Antipsychotic drugs in first-episode psychosis: A target trial emulation in the FEP-CAUSAL Collaboration.

Good adherence to antipsychotic therapy helps prevent relapses in First Episode Psychosis (FEP). We used data from the FEP-CAUSAL Collaboration, an international consortium of observational cohorts to emulate a target trial comparing antipsychotics with treatment discontinuation as the primary outcome. Other outcomes included all-cause hospitalization. We benchmarked our results to estimates from EUFEST, a randomized trial conducted in the 2000s. We included 1097 patients with a psychotic disorder and less than 2 years since psychosis onset. Inverse probability weighting was used to control for confounding. The estimated 12-month risks of discontinuation for aripiprazole, first-generation agents, olanzapine, paliperidone, quetiapine, and risperidone (95% CI) were: 61.5% (52.5-70.6), 73.5% (60.5-84.9), 76.8% (67.2-85.3), 58.4% (40.4-77.4), 76.5% (62.1-88.5), and 74.4% (67.0-81.2) respectively. Compared with aripiprazole, the 12-month risk differences (95% CI) were -15.3% (-30.0, 0.0) for olanzapine, -12.8% (-25.7, -1.0) for risperidone, and 3.0% (-21.5, 30.8) for paliperidone. The 12-month risks of hospitalization were similar between agents. Our estimates support use of aripiprazole and paliperidone as first-line therapies for FEP. Benchmarking yielded similar results for discontinuation and absolute risks of hospitalization as in the original trial, suggesting that data from the FEP-CAUSAL Collaboration data sufficed to approximately remove confounding for these clinical questions.

Internal validation of gestational age estimation algorithms in healthcare databases using pregnancies conceived with fertility procedures.

Fertility procedures recorded in healthcare databases can be used to estimate the start of pregnancy, which can serve as a reference standard to validate gestational age estimates based on International Classification of Diseases (ICD) codes. In a cohort of 17,398 pregnancies conceived by fertility procedures in MarketScan (2011-2020), we estimated gestational age at the end of pregnancy using algorithms based on (1) days since fertility procedure (the reference); (2) ICD-9/ICD-10 (before/after October 2015) codes indicating gestational length recorded at the end of pregnancy (method A); and (3) ICD-10 enhanced with Z3A codes denoting specific gestation weeks recorded at prenatal visits (method B). We calculated the proportion of pregnancies with an estimated gestational age within 14 days of the reference. Method A accuracy was similar for ICD-9 and ICD-10 codes. After 2015, method B was more accurate than method A: For term births, within-14-day-agreements were 90.8% for method A and 98.7% for method B. Corresponding estimates were 70.1% and 95.6% for preterm births; 35.3% and 92.6% for stillbirths; 54.3% and 64.2% for spontaneous abortions; and 16.7% and 84.6% for elective terminations. ICD-10-based algorithms that incorporate Z3A codes improve the accuracy of gestational age estimation in healthcare databases, especially for preterm and non-live births.

Integrase strand-transfer inhibitor use and cardiovascular events in adults with HIV: an emulation of target trials in the HIV-CAUSAL Collaboration and the Antiretroviral Therapy Cohort Collaboration.

