challenges. To illustrate the clone-censor-weight approach, we describe published examples demonstrating its application during the COVID-19 pandemic. Target trial emulation is an important framework for evaluating treatment effects using observational data, but it requires careful implementation to mitigate methodological biases. Identifying and addressing confounding, immortal time bias, and competing risks during study design and analysis are important in any causal study evaluating treatment effects. This framework can improve the quality of observational studies and complement evidence from clinical trials, particularly when evidence is urgently needed, as during the first waves of the COVID-19 pandemic.

BackgroundImmune checkpoint inhibitors (ICIs) are an important therapeutic pillar in metastatic urothelial carcinoma (mUC). The occurrence of immune-related adverse events (irAEs) appears to be associated with improved outcomes in observational studies. However, these associations are likely affected by immortal time bias and do not represent causal effects. The aim of this study was to assess the effect of irAEs on outcomes while correcting for immortal time bias, using target trial emulation (TTE).MethodsTTE was contrasted to adjusted naïve and time-updated Cox models. We performed a multi-institutional retrospective study involving mUC patients under ICI. The primary objective was to assess the impact of irAEs on progression-free survival (PFS) and overall survival (OS). Secondary endpoints included the influence of irAEs on objective response rates (ORRs) to ICI and the influence of systemic corticosteroids on outcomes.ResultsAmong 335 patients (median age: 69 yrs), 69.6% received ICI in the second line or further lines. During a median follow-up of 21.1 months, 122 (36.4%) patients developed irAEs of any grade (grade ≥ 3: 14.9%). Hazard ratios (HRs) for PFS ranged from 0.37 for naïve adjusted Cox model to 0.88 (95% confidence interval (CI), 0.59–1.30) with time-updated covariates, and from 0.41 to 1.10 (95% CI, 0.69–1.75) for OS. TTE accounting for immortal time bias yielded a HR of 1.02 (95% CI, 0.72–1.44) for PFS, and 0.90 (95% CI, 0.62–1.30) for OS. In contrast to the naïve Cox model (HR = 2.26, 95% CI 1.26–4.05), the presence of irAEs was no longer a predictive factor for improved ORR in time-updated Cox models (HR = 1.27, 95% CI 0.68–2.36) and TTE (HR = 1.43, 95% CI 0.89–2.29). In patients with irAEs, systemic corticosteroids did not negatively impact survival.ConclusionUsing TTE, we were able to show that the occurrence of irAEs is no longer associated with better survival or improved response rates to ICI in mUC patients, in contrast to the naïve analysis. These findings demonstrate that TTE is a suitable formal framework to avoid immortal time bias in studies with time-dependent non-interventional exposures.Graphical abstract

Target trial emulation (TTE) is a powerful framework for addressing causal questions using observational data. By explicitly designing analyses to mimic a hypothetical randomized trial, TTE enables researchers to more precisely define their research questions, leading to more clinically meaningful conclusions. Its forward-looking design also helps limit common biases, such as immortal time bias and selection bias. Understanding TTE principles is essential not only for researchers working with observational data but also for clinicians who aim to critically interpret the growing number of TTE studies, as well as studies addressing causal questions without explicit use of the TTE framework. In this review, using the timing of switch from controlled to assisted ventilation as a key example, we outline the core assumptions underpinning valid causal inference in TTE: consistency, conditional exchangeability, and positivity. We discuss practical challenges in dynamic critical care settings, including defining a meaningful time zero, handling grace periods, and selecting and properly adjusting for confounders. We also discuss caveats, such as TTE’s applicability to non-modifiable interventions, limited applicability for intention-to-treat effects, and the need for high-resolution longitudinal data. Finally, we provide a visual summary linking each trial component to key indicators of high-quality emulation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13054-025-05723-x.

Immortal time bias is a spurious or exaggerated protective association that commonly arises in naive analyses of observational data. It occurs when people receive the intervention because they survive, rather than survive because they received the intervention. Studies in conditions with substantial early mortality, such as acute severe infections, are particularly vulnerable. We developed IMMORTOOL, an R package accessible via a user-friendly web interface (https://petedodd.github.io/IMMORTOOL-live/). This tool will estimate the potential for immortal time bias using empiric or assumed data on the distributions of time to intervention and time to event. Assumptions are that no other biases are present and that the intervention does not affect the outcome. The tool was benchmarked using studies presenting both naive analyses and analyses with the intervention fit as a time-varying exposure. We applied IMMORTOOL to a set of influential observational studies that used naive analyses when estimating the impact of polyclonal intravenous immunoglobulin (IVIG) on survival in streptococcal toxic shock syndrome (STSS). IMMORTOOL demonstrated that published estimates suggesting a survival advantage from giving IVIG in STSS are explained, at least in part, by immortal time bias. IMMORTOOL can quantify the potential for immortal time bias in observational analyses. Importantly, it simulates only bias resulting from misallocation of person-time, not other related selection biases. The tool may help readers interrogate published studies. We do not advocate IMMORTOOL being used to justify naive analyses where robust analyses are possible. To what extent giving IVIG in STSS improves survival remains uncertain.

