Matching Covariates Research Articles

Abstract P303: Low Albumin and Risk of Type 2 Diabetes

Type 2 diabetes (T2DM) is a burdensome, costly and debilitating disease that causes premature morbidity and mortality. Recent evidence suggests that T2DM is in part an inflammatory disease. It is well known that serum albumin is a negative acute phase reactant, yet few studies have examined the association between albumin and risk of T2DM. The purpose of this analysis was to prospectively examine the association between serum albumin and incident type 2 diabetes in middle-aged and older adults from a defined population. The Western New York Health Study enrolled apparently healthy participants ages 35-79 yrs from Niagara and Erie counties, NY. For this report, at baseline (1996-2001) all participants from the Western New York Health Study had fasting plasma glucose <110 mg/dl. At follow-up, six years later each case (n=61) was matched with up to three controls (n=158) on gender, race/ethnicity, the year of study enrollment, and baseline fasting glucose <110 mg/dl or 110-125 mg/dl. Diabetes at follow-up was defined as: 1) fasting plasma glucose > 125 mg/dl; 2) random glucose >199 mg/dl; or 3) a physician diagnosis and receiving antidiabetic medication. Conditional logistic regression was used to estimate the odds ratio (95% CI) of developing type 2 diabetes by tertile of baseline albumin adjusted for the matching criteria and additional covariates including BMI, smoking and physical activity. Mean age was 58 years; 60% women; 93% White. Multivariate analyses indicated that cases had lower baseline albumin concentrations (OR = 0.41, 95%CI: 0.17, 0.99 (p=0.04; lowest tertile vs highest) (p trend < 0.05). Further analyses indicated that this relation was independent of hs-CRP and IL-6. Stratification by smoking status did not alter the findings. The results of this analysis support the hypothesis that individuals with lower albumin concentrations were more likely to develop type 2 diabetes than those with higher levels.

Matching methods to create paired survival data based on an exposure occurring over time: a simulation study with application to breast cancer.

BackgroundPaired survival data are often used in clinical research to assess the prognostic effect of an exposure. Matching generates correlated censored data expecting that the paired subjects just differ from the exposure. Creating pairs when the exposure is an event occurring over time could be tricky. We applied a commonly used method, Method 1, which creates pairs a posteriori and propose an alternative method, Method 2, which creates pairs in “real-time”. We used two semi-parametric models devoted to correlated censored data to estimate the average effect of the exposure : the Holt and Prentice (HP), and the Lee Wei and Amato (LWA) models. Contrary to the HP, the LWA allowed adjustment for the matching covariates (LWA a ) and for an interaction (LWA i ) between exposure and covariates (assimilated to prognostic profiles). The aim of our study was to compare the performances of each model according to the two matching methods.MethodsExtensive simulations were conducted. We simulated cohort data sets on which we applied the two matching methods, the HP and the LWA. We used our conclusions to assess the prognostic effect of subsequent pregnancy after treatment for breast cancer in a female cohort treated and followed up in eight french hospitals.ResultsIn terms of bias and RMSE, Method 2 performed better than Method 1 in designing the pairs, and LWA a was the best model for all the situations except when there was an interaction between exposure and covariates, for which LWA i was more appropriate. On our real data set, we found opposite effects of pregnancy according to the six prognostic profiles, but none were statistically significant. We probably lacked statistical power or reached the limits of our approach. The pairs’ censoring options chosen for combination Method 2 - LWA had to be compared with others.ConclusionsCorrelated censored data designing by Method 2 seemed to be the most pertinent method to create pairs, when the criterion, which characterized the pair, was an exposure occurring over time. In such a setting, the LWA was the most appropriate model.

Evaluating the Effect of Hospital and Insurance Type on the Risk of 1-year Mortality of Very Low Birth Weight Infants