BACKGROUND: A recent observational study suggested that the risk of cardiovascular events could be higher among antiretroviral therapy (ART)-naive individuals with HIV who receive integrase strand-transfer inhibitor (INSTI)-based ART than among those who receive other ART regimens. We aimed to emulate target trials separately in ART-naive and ART-experienced individuals with HIV to examine the effect of using INSTI-based regimens versus other ART regimens on the 4-year risk of cardiovascular events. METHODS: We used routinely recorded clinical data from 12 cohorts that collected information on cardiovascular events, BMI, and blood pressure from two international consortia of cohorts of people with HIV from Europe and North America. For the target trial in individuals who had previously never used ART (ie, ART-naive), eligibility criteria were aged 18 years or older, a detectable HIV-RNA measurement while ART-naive (>50 copies per mL), and no history of a cardiovascular event or cancer. Eligibility criteria for the target trial in those with previous use of non-INSTI-based ART (ie, ART-experienced) were the same except that individuals had to have been on at least one non-INSTI-based ART regimen and be virally suppressed (≤50 copies per mL). We assessed eligibility for both trials for each person-month between January, 2013, and January, 2023, and assigned individuals to the treatment strategy that was compatible with their data. We estimated the standardised 4-year risks of cardiovascular events (myocardial infarction, stroke, or invasive cardiovascular procedure) via pooled logistic regression models adjusting for time and baseline covariates. In per-protocol analyses, we censored individuals if they deviated from their assigned treatment strategy for more than 2 months and weighted uncensored individuals by the inverse of their time-varying probability of remaining uncensored. The denominator of the weight was estimated via a pooled logistic model that included baseline and time-varying covariates. FINDINGS: The analysis in ART-naive individuals included 10 767 INSTI initiators and 8292 non-initiators of INSTI. There were 43 cardiovascular events in INSTI initiators (median follow-up of 29 months; IQR 15-45) and 52 in non-initiators (39 months; 18-47): standardised 4-year risks were 0·76% (95% CI 0·51 to 1·04) in INSTI initiators and 0·75% (0·54 to 0·98) in non-INSTI initiators; risk ratio 1·01 (0·57 to 1·57); risk difference 0·0089% (-0·43 to 0·36). The analysis in ART-experienced individuals included 7875 INSTI initiators and 373 965 non-initiators. There were 56 events in INSTI initiators (median follow-up 18 months; IQR 9-29) and 3103 events (808 unique) in non-INSTI initiators (26 months; 15-37) in non-initiators: standardised 4-year risks 1·41% (95% CI 0·88 to 2·03) in INSTI initiators and 1·48% (1·28 to 1·71) in non-initiators; risk ratio 0·95 (0·60 to 1·36); risk difference -0·068% (-0·60 to 0·52). INTERPRETATION: We estimated that INSTI use did not result in a clinically meaningful increase of cardiovascular events in ART-naive and ART-experienced individuals with HIV. FUNDING: National Institute of Allergy and Infectious Diseases and National Institute on Alcohol Abuse and Alcoholism.

Evaluating Model Specification When Using the Parametric G-Formula in the Presence of Censoring.

The noniterative conditional expectation (NICE) parametric g-formula can be used to estimate the causal effect of sustained treatment strategies. In addition to identifiability conditions, the validity of the NICE parametric g-formula generally requires the correct specification of models for time-varying outcomes, treatments, and confounders at each follow-up time point. An informal approach for evaluating model specification is to compare the observed distributions of the outcome, treatments, and confounders with their parametric g-formula estimates under the "natural course." In the presence of loss to follow-up, however, the observed and natural-course risks can differ even if the identifiability conditions of the parametric g-formula hold and there is no model misspecification. Here, we describe 2 approaches for evaluating model specification when using the parametric g-formula in the presence of censoring: 1) comparing factual risks estimated by the g-formula with nonparametric Kaplan-Meier estimates and 2) comparing natural-course risks estimated by inverse probability weighting with those estimated by the g-formula. We also describe how to correctly compute natural-course estimates of time-varying covariate means when using a computationally efficient g-formula algorithm. We evaluate the proposed methods via simulation and implement them to estimate the effects of dietary interventions in 2 cohort studies.

Evaluating Metformin Strategies for Cancer Prevention: A Target Trial Emulation Using Electronic Health Records.