KEY POINTSImmortal time bias exaggerates estimates of treatment efficacy in naïve analyses of observational data. We developed a tool to estimate the extent of this bias. The benefits of giving intravenous immunoglobulin in streptococcal toxic shock syndrome have likely been overstatedBackgroundImmortal time bias is a spurious or exaggerated protective association that commonly arises in naive analyses of observational data. It occurs when people receive the intervention because they survive, rather than survive because they received the intervention. Studies in conditions with substantial early mortality, such as acute severe infections, are particularly vulnerable. The bias can be avoided by fitting the intervention as a time-varying exposure.MethodsWe developed IMMORTOOL, an R package accessible via a user-friendly web interface. This tool will estimate the potential for immortal time bias using empiric or assumed data on the distributions of time to intervention and time to event. Assumptions are that no other biases are present and that the intervention does not impact the outcome. The tool was benchmarked using studies presenting both naive analyses and analyses with the intervention fit as a time-varying exposure. We applied IMMORTOOL to a set of influential observational studies that used naive analyses when estimating the impact of polyclonal intravenous immunoglobulin (IVIG) on survival in streptococcal toxic shock syndrome (STSS).ResultsIMMORTOOL demonstrated that published estimates suggesting a survival advantage from giving IVIG in STSS are explained, at least in part, by immortal time bias.ConclusionsIMMORTOOL can quantify the potential for immortal time bias in observational analyses. This may help readers interrogate published studies. We do not advocate IMMORTOOL being used to justify naive analyses where the intervention could be fit as a time-varying exposure. To what extent giving IVIG in STSS improves survival remains uncertain.

Motivated by recent reports on associations between diabetes and cancer, many researchers have used administrative databases to examine risk association of cancer with drug use in patients with diabetes. Many of these studies suffered from major biases in study design and data analysis, which can lead to erroneous conclusions if these biases are not adjusted. This article discusses the sources and impacts of these biases and methods for correction of these biases. To avoid erroneous results, this article suggests performing sensitivity and specificity analysis as well as using a drug with a known effect on an outcome to ascertain the validity of the proposed methods. Using the Hong Kong Diabetes Registry, we illustrated the impacts of biases of drug use indication and prevalent user by examining the effects of statins on cardiovascular disease. We further showed that 'immortal time bias' may have a neutral impact on the estimated drug effect if the hazard is assumed to be constant over time. On the contrary, adjustment for 'immortal time bias' using time-dependent models may lead to misleading results biased towards against the treatment. However, artificial inclusion of immortal time in non-drug users to correct for immortal time bias may bias the result in favour of the therapy. In conclusion, drug use indication bias and prevalent user bias but not immortal time bias are major biases in the design and analysis of drug use effects among patients with diabetes in non-clinical trial settings.

Analysing electronic health records: The benefits of target trial emulation

Reply to Leah Bensimon, Samy Suissa, and Laurent Azoulay's Letter to the Editor re: Daniel E. Spratt, Chi Zhang, Zachary S. Zumsteg, Xin Pei, Zhigang Zhang, Michael J. Zelefsky. Metformin and Prostate Cancer: Reduced Development of Castration-resistant Disease and Prostate Cancer Mortality. Eur Urol 2013;63:709–16

Trials that use a randomization process are preferred for evaluating causal effect. However, these interventional studies are not always feasible due to ethical, time, and cost constraints. Additionally, they are often criticized for not accurately representing patients seen in clinical practice. Observational studies can help bridge this gap, but they are often affected by several sources of bias. A growing framework known as target trial emulation (TTE) is increasingly being applied to real-world data to minimize these common biases in observational studies and, under certain conditions, allow for causal interpretation. In this study, we assessed the 5-y effect of thiazide monotherapy, an antihypertensive medication known to cause xerostomia, on the risk of developing dental caries. We applied the target trial framework to build a TTE cohort using an active comparator new user design. We compared thiazide vs nonthiazide monotherapy using electronic health records from the All of Us Research Program. Furthermore, we developed 2 naïve cohorts: a prevalent user cohort, which included existing medication users, and a nonaligned cohort, which was affected by immortal time bias. These cohorts helped illustrate the advantages of TTE. In the TTE cohort, we found no significant difference in the 5-y risk of dental caries between groups, with a risk difference of 0.29% (95% CI, -0.36% to 0.95%). In contrast, the naïve cohorts showed directionally opposite effects, with 5-y risk differences of 0.65% (95% CI, 0.22% to 1.09%) in the prevalent user cohort and -0.24% (95% CI, -0.55% to 0.07%) in the nonaligned cohort, highlighting the impact of design-related biases on the observed outcomes. We demonstrated how the TTE framework can help avoid common biases in observational research, enabling researchers to answer important oral health questions.

