Semiparametric Theory and Empirical Processes in Causal Inference

Covariate adjustment and estimation of difference in proportions in randomized clinical trials

Difference in proportions is frequently used to measure treatment effect for binary outcomes in randomized clinical trials. The estimation of difference in proportions can be assisted by adjusting for prognostic baseline covariates to enhance precision and bolster statistical power. Standardization or g-computation is a widely used method for covariate adjustment in estimating unconditional difference in proportions, because of its robustness to model misspecification. Various inference methods have been proposed to quantify the uncertainty and confidence intervals based on large-sample theories. However, their performances under small sample sizes and model misspecification have not been comprehensively evaluated. We propose an alternative approach to estimate the unconditional variance of the standardization estimator based on the robust sandwich estimator to further enhance the finite sample performance. Extensive simulations are provided to demonstrate the performances of the proposed method, spanning a wide range of sample sizes, randomization ratios, and model specification. We apply the proposed method in a real data example to illustrate the practical utility.

Multiply robust estimators in longitudinal studies with missing data under control-based imputation

Longitudinal studies are often subject to missing data. The recent guidance from regulatory agencies, such as the ICH E9(R1) addendum addresses the importance of defining a treatment effect estimand with the consideration of intercurrent events. Jump-to-reference (J2R) is one classical control-based scenario for the treatment effect evaluation, where the participants in the treatment group after intercurrent events are assumed to have the same disease progress as those with identical covariates in the control group. We establish new estimators to assess the average treatment effect based on a proposed potential outcomes framework under J2R. Various identification formulas are constructed, motivating estimators that rely on different parts of the observed data distribution. Moreover, we obtain a novel estimator inspired by the efficient influence function, with multiple robustness in the sense that it achieves n1/2-consistency if any pairs of multiple nuisance functions are correctly specified, or if the nuisance functions converge at a rate not slower than n-1/4 when using flexible modeling approaches. The finite-sample performance of the proposed estimators is validated in simulation studies and an antidepressant clinical trial.

Longitudinal plasmode algorithms to evaluate statistical methods in realistic scenarios: an illustration applied to occupational epidemiology

INTRODUCTION

Plasmode simulations are a type of simulations that use real data to determine the synthetic data-generating equations. Such simulations thus allow evaluating statistical methods under realistic conditions. As far as we know, no plasmode algorithm has been proposed for simulating longitudinal data. In this paper, we propose a longitudinal plasmode framework to generate realistic data with both a time-varying exposure and time-varying covariates. This work was motivated by the objective of comparing different methods for estimating the causal effect of a cumulative exposure to psychosocial stressors at work over time.

METHODS

We developed two longitudinal plasmode algorithms: a parametric and a nonparametric algorithms. Data from the PROspective Québec (PROQ) Study on Work and Health were used as an input to generate data with the proposed plasmode algorithms. We evaluated the performance of multiple estimators of the parameters of marginal structural models (MSMs): inverse probability of treatment weighting, g-computation and targeted maximum likelihood estimation. These estimators were also compared to standard regression approaches with either adjustment for baseline covariates only or with adjustment for both baseline and time-varying covariates.

RESULTS

Standard regression methods were susceptible to yield biased estimates with confidence intervals having coverage probability lower than their nominal level. The bias was much lower and coverage of confidence intervals was much closer to the nominal level when considering MSMs. Among MSM estimators, g-computation overall produced the best results relative to bias, root mean squared error and coverage of confidence intervals. No method produced unbiased estimates with adequate coverage for all parameters in the more realistic nonparametric plasmode simulation.

CONCLUSION

The proposed longitudinal plasmode algorithms can be important methodological tools for evaluating and comparing analytical methods in realistic simulation scenarios. To facilitate the use of these algorithms, we provide R functions on GitHub. We also recommend using MSMs when estimating the effect of cumulative exposure to psychosocial stressors at work.