BACKGROUND: Metformin users appear to have a substantially lower risk of cancer than nonusers in many observational studies. These inverse associations may be explained by common flaws in observational analyses that can be avoided by explicitly emulating a target trial. METHODS: We emulated target trials of metformin therapy and cancer risk using population-based linked electronic health records from the UK (2009-2016). We included individuals with diabetes, no history of cancer, no recent prescription for metformin or other glucose-lowering medication, and hemoglobin A1c (HbA1c) <64 mmol/mol (<8.0%). Outcomes included total cancer and 4 site-specific cancers (breast, colorectal, lung, and prostate). We estimated risks using pooled logistic regression with adjustment for risk factors via inverse-probability weighting. We emulated a second target trial among individuals regardless of diabetes status. We compared our estimates with those obtained using previously applied analytic approaches. RESULTS: Among individuals with diabetes, the estimated 6-year risk differences (metformin - no metformin) were -0.2% (95% CI = -1.6%, 1.3%) in the intention-to-treat analysis and 0.0% (95% CI = -2.1%, 2.3%) in the per-protocol analysis. The corresponding estimates for all site-specific cancers were close to zero. Among individuals regardless of diabetes status, these estimates were also close to zero and more precise. By contrast, previous analytic approaches yielded estimates that appeared strongly protective. CONCLUSIONS: Our findings are consistent with the hypothesis that metformin therapy does not meaningfully influence cancer incidence. The findings highlight the importance of explicitly emulating a target trial to reduce bias in the effect estimates derived from observational analyses.

Sustained virological response after treatment with direct antiviral agents in individuals with HIV and hepatitis C co-infection.

INTRODUCTION: Randomized trials and observational studies have consistently reported rates of sustained virological response (SVR), equivalent to hepatitis C virus (HCV) cure, as high as 95% following treatment with direct-acting antiviral (DAA) treatment in individuals with HIV and HCV co-infection. However, large studies assessing whether SVR rates differ according to demographic and clinical strata are lacking. Additionally, the SVR rates reported in the literature were typically computed in non-random samples of individuals with available post-DAA HCV-RNA measures. Here, we aimed to estimate the probability of SVR after DAA treatment initiation in persons with HIV and HCV co-infection overall and by demographic and clinical characteristics with and without adjustment for missing HCV-RNA testing. METHODS: We included adults with HIV-HCV co-infection who received DAA treatment between 2014 and 2020 in HepCAUSAL, an international collaboration of cohorts from Europe and North America. We estimated the proportions of DAA recipients who had documented SVR (defined as an undetectable HCV-RNA at least 12 weeks after the end of DAA treatment) overall and by strata defined by age, sex, presence of cirrhosis, calendar period, mode of HIV acquisition, CD4 cell count and HCV genotype at DAA treatment. We then compared these rates with those obtained using the parametric g-formula to impute SVR status for individuals with no SVR assessment. RESULTS AND DISCUSSION: A total of 4527 individuals who initiated DAA treatment (88% males, median [IQR] age 56 [50, 62] years) were included. Of the total of 642 (14%) individuals had no HCV-RNA test on or after 12 weeks after the end of treatment. The overall observed and g-formula imputed SVR rates were 93% (95% CI 93, 94) and 94% (95% CI 92, 95), respectively. SVR estimates were similarly high across all strata. A substantial proportion of individuals who received DAA treatment were never assessed for SVR post-DAA and strategies for more systematic routine HCV-RNA testing should be considered. CONCLUSIONS: Our estimates with and without adjustment for missing HCV-RNA testing indicate SVR rates of approximately 95%, like those reported in clinical trials.

Tenofovir disoproxil fumarate and coronavirus disease 2019 outcomes in men with HIV.

