Daniel O. Scharfstein

The Prevention and Treatment of Missing Data in Clinical Trials

Missing data in clinical trials can have a major effect on the validity of the inferences that can be drawn from the trial. This article reviews methods for preventing missing data and, failing that, dealing with data that are missing.

Adjusting for Nonignorable Drop-Out Using Semiparametric Nonresponse Models

Abstract Consider a study whose design calls for the study subjects to be followed from enrollment (time t = 0) to time t = T, at which point a primary endpoint of interest Y is to be measured. The design of the study also calls for measurements on a vector V t) of covariates to be made at one or more times t during the interval [0, T). We are interested in making inferences about the marginal mean μ0 of Y when some subjects drop out of the study at random times Q prior to the common fixed end of follow-up time T. The purpose of this article is to show how to make inferences about μ0 when the continuous drop-out time Q is modeled semiparametrically and no restrictions are placed on the joint distribution of the outcome and other measured variables. In particular, we consider two models for the conditional hazard of drop-out given (V(T), Y), where V(t) denotes the history of the process V t) through time t, t ∈ [0, T). In the first model, we assume that λQ(t|V(T), Y) exp(α0 Y), where α0 is a scalar paramet...

Semiparametric Regression for Repeated Outcomes with Nonignorable Nonresponse

Abstract We consider inference about the parameter β* indexing the conditional mean of a vector of correlated outcomes given a vector of explanatory variables when some of the outcomes are missing in a subsample of the study and the probability of response depends on both observed and unobserved data values; that is, nonresponse is nonignorable. We propose a class of augmented inverse probability of response weighted estimators that are consistent and asymptotically normal (CAN) for estimating β* when the response probabilities can be parametrically modeled and a CAN estimator exists. The proposed estimators do not require full specification of a parametric likelihood, and their computation does not require numerical integration. Our estimators can be viewed as an extension of generalized estimating equation estimators that allows for nonignorable nonresponse. We show that our class essentially consists of all CAN estimators of β*. We also show that the asymptotic variance of the optimal estimator in our ...

Sensitivity Analysis for Selection bias and unmeasured Confounding in missing Data and Causal inference models

In both observational and randomized studies, subjects commonly drop out of the study (i.e., become censored) before end of follow-up. If, conditional on the history of the observed data up to t, the hazard of dropping out of the study (i.e., censoring) at time t does not depend on the possibly unobserved data subsequent to t, we say drop-out is ignorable or explainable (Rubin, 1976). On the other hand, if the hazard of drop-out depends on the possibly unobserved future, we say drop-out is non-ignorable or, equivalently, that there is selection bias on unobservables. Neither the existence of selection bias on unobservables nor its magnitude is identifiable from the joint distribution of the observables. In view of this fact, we argue that the data analyst should conduct a “sensitivity analysis” to quantify how one’s inference concerning an outcome of interest varies as a function of the magnitude of non-identifiable selection bias.

The Value of Trauma Center Care

BACKGROUND The cost of trauma center care is high, raising questions about the value of a regionalized approach to trauma care. To address these concerns, we estimate 1-year and lifetime treatment costs and measure the cost-effectiveness of treatment at a Level I trauma center (TC) compared with a nontrauma center hospital (NTC). METHODS Estimates of cost-effectiveness were derived using data on 5,043 major trauma patients enrolled in the National Study on Costs and Outcomes of Trauma, a prospective cohort study of severely injured adult patients cared for in 69 hospitals in 14 states. Data on costs were derived from multiple sources including claims data from the Centers for Medicare and Medicaid Services, UB92 hospital bills, and patient interviews. Cost-effectiveness was estimated as the ratio of the difference in costs (for treatment at a TC vs. NTC) divided by the difference in life years gained (and lives saved). We also measured cost-effectiveness per quality-adjusted life year gained where quality of life was measured using the SF-6D. We used inverse probability of treatment weighting to adjust for observable differences between patients treated at TCs and NTCs. RESULTS The added cost for treatment at a TC versus NTC was

The Effects of Guided Care on the Perceived Quality of Health Care for Multi-morbid Older Persons: 18-Month Outcomes from a Cluster-Randomized Controlled Trial

36,319 per life-year gained (

Semiparametric Efficiency and its Implication on the Design and Analysis of Group-Sequential Studies

790,931 per life saved) and

A Longitudinal Study of Vaginal Douching and Bacterial Vaginosis—A Marginal Structural Modeling Analysis

36,961 per quality-adjusted life years gained. Cost-effectiveness was more favorable for patients with injuries of higher versus lower severity and for younger versus older patients. CONCLUSIONS Our findings provide evidence that regionalization of trauma care is not only effective but also it is cost-effective.

Evaluating Health Outcomes in the Presence of Competing Risks A Review of Statistical Methods and Clinical Applications

BACKGROUNDThe quality of health care for older Americans with chronic conditions is suboptimal.OBJECTIVETo evaluate the effects of “Guided Care” on patient-reported quality of chronic illness care.DESIGNCluster-randomized controlled trial of Guided Care in 14 primary care teams.PARTICIPANTSOlder patients of these teams were eligible to participate if, based on analysis of their recent insurance claims, they were at risk for incurring high health-care costs during the coming year. Small teams of physicians and their at-risk older patients were randomized to receive either Guided Care (GC) or usual care (UC).INTERVENTION“Guided Care” is designed to enhance the quality of health care by integrating a registered nurse, trained in chronic care, into a primary care practice to work with 2–5 physicians in providing comprehensive chronic care to 50–60 multi-morbid older patients.MEASUREMENTSEighteen months after baseline, interviewers blinded to group assignment administered the Patient Assessment of Chronic Illness Care (PACIC) survey by telephone. Logistic and linear regression was used to evaluate the effect of the intervention on patient-reported quality of chronic illness care.RESULTSOf the 13,534 older patients screened, 2,391 (17.7%) were eligible to participate in the study, of which 904 (37.8%) gave informed consent and were cluster-randomized. After 18 months, 95.3% and 92.2% of the GC and UC recipients who remained alive and eligible completed interviews. Compared to UC recipients, GC recipients had twice greater odds of rating their chronic care highly (aOR = 2.13, 95% CI = 1.30–3.50, p = 0.003).CONCLUSIONGuided Care improves self-reported quality of chronic health care for multi-morbid older persons.

The impact of an intensivist-model ICU on trauma-related mortality

Abstract Authors have shown that the time-sequential joint distributions of many statistics used to analyze data arising from group-sequential time-to-event and longitudinal studies are multivariate normal with an independent increments covariance structure. In Theorem 1 of this article, we demonstrate that this limiting distribution arises naturally when one uses an efficient test statistic to test a single parameter in a semiparametric or parametric model. Because we are able to think of many of the statistics in the literature in this fashion, the limiting distribution under investigation is just a special case of Theorem 1. Using this general structure, we then develop an information-based design and monitoring procedure that can be applied to any type of model for any type of group-sequential study provided that there is a unique parameter of interest that can be efficiently tested.