Challenges in Obtaining Valid Causal Effect Estimates with Machine Learning Algorithms

Unlike parametric regression, machine learning (ML) methods do not generally require precise knowledge of the true data generating mechanisms. As such, numerous authors have advocated for ML methods to estimate causal effects. Unfortunately, ML algorithmscan perform worse than parametric regression. We demonstrate the performance of ML-based single- and double-robust estimators. We use 100 Monte Carlo samples with sample sizes of 200, 1200, and 5000 to investigate bias and confidence interval coverage under several scenarios. In a simple confounding scenario, confounders were related to the treatment and the outcome via parametric models. In a complex confounding scenario, the simple confounders were transformed to induce complicated nonlinear relationships. In the simple scenario, when ML algorithms were used, double-robust estimators were superior to single-robust estimators. In the complex scenario, single-robust estimators with ML algorithms were at least as biased as estimators using misspecified parametric models. Double-robust estimators were less biased, but coverage was well below nominal. The use of sample splitting, inclusion of confounder interactions, reliance on a richly specified ML algorithm, and use of doubly robust estimators was the only explored approach that yielded negligible bias and nominal coverage. Our results suggest that ML based singly robust methods should be avoided.

Robust Variance Estimation for Covariate-Adjusted Unconditional Treatment Effect in Randomized Clinical Trials with Binary Outcomes

To improve precision of estimation and power of testing hypothesis for an unconditional treatment effect in randomized clinical trials with binary outcomes, researchers and regulatory agencies recommend using g-computation as a reliable method of covariate adjustment. However, the practical application of g-computation is hindered by the lack of an explicit robust variance formula that can be used for different unconditional treatment effects of interest. To fill this gap, we provide explicit and robust variance estimators for g-computation estimators and demonstrate through simulations that the variance estimators can be reliably applied in practice.

A generalization of moderated statistics to data adaptive semiparametric estimation in high-dimensional biology

The widespread availability of high-dimensional biological data has made the simultaneous screening of many biological characteristics a central problem in computational and high-dimensional biology. As the dimensionality of datasets continues to grow, so too does the complexity of identifying biomarkers linked to exposure patterns. The statistical analysis of such data often relies upon parametric modeling assumptions motivated by convenience, inviting opportunities for model misspecification. While estimation frameworks incorporating flexible, data adaptive regression strategies can mitigate this, their standard variance estimators are often unstable in high-dimensional settings, resulting in inflated Type-I error even after standard multiple testing corrections. We adapt a shrinkage approach compatible with parametric modeling strategies to semiparametric variance estimators of a family of efficient, asymptotically linear estimators of causal effects, defined by counterfactual exposure contrasts. Augmenting the inferential stability of these estimators in high-dimensional settings yields a data adaptive approach for robustly uncovering stable causal associations, even when sample sizes are limited. Our generalized variance estimator is evaluated against appropriate alternatives in numerical experiments, and an open source R/Bioconductor package, biotmle, is introduced. The proposal is demonstrated in an analysis of high-dimensional DNA methylation data from an observational study on the epigenetic effects of tobacco smoking.

Causal Inference for Social Network Data

We describe semiparametric estimation and inference for causal effects using observational data from a single social network. Our asymptotic results are the first to allow for dependence of each observation on a growing number of other units as sample size increases. In addition, while previous methods have implicitly permitted only one of two possible sources of dependence among social network observations, we allow for both dependence due to transmission of information across network ties and for dependence due to latent similarities among nodes sharing ties. We propose new causal effects that are specifically of interest in social network settings, such as interventions on network ties and network structure. We use our methods to reanalyze an influential and controversial study that estimated causal peer effects of obesity using social network data from the Framingham Heart Study; after accounting for network structure we find no evidence for causal peer effects.