OBJECTIVE: To compare the risk of coronavirus disease 2019 (COVID-19) outcomes by antiretroviral therapy (ART) regimens among men with HIV. DESIGN: We included men with HIV on ART in the Veterans Aging Cohort Study who, between February 2020 and October 2021, were 18 years or older and had adequate virological control, CD4 + cell count, and HIV viral load measured in the previous 12 months, and no previous COVID-19 diagnosis or vaccination. METHODS: We compared the adjusted risks of documented severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, COVID-19-related hospitalization, and intensive care unit (ICU) admission by baseline ART regimen: tenofovir alafenamide (TAF)/emtricitabine (FTC), tenofovir disoproxil fumarate (TDF)/FTC, abacavir (ABC)/lamivudine (3TC), and other. We fit pooled logistic regressions to estimate the 18-month risks standardized by demographic and clinical factors. RESULTS: Among 20 494 eligible individuals, the baseline characteristics were similar across regimens, except that TDF/FTC and TAF/FTC had lower prevalences of chronic kidney disease and estimated glomerular filtration rate <60 ml/min. Compared with TAF/FTC, the estimated 18-month risk ratio (95% confidence interval) of documented SARS-CoV-2 infection was 0.65 (0.43, 0.89) for TDF/FTC, 1.00 (0.85, 1.18) for ABC/3TC, and 0.87 (0.70, 1.04) for others. The corresponding risk ratios for COVID-19 hospitalization were 0.43 (0.07, 0.87), 1.09 (0.79, 1.48), and 1.21 (0.88, 1.62). The risk of COVID-19 ICU admission was lowest for TDF/FTC, but the estimates were imprecise. CONCLUSION: Our study suggests that, in men living with HIV, TDF/FTC may protect against COVID-19-related events. Randomized trials are needed to investigate the effectiveness of TDF as prophylaxis for, and early treatment of, COVID-19 in the general population.

Predicting counterfactual risks under hypothetical treatment strategies: an application to HIV.

The accuracy of a prediction algorithm depends on contextual factors that may vary across deployment settings. To address this inherent limitation of prediction, we propose an approach to counterfactual prediction based on the g-formula to predict risk across populations that differ in their distribution of treatment strategies. We apply this to predict 5-year risk of mortality among persons receiving care for HIV in the U.S. Veterans Health Administration under different hypothetical treatment strategies. First, we implement a conventional approach to develop a prediction algorithm in the observed data and show how the algorithm may fail when transported to new populations with different treatment strategies. Second, we generate counterfactual data under different treatment strategies and use it to assess the robustness of the original algorithm's performance to these differences and to develop counterfactual prediction algorithms. We discuss how estimating counterfactual risks under a particular treatment strategy is more challenging than conventional prediction as it requires the same data, methods, and unverifiable assumptions as causal inference. However, this may be required when the alternative assumption of constant treatment patterns across deployment settings is unlikely to hold and new data is not yet available to retrain the algorithm.

Risk for Non-AIDS-Defining and AIDS-Defining Cancer of Early Versus Delayed Initiation of Antiretroviral Therapy : A Multinational Prospective Cohort Study.

BACKGROUND: Immediate initiation of antiretroviral therapy (ART) regardless of CD4 cell count reduces risk for AIDS and non-AIDS-related events in asymptomatic, HIV-positive persons and is the standard of care. However, most HIV-positive persons initiate ART when their CD4 count decreases below 500 × 109 cells/L. Consequences of delayed ART on risk for non-AIDS-defining and AIDS-defining cancer, one of the most common reasons for death in HIV, are unclear. OBJECTIVE: To estimate the long-term risk difference for cancer with the immediate ART strategy. DESIGN: Multinational prospective cohort study. SETTING: The D:A:D (Data collection on Adverse events of anti-HIV Drugs) study, which included HIV-positive persons from Europe, Australia, and the United States. PARTICIPANTS: 8318 HIV-positive persons with at least 1 measurement each of CD4 cell count and viral load while ART-naive (study period, 2006 to 2016). MEASUREMENTS: The parametric g-formula was used, with adjustment for baseline and time-dependent confounders (CD4 cell count and viral load), to assess the 10-year risk for non-AIDS-defining and AIDS-defining cancer of immediate versus deferred (at CD4 counts <350 and <500 × 109 cells/L) ART initiation strategies. RESULTS: During 64 021 person-years of follow-up, 231 cases of non-AIDS-defining cancer and 272 of AIDS-defining cancer occurred among HIV-positive persons with a median age of 36 years (interquartile range, 29 to 43 years). With immediate ART, the 10-year risk for non-AIDS-defining cancer was 2.97% (95% CI, 2.37% to 3.50%) and that for AIDS-defining cancer was 2.50% (CI, 2.37% to 3.38%). Compared with immediate ART initiation, the 10-year absolute risk differences when deferring ART to CD4 counts less than 500 × 109 cells/L and less than 350 × 109 cells/L were 0.12 percentage point (CI, -0.01 to 0.26 percentage point) and 0.29 percentage point (CI, -0.03 to 0.73 percentage point), respectively, for non-AIDS-defining cancer and 0.32 percentage point (CI, 0.21 to 0.44 percentage point) and 1.00 percentage point (CI, 0.67 to 1.44 percentage points), respectively, for AIDS-defining cancer. LIMITATION: Potential residual confounding due to observational study design. CONCLUSION: In this young cohort, effects of immediate ART on 10-year risk for cancer were small, and further supportive data are needed for non-AIDS-defining cancer. PRIMARY FUNDING SOURCE: Highly Active Antiretroviral Therapy Oversight Committee.