We examined the effect of hospital type and medical coverage on the risk of 1-year mortality of very low birth weight (VLBW) infants while adjusting for possible selection bias. The study population was limited to singleton live birth infants having birth weight between 500 and 1500 g with no congenital anomalies who were born in Arkansas hospitals between 2001 and 2007. Propensity score (PS) matching and PS covariate adjustment were used to mitigate selection bias. In addition, a conventional multivariable logistic regression model was used for comparison purposes. Generally, all 3 analytical approaches provided consistent results in terms of the estimated relative risk, absolute risk reduction, and the number needed to treat. Using the PS matching method, VLBW infants delivered at a hospital with a neonatal intensive care unit (NICU) were associated with a 35% relative decrease (95% bootstrap confidence interval, 18.5%-48.9%) in the risk of 1-year mortality as compared with those infants delivered at non-NICU hospitals. Furthermore, our results showed that on average, 16 VLBW infants (95% bootstrap confidence interval, 11-32), would need to be delivered at a hospital with an NICU to prevent 1 additional death at 1 year. However, there was not a difference in the risk of 1-year mortality between VLBW infants born to Medicaid-insured versus non-Medicaid-insured women. Estimated relative risk of infant mortality was significantly lower for births that occurred in hospitals with an NICU; therefore, greater efforts should be made to deliver VLBW neonates in an NICU hospital.

Semiparametric regression models for detecting effect modification in matched case‐crossover studies

In matched case-crossover studies, it is generally accepted that covariates on which a case and associated controls are matched cannot exert a confounding effect on independent predictors included in the conditional logistic regression model because any stratum effect is removed by the conditioning on the fixed number of sets of a case and controls in the stratum. Hence, the conditional logistic regression model is not able to detect any effects associated with the matching covariates by stratum. In addition, the matching covariates may be effect modification and the methods for assessing and characterizing effect modification by matching covariates are quite limited. In this article, we propose a unified approach in its ability to detect both parametric and nonparametric relationships between the predictor and the relative risk of disease or binary outcome, as well as potential effect modifications by matching covariates. Two methods are developed using two semiparametric models: (1) the regression spline varying coefficients model and (2) the regression spline interaction model. Simulation results show that the two approaches are comparable. These methods can be used in any matched case-control study and extend to multilevel effect modification studies. We demonstrate the advantage of our approach using an epidemiological example of a 1-4 bi-directional case-crossover study of childhood aseptic meningitis associated with drinking water turbidity.

The Economic Value of a Wellness and Disease Prevention Program

The objective of this study was to determine the economic impact of the Hawaii Medical Service Association's health promotion/disease prevention program. A retrospective analysis of health risk, health claims, and cost was performed using a mixed model factorial design for the years 2002-2005 that compared program participants to nonparticipants. All analyses were adjusted for preexisting observed differences based on sex, age, baseline morbidity, and health care costs between participants and nonparticipants using propensity score matching method and/or covariates as appropriate. After analyzing data from more than 166,000 HMSA members over a 4-year period, participants were found to incur consistently lower costs. Predictive modeling of upward cost trajectories relative to actual health care costs for participants and risk-matched nonparticipants indicated a savings of $350 per participant per year. Those who participated in additional wellness programming demonstrated additional cost savings. This study illustrates the economic value of a comprehensive health promotion program.

Bootstrap Inference for K-Nearest Neighbour Matching Estimators

Abadie and Imbens (2008, Econometrica) showed that classical bootstrap schemes fail to provide correct inference for K-nearest neighbour (KNN) matching estimators of average causal effects. This is an interesting result showing that bootstrap should not be applied without theoretical justification. In this paper, we present two resampling schemes, which we show provide valid inference for KNN matching estimators. We resample estimated individual causal (EICE), i.e. the difference in outcome between matched pairs, instead of the original data. Moreover, by taking differences in EICEs ordered with respect to the matching covariate, we obtain a bootstrap scheme valid also with heterogeneous causal effects where mild assumptions on the heterogeneity are imposed. We provide proofs of the validity of the proposed resampling based inferences. A simulation study illustrates finite sample properties.

Why IQ is not a covariate in cognitive studies of neurodevelopmental disorders

IQ scores are volatile indices of global functional outcome, the final common path of an individual's genes, biology, cognition, education, and experiences. In studying neurocognitive outcomes in children with neurodevelopmental disorders, it is commonly assumed that IQ can and should be partialed out of statistical relations or used as a covariate for specific measures of cognitive outcome. We propose that it is misguided and generally unjustified to attempt to control for IQ differences by matching procedures or, more commonly, by using IQ scores as covariates. We offer logical, statistical, and methodological arguments, with examples from three neurodevelopmental disorders (spina bifida meningomyelocele, learning disabilities, and attention deficit hyperactivity disorder) that: (1) a historical reification of general intelligence, g, as a causal construct that measures aptitude and potential rather than achievement and performance has fostered the idea that IQ has special status and that in studying neurocognitive function in neurodevelopmental disorders; (2) IQ does not meet the requirements for a covariate; and (3) using IQ as a matching variable or covariate has produced overcorrected, anomalous, and counterintuitive findings about neurocognitive function.