Doubly robust estimators for generalizing treatment effects on survival outcomes from randomized controlled trials to a target population

In the presence of heterogeneity between the randomized controlled trial (RCT) participants and the target population, evaluating the treatment effect solely based on the RCT often leads to biased quantification of the real-world treatment effect. To address the problem of lack of generalizability for the treatment effect estimated by the RCT sample, we leverage observational studies with large samples that are representative of the target population. This article concerns evaluating treatment effects on survival outcomes for a target population and considers a broad class of estimands that are functionals of treatment-specific survival functions, including differences in survival probability and restricted mean survival times. Motivated by two intuitive but distinct approaches, i.e., imputation based on survival outcome regression and weighting based on inverse probability of sampling, censoring, and treatment assignment, we propose a semiparametric estimator through the guidance of the efficient influence function. The proposed estimator is doubly robust in the sense that it is consistent for the target population estimands if either the survival model or the weighting model is correctly specified and is locally efficient when both are correct. In addition, as an alternative to parametric estimation, we employ the nonparametric method of sieves for flexible and robust estimation of the nuisance functions and show that the resulting estimator retains the root-n consistency and efficiency, the so-called rate-double robustness. Simulation studies confirm the theoretical properties of the proposed estimator and show that it outperforms competitors. We apply the proposed method to estimate the effect of adjuvant chemotherapy on survival in patients with early-stage resected non-small cell lung cancer.

Targeted maximum likelihood estimation of causal effects with interference: A simulation study

Interference, the dependency of an individual's potential outcome on the exposure of other individuals, is a common occurrence in medicine and public health. Recently, targeted maximum likelihood estimation (TMLE) has been extended to settings of interference, including in the context of estimation of the mean of an outcome under a specified distribution of exposure, referred to as a policy. This paper summarizes how TMLE for independent data is extended to general interference (network-TMLE). An extensive simulation study is presented of network-TMLE, consisting of four data generating mechanisms (unit-treatment effect only, spillover effects only, unit-treatment and spillover effects, infection transmission) in networks of varying structures. Simulations show that network-TMLE performs well across scenarios with interference, but issues manifest when policies are not well-supported by the observed data, potentially leading to poor confidence interval coverage. Guidance for practical application, freely available software, and areas of future work are provided.

Cumming S, Diop A, Talbot D. Confounding adjustment methods for multi-level treatment comparisons under lack of positivity and unknown model specification

Imbalances in covariates between treatment groups are frequent in observational studies and can lead to biased comparisons. Various adjustment methods can be employed to correct these biases in the context of multi-level treatments (> 2). Analytical challenges, such as positivity violations and incorrect model specification due to unknown functional relationships between covariates and treatment or outcome, may affect their ability to yield unbiased results. Such challenges were expected in a comparison of fire-suppression interventions for preventing fire growth. We identified the overlap weights, augmented overlap weights, bias-corrected matching and targeted maximum likelihood as methods with the best potential to address those challenges. A simple variance estimator for the overlap weight estimators that can naturally be combined with machine learning is proposed. In a simulation study, we investigated the performance of these methods as well as those of simpler alternatives. Adjustment methods that included an outcome modeling component performed better than those that focused on the treatment mechanism in our simulations. Additionally, machine learning implementation was observed to efficiently compensate for the unknown model specification for the former methods, but not the latter. Based on these results, we compared the effectiveness of fire-suppression interventions using the augmented overlap weight estimator.

Machine learning for improving high-dimensional proxy confounder adjustment in healthcare database studies: an overview of the current literature

Controlling for large numbers of variables that collectively serve as 'proxies' for unmeasured factors can often improve confounding control in pharmacoepidemiologic studies utilizing administrative healthcare databases. There is a growing body of evidence showing that data-driven machine learning algorithms for high-dimensional proxy confounder adjustment can supplement investigator-specified variables to improve confounding control compared to adjustment based on investigator-specified variables alone. Consequently, there has been a recent focus on the development of data-driven methods for high-dimensional proxy confounder adjustment. In this paper, we discuss the considerations underpinning three areas for data-driven high-dimensional proxy confounder adjustment: 1) feature generation-transforming raw data into covariates (or features) to be used for proxy adjustment; 2) covariate prioritization, selection and adjustment; and 3) diagnostic assessment. We survey current approaches and recent advancements within each area, including the most widely used approach to proxy confounder adjustment in healthcare database studies (the high-dimensional propensity score or hdPS). We also discuss limitations of the hdPS and outline recent advancements that incorporate the principles of proxy adjustment with machine learning extensions to improve performance. We further discuss challenges and avenues of future development within each area. This manuscript is endorsed by the International Society for Pharmacoepidemiology (ISPE). This article is protected by copyright. All rights reserved.