Hypothetical blood-pressure-lowering interventions and risk of stroke and dementia.

We aimed to study the effects of hypothetical interventions on systolic blood pressure (SBP) and smoking on risk of stroke and dementia using data from 15 years of follow-up in the Rotterdam Study. We used data from 4930 individuals, aged 55-80 years, with no prior history of stroke, dementia or cognitive impairment, followed for 15 years within the Rotterdam Study, a population-based cohort. We defined the following sustained interventions on SBP: (1) maintaining SBP below 120 mmHg, (2) maintaining SBP below 140 mmHg, (3) reducing SBP by 10% if above 140 mmHg, (4) reducing SBP by 20% if above 140 mmHg, and a combined intervention of quitting smoking with each of these SBP-lowering strategies. We considered incident stroke and incident dementia diagnoses as outcomes. We applied the parametric g-formula to adjust for baseline and time-varying confounding. The observed 15-year risk for stroke was 10.7%. Compared to no specified intervention (i.e., the "natural course"), all interventions that involved reducing SBP were associated with a stroke risk reduction of about 10% (e.g., reducing SBP by 20% if above 140 mmHg risk ratio: 0.89; 95% CI 0.76, 1). Jointly intervening on SBP and smoking status further decreased the risk of stroke (e.g., risk ratio: 0.83; 95% CI 0.71, 0.94). None of the specified interventions were associated with a substantive change in dementia risk. Our study suggests that a joint intervention on SBP and smoking cessation during later life may reduce stroke risk, while the potential for reducing dementia risk were not observed.

Emulating a target trial in case-control designs: an application to statins and colorectal cancer.

BACKGROUND: Previous case-control studies have reported a strong association between statin use and lower cancer risk. It is unclear whether this association reflects a benefit of statins or is the result of design decisions that cannot be mapped to a (hypothetical) target trial (that would answer the question of interest). METHODS: We outlined the protocol of a target trial to estimate the effect of statins on colorectal cancer incidence among adults with low-density lipoprotein (LDL) cholesterol below 5 mmol/L. We then emulated the target trial using linked electronic health records of 752 469 eligible UK adults (CALIBER 1999-2016) under both a cohort design and a case-control sampling of the cohort. We used pooled logistic regression to estimate intention-to-treat and per-protocol effects of statins on colorectal cancer, with adjustment for baseline and time-varying risk factors via inverse-probability weighting. Finally, we compared our case-control effect estimates with those obtained using previous case-control procedures. RESULTS: Over the 6-year follow-up, 3596 individuals developed colorectal cancer. Estimated intention-to-treat and per-protocol hazard ratios were 1.00 (95% confidence interval [CI]: 0.87, 1.16) and 0.90 (95% CI: 0.71, 1.12), respectively. As expected, adequate case-control sampling yielded the same estimates. By contrast, previous case-control analytical approaches yielded estimates that appeared strongly protective (odds ratio 0.57, 95% CI: 0.36, 0.91, for ≥5 vs. <5 years of statin use). CONCLUSIONS: Our study demonstrates how to explicitly emulate a target trial using case-control data to reduce discrepancies between observational and randomized trial evidence. This approach may inform future case-control analyses for comparative effectiveness research.

Emulating a target trial of statin use and risk of dementia using cohort data.