BackgroundTo assess the completeness of reporting, research transparency practices, and risk of selection and immortal bias in observational studies using routinely collected data for comparative effectiveness research.MethodWe performed a meta-research study by searching PubMed for comparative effectiveness observational studies evaluating therapeutic interventions using routinely collected data published in high impact factor journals from 01/06/2018 to 30/06/2020. We assessed the reporting of the study design (i.e., eligibility, treatment assignment, and the start of follow-up). The risk of selection bias and immortal time bias was determined by assessing if the time of eligibility, the treatment assignment, and the start of follow-up were synchronized to mimic the randomization following the target trial emulation framework.ResultSeventy-seven articles were identified. Most studies evaluated pharmacological treatments (69%) with a median sample size of 24,000 individuals. In total, 20% of articles inadequately reported essential information of the study design. One-third of the articles (n = 25, 33%) raised some concerns because of unclear reporting (n = 6, 8%) or were at high risk of selection bias and/or immortal time bias (n = 19, 25%). Only five articles (25%) described a solution to mitigate these biases. Six articles (31%) discussed these biases in the limitations section.ConclusionReporting of essential information of study design in observational studies remained suboptimal. Selection bias and immortal time bias were common methodological issues that researchers and physicians should be aware of when interpreting the results of observational studies using routinely collected data.

Purpose of ReviewThe evidence regarding the clinical effects of drug-drug interactions (DDIs) is scarce and limited. Pharmacoepidemiologic studies could help fill in this important knowledge gap. Here, we review the pharmacoepidemiology of DDIs with a focus on cohort designs. We also highlight the decision-making process with respect to different aspects of cohort study design, potential biases that may arise during this decision process, and mitigation strategies.Recent FindingsConsidering the pharmacologic mechanism of the DDI of interest as well as of the object drug and the precipitant drug separately at the design stage of cohort studies for DDIs will help minimize major biases such as prevalent user bias and confounding by indication. Confounding by indication could also be mitigated by using control precipitants. Further, the correct assignment of the cohort entry date via the application of a time-varying exposure definition can help minimize immortal time bias and prevalent user bias. Minimization of these biases may also potentially be achieved with recently developed tools such as target trial emulation and the prevalent new-user design; however, more research is needed in the area.SummaryCareful consideration of the underlying pharmacology and the specifics of study design will help minimize major biases in cohort studies that aim to assess the clinical effects of DDIs. Recent methodological developments from other areas of pharmacoepidemiology could further improve the internal validity of DDI studies.

BackgroundReal-world observational data are an important source of evidence on the treatment effectiveness for patients hospitalized with coronavirus disease 2019 (COVID-19). However, observational studies evaluating treatment effectiveness based on longitudinal data are often prone to methodological biases such as immortal time bias, confounding bias, and competing risks.MethodsFor exemplary target trial emulation, we used a cohort of patients hospitalized with COVID-19 (n = 501) in a single centre. We described the methodology for evaluating the effectiveness of a single-dose treatment, emulated a trial using real-world data, and drafted a hypothetical study protocol describing the main components. To avoid immortal time and time-fixed confounding biases, we applied the clone-censor-weight technique. We set a 5-day grace period as a period of time when treatment could be initiated. We used the inverse probability of censoring weights to account for the selection bias introduced by artificial censoring. To estimate the treatment effects, we took the multi-state model approach. We considered a multi-state model with five states. The primary endpoint was defined as clinical severity status, assessed by a 5-point ordinal scale on day 30. Differences between the treatment group and standard of care treatment group were calculated using a proportional odds model and shown as odds ratios. Additionally, the weighted cause-specific hazards and transition probabilities for each treatment arm were presented.ResultsOur study demonstrates that trial emulation with a multi-state model analysis is a suitable approach to address observational data limitations, evaluate treatment effects on clinically heterogeneous in-hospital death and discharge alive endpoints, and consider the intermediate state of admission to ICU. The multi-state model analysis allows us to summarize results using stacked probability plots that make it easier to interpret results.ConclusionsExtending the emulated target trial approach to multi-state model analysis complements treatment effectiveness analysis by gaining information on competing events. Combining two methodologies offers an option to address immortal time bias, confounding bias, and competing risk events. This methodological approach can provide additional insight for decision-making, particularly when data from randomized controlled trials (RCTs) are unavailable.