Matching in Studies of Classification Accuracy: Implications for Analysis, Efficiency, and Assessment of Incremental Value

In case-control studies evaluating the classification accuracy of a marker, controls are often matched to cases with respect to factors associated with the marker and disease status. In contrast with matching in epidemiologic etiology studies, matching in the classification setting has not been rigorously studied. In this article, we consider the implications of matching in terms of the choice of statistical analysis, efficiency, and assessment of the incremental value of the marker over the matching covariates. We find that adjustment for the matching covariates is essential, as unadjusted summaries of classification accuracy can be biased. In many settings, matching is the most efficient covariate-dependent sampling scheme, and we provide an expression for the optimal matching ratio. However, we also show that matching greatly complicates estimation of the incremental value of the marker. We recommend that matching be carefully considered in the context of these findings.

Using Matching to Estimate Treatment Effects: Data Requirements, Matching Metrics, and Monte Carlo Evidence

We compare propensity-score matching methods with covariatematching estimators. We first discuss the data requirements of propensity-score matching estimators and covariate matching estimators. Then we propose two new matching metrics incorporating the treatment outcome information and participation indicator information, and discuss the motivations of different metrics. Next we study the small-sample properties of propensity-score matching versus covariate matching estimators, and of different matching metrics, through Monte Carlo experiments. Through a series of simulations, we provide some guidance to practitioners on how to choose among different matching estimators and matching metrics.

Semiparametric and nonparametric modeling for effect modification in matched studies

This study describes a new graphical method for assessing and characterizing effect modification by a matching covariate in matched case–control studies. This method to understand effect modification is based on a semiparametric model using a varying coefficient model. The method allows for nonparametric relationships between effect modification and other covariates, or can be useful in suggesting parametric models. This method can be applied to examining effect modification by any ordered categorical or continuous covariates for which cases have been matched with controls. The method applies to effect modification when causality might be reasonably assumed. An example from veterinary medicine is used to demonstrate our approach. The simulation results show that this method, when based on linear, quadratic and nonparametric effect modification, can be more powerful than both a parametric multiplicative model fit and a fully nonparametric generalized additive model fit.

Effect heterogeneity by a matching covariate in matched case-control studies: a method for graphs-based representation.

The authors describe a method for assessing and characterizing effect heterogeneity related to a matching covariate in case-control studies, using an example from veterinary medicine. Data are from a case-control study conducted in Texas during 1997-1998 of 498 pairs of horses with colic and their controls. Horses were matched by veterinarian and by month of examination. The number of matched pairs of cases and controls varied by veterinarian. The authors demonstrate that there is effect heterogeneity related to this characteristic (i.e., cluster size of veterinarians) for the association of colic with certain covariates, using a moving average approach to conditional logistic regression and graphs-based methods. The method described in this report can be applied to examining effect heterogeneity (or effect modification) by any ordered categorical or continuous covariates for which cases have been matched with controls. The method described enables one to understand the pattern of variation across ordered categorical or continuous matching covariates and allows for any shape for this pattern. This method applies to effect modification when causality might be reasonably assumed.

Instrumental Variables: Response to Angrist and Imbens

We compare propensity-score matching methods with covariate matching estimators. We first discuss the data requirements of propensity-score matching estimators and covariate matching estimators. Then we propose two new matching metrics incorporating the ...

OPTIMAL CASE-CONTROL MATCHING IN PRACTICE

We illustrate modern matching techniques and discuss practical issues in defining the closeness of matching for retrospective case-control designs (in which the pool of subjects already exists when the study commences). We empirically compare matching on a balancing score, analogous to the propensity score for treated/control matching, with matching on a weighted distance measure. Although both methods in principle produce balance between cases and controls in the marginal distributions of the matching covariates, the weighted distance measure provides better balance in practice because the balancing score can be poorly estimated. We emphasize the use of optimal matching based on efficient network algorithms. An illustration is based on the design of a case-control study of hepatitis B virus infection as a possible confounder and/or effect modifier of radiation-related primary liver cancer in atomic bomb survivors.

Detecting effect modification due to the variables controlled in multiple matching

A new approach is presented to the study of heterogeneity of the odds ratio in a case-control design when the possible interaction effects are associated with the variables used for multiple matching. This analysis does not require stratification on the matching covariates.