OBJECTIVE: Observational data can be used to attempt to emulate a target trial of statin use and estimate analogues of intention-to-treat and per protocol effects on dementia risk. METHODS: Using data from a prospective cohort study in the Netherlands, we conceptualized a sequence of "trials" in which eligible individuals ages 55-80 years were classified as statin initiators or noninitiators for every consecutive month between 1993 and 2007 and were followed until diagnosis of dementia, death, loss to follow-up, or the end of follow-up. We estimated 2 types of effects of statin use on dementia and a combined endpoint of dementia or death: the effect of initiation vs no initiation and the effect of sustained use vs no use. We estimated risk by statin treatment strategy over time via pooled logistic regression. We used inverse-probability weighting to account for treatment-confounder feedback in estimation of per-protocol effects. RESULTS: Of 233,526 eligible person-trials (6,373 individuals), there were 622 initiators and 232,904 noninitiators. Comparing statin initiation with no initiation, the 10-year risk differences (95% confidence interval) were -0.1% (-2.3% to 1.8%) for dementia and 0.3% (-2.7% to 3.3%) for dementia or death. Comparing sustained statin use vs no use, the 10-year risk differences were -2.2% (-5.2% to 1.6%) for dementia and -5.1% (-10.5% to -1.1%) for dementia or death. CONCLUSIONS: Individuals with sustained statin use, but not statin initiation alone, had reduced 10-year risks of dementia and dementia or death. Our results should be interpreted with caution due to the small number of initiators and events and potential for residual confounding.

gfoRmula: An R Package for Estimating the Effects of Sustained Treatment Strategies via the Parametric g-formula.

Researchers are often interested in estimating the causal effects of sustained treatment strategies, i.e., of (hypothetical) interventions involving time-varying treatments. When using observational data, estimating those effects requires adjustment for confounding. However, conventional regression methods cannot appropriately adjust for confounding in the presence of treatment-confounder feedback. In contrast, estimators derived from Robins's g-formula may correctly adjust for confounding even if treatment-confounder feedback exists. The package gfoRmula implements in R one such estimator: the parametric g-formula. This estimator can be used to estimate the effects of binary or continuous time-varying treatments as well as contrasts defined by static or dynamic, deterministic, or random interventions, as well as interventions that depend on the natural value of treatment. The package accommodates survival outcomes as well as binary or continuous outcomes measured at the end of follow-up. This paper describes the gfoRmula package, along with motivating background, features, and examples.

Estimates of Overall Survival in Patients With Cancer Receiving Different Treatment Regimens: Emulating Hypothetical Target Trials in the Surveillance, Epidemiology, and End Results (SEER)-Medicare Linked Database.

Importance: The Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database may provide insights into the comparative effectiveness of oncological treatments for elderly individuals who are underrepresented in clinical trials. Objective: To evaluate the suitability of SEER-Medicare data for assessing the effectiveness of adding a drug to an existing treatment regimen on the overall survival of elderly patients with cancer. Design, Setting, and Participants: This comparative effectiveness study analyzed SEER-Medicare data from 9549 individuals who received a new diagnosis of stage II colorectal cancer (2008-2012) and 940 patients who received a new diagnosis of advanced pancreatic adenocarcinoma (2007-2012), with follow-up to December 31, 2013 (SEER-Medicare data released in 2015). Two (hypothetical) target trials were designed and emulated based on 2 existing randomized clinical trials: (1) adjuvant fluorouracil after curative surgery for individuals with stage II colorectal cancer and (2) erlotinib added to gemcitabine for individuals with advanced pancreatic adenocarcinoma. Data were analyzed January 2018 to March 2019. Exposures: The following treatment strategies were compared: (1) fluorouracil initiation vs no initiation within 3 months of tumor resection and (2) erlotinib initiation vs no initiation within 12 weeks of gemcitabine initiation. Main Outcomes and Measures: All-cause mortality within 60 months of baseline for the fluorouracil trial and within 72 weeks for the erlotinib trial. Results: Compared with 3293 individuals in the existing fluorouracil trial, 9549 eligible individuals included in the present analyses were more likely to have colon cancer (8565 [90%] vs 2291 [71%]) and were older (median [interquartile range], 79 [73-84] vs 63 [56-68] years). The 5-year risk difference for initiation vs noninitiation of fluorouracil after surgery was -3.8% (95% CI, -14.8% to 12.6%), and the mortality hazard ratio (HR) was 0.95 (95% CI, 0.85-1.04). Compared with 569 individuals in the existing erlotinib trial, 940 eligible patients included in the present analysis were older (median [range], 74 [66-93] vs 64 [36-92] years) and more likely to be male (547 [58%] vs 298 [52%]). The 1-year risk difference for initiation vs noninitiation of erlotinib was 4.7% (95% CI, -9.4% to 18.0%), and the corresponding mortality HR was 1.04 (95% CI, 0.86-1.42). In naive analyses, the mortality HR estimate was 1.14 (95% CI, 0.95-1.36) for the fluorouracil emulation and 0.68 (95% CI, 0.54-0.87) for the erlotinib emulation. Conclusions and Relevance: The present estimates were similar to those from randomized clinical trials that studied adding the same cancer drugs to existing regimens. The published HR was 1.02 (95% CI, 0.70-1.48) in the fluorouracil trial for individuals aged 70 or older and 0.96 (95% CI, 0.74-1.24) in the erlotinib trial for individuals aged 65 years or older. The SEER-Medicare database may be adequate for studying the real-world effectiveness of adding a drug to treatment regimens used for elderly individuals with cancer.

Continuation of Annual Screening Mammography and Breast Cancer Mortality in Women Older Than 70 Years.

Background: Randomized trials have shown that initiating breast cancer screening between ages 50 and 69 years and continuing it for 10 years decreases breast cancer mortality. However, no trials have studied whether or when women can safely stop screening mammography. An estimated 52% of women aged 75 years or older undergo screening mammography in the United States. Objective: To estimate the effect of breast cancer screening on breast cancer mortality in Medicare beneficiaries aged 70 to 84 years. Design: Large-scale, population-based, observational study of 2 screening strategies: continuing annual mammography, and stopping screening. Setting: U.S. Medicare program, 2000 to 2008. Participants: 1 058 013 beneficiaries aged 70 to 84 years who had a life expectancy of at least 10 years, had no previous breast cancer diagnosis, and underwent screening mammography. Measurements: Eight-year breast cancer mortality, incidence, and treatments, plus the positive predictive value of screening mammography by age group. Results: In women aged 70 to 74 years, the estimated difference in 8-year risk for breast cancer death between continuing and stopping screening was -1.0 (95% CI, -2.3 to 0.1) death per 1000 women (hazard ratio, 0.78 [CI, 0.63 to 0.95]) (a negative risk difference favors continuing). In those aged 75 to 84 years, the corresponding risk difference was 0.07 (CI, -0.93 to 1.3) death per 1000 women (hazard ratio, 1.00 [CI, 0.83 to 1.19]). Limitations: The available Medicare data permit only 8 years of follow-up after screening. As with any study using observational data, the estimates could be affected by residual confounding. Conclusion: Continuing annual breast cancer screening past age 75 years did not result in substantial reductions in 8-year breast cancer mortality compared with stopping screening. Primary Funding Source: National Institutes of Health.

Inverse probability weighted estimation of risk under representative interventions in observational studies.

Researchers are often interested in using observational data to estimate the effect on a health outcome of maintaining a continuous treatment within a pre-specified range over time; e.g. "always exercise at least 30 minutes per day". There may be many precise interventions that could achieve this range. In this paper we consider representative interventions. These are special cases of random dynamic interventions; interventions under which treatment at each time is assigned according to a random draw from a distribution that may depend on a subject's measured past. Estimators of risk under representative interventions on a time-varying treatment have previously been described based on g-estimation of structural nested cumulative failure time models. In this paper, we consider an alternative approach based on inverse probability weighting (IPW) of marginal structural models. In particular, we show that the risk under a representative intervention on a time-varying continuous treatment can be consistently estimated via computationally simple IPW methods traditionally used for deterministic static (i.e. "nonrandom" and "nondynamic") interventions for binary treatments. We present an application of IPW in this setting to estimate the 28-year risk of coronary heart disease under various representative interventions on lifestyle behaviors in the Nurses Health Study.

Effectiveness of Transmitted Drug Resistance Testing Before Initiation of Antiretroviral Therapy in HIV-Positive Individuals.

BACKGROUND: For people living with HIV, major guidelines in high-income countries recommend testing for transmitted drug resistance (TDR) to guide the choice of first-line antiretroviral therapy (ART). However, individuals who fail a first-line regimen can now be switched to one of several effective regimens. Therefore, the virological and clinical benefit of TDR testing needs to be evaluated. METHODS: We included individuals from the HIV-CAUSAL Collaboration who enrolled <6 months of HIV diagnosis between 2006 and 2015, were ART-naive, and had measured CD4 count and HIV-RNA. Follow-up started at the date when all inclusion criteria were first met (baseline). We compared 2 strategies: (1) TDR testing within 3 months of baseline versus (2) no TDR testing. We used inverse probability weighting to estimate the 5-year proportion and hazard ratios (HRs) of virological suppression (confirmed HIV-RNA <50 copies/mL), and of AIDS or death under both strategies. RESULTS: Of 25,672 eligible individuals (82% males, 52% diagnosed in 2010 or later), 17,189 (67%) were tested for TDR within 3 months of baseline. Of these, 6% had intermediate- or high-level TDR to any antiretroviral drug. The estimated 5-year proportion virologically suppressed was 77% under TDR testing and 74% under no TDR testing; HR 1.06 (95% confidence interval: 1.03 to 1.19). The estimated 5-year risk of AIDS or death was 6% under both strategies; HR 1.03 (95% confidence interval: 0.95 to 1.12). CONCLUSIONS: TDR prevalence was low. Although TDR testing improved virological response, we found no evidence that it reduced the incidence of AIDS or death in first 5 years after diagnosis.

Avoidable flaws in observational analyses: an application to statins and cancer.

The increasing availability of large healthcare databases is fueling an intense debate on whether real-world data should play a role in the assessment of the benefit-risk of medical treatments. In many observational studies, for example, statin users were found to have a substantially lower risk of cancer than in meta-analyses of randomized trials. Although such discrepancies are often attributed to a lack of randomization in the observational studies, they might be explained by flaws that can be avoided by explicitly emulating a target trial (the randomized trial that would answer the question of interest). Using the electronic health records of 733,804 UK adults, we emulated a target trial of statins and cancer and compared our estimates with those obtained using previously applied analytic approaches. Over the 10-yr follow-up, 28,408 individuals developed cancer. Under the target trial approach, estimated observational analogs of intention-to-treat and per-protocol 10-yr cancer-free survival differences were -0.5% (95% confidence interval (CI) -1.0%, 0.0%) and -0.3% (95% CI -1.5%, 0.5%), respectively. By contrast, previous analytic approaches yielded estimates that appeared to be strongly protective. Our findings highlight the importance of explicitly emulating a target trial to reduce bias in the effect estimates derived from observational analyses.

Effect Estimates in Randomized Trials and Observational Studies: Comparing Apples With Apples.

Effect estimates from randomized trials and observational studies might not be directly comparable because of differences in study design, other than randomization, and in data analysis. We propose a 3-step procedure to facilitate meaningful comparisons of effect estimates from randomized trials and observational studies: 1) harmonization of the study protocols (eligibility criteria, treatment strategies, outcome, start and end of follow-up, causal contrast) so that the studies target the same causal effect, 2) harmonization of the data analysis to estimate the causal effect, and 3) sensitivity analyses to investigate the impact of discrepancies that could not be accounted for in the harmonization process. To illustrate our approach, we compared estimates of the effect of immediate with deferred initiation of antiretroviral therapy in individuals positive for the human immunodeficiency virus from the Strategic Timing of Antiretroviral Therapy (START) randomized trial and the observational HIV-CAUSAL Collaboration.