Introduction to computational causal inference using reproducible Stata, R, and Python code: A tutorial

The main purpose of many medical studies is to estimate the effects of a treatment or exposure on an outcome. However, it is not always possible to randomize the study participants to a particular treatment, therefore observational study designs may be used. There are major challenges with observational studies; one of which is confounding. Controlling for confounding is commonly performed by direct adjustment of measured confounders; although, sometimes this approach is suboptimal due to modeling assumptions and misspecification. Recent advances in the field of causal inference have dealt with confounding by building on classical standardization methods. However, these recent advances have progressed quickly with a relative paucity of computational-oriented applied tutorials contributing to some confusion in the use of these methods among applied researchers. In this tutorial, we show the computational implementation of different causal inference estimators from a historical perspective where new estimators were developed to overcome the limitations of the previous estimators (ie, nonparametric and parametric g-formula, inverse probability weighting, double-robust, and data-adaptive estimators). We illustrate the implementation of different methods using an empirical example from the Connors study based on intensive care medicine, and most importantly, we provide reproducible and commented code in Stata, R, and Python for researchers to adapt in their own observational study. The code can be accessed at https://github.com/migariane/Tutorial_Computational_Causal_Inference_Estimators.

A potential outcomes approach to defining and estimating gestational age-specific exposure effects during pregnancy

Many studies seek to evaluate the effects of potentially harmful pregnancy exposures during specific gestational periods. We consider an observational pregnancy cohort where pregnant individuals can initiate medication usage or become exposed to a drug at various times during their pregnancy. An important statistical challenge involves how to define and estimate exposure effects when pregnancy loss or delivery can occur over time. Without proper consideration, the results of standard analysis may be vulnerable to selection bias, immortal time-bias, and time-dependent confounding. In this study, we apply the “target trials” framework of Hernán and Robins in order to define effects based on the counterfactual approach often used in causal inference. This effect is defined relative to a hypothetical randomized trial of timed pregnancy exposures where delivery may precede and thus potentially interrupt exposure initiation. We describe specific implementations of inverse probability weighting, G-computation, and Targeted Maximum Likelihood Estimation to estimate the effects of interest. We demonstrate the performance of all estimators using simulated data and show that a standard implementation of inverse probability weighting is biased. We then apply our proposed methods to a pharmacoepidemiology study to evaluate the potentially time-dependent effect of exposure to inhaled corticosteroids on birthweight in pregnant people with mild asthma.

Performance Evaluation of Parametric and Nonparametric Methods When Assessing Effect Measure Modification

Effect measure modification is often evaluated using parametric models. These models, although efficient when correctly specified, make strong parametric assumptions. While nonparametric models avoid important functional form assumptions, they often require larger samples to achieve a given accuracy. We conducted a simulation study to evaluate performance tradeoffs between correctly specified parametric and nonparametric models to detect effect modification of a binary exposure by both binary and continuous modifiers. We evaluated generalized linear models and doubly robust (DR) estimators, with and without sample splitting. Continuous modifiers were modeled with cubic splines, fractional polynomials, and nonparametric DR-learner. For binary modifiers, generalized linear models showed the greatest power to detect effect modification, ranging from 0.42 to 1.00 in the worst and best scenario, respectively. Augmented inverse probability weighting had the lowest power, with an increase of 23% when using sample splitting. For continuous modifiers, the DR-learner was comparable to flexible parametric models in capturing quadratic and nonlinear monotonic functions. However, for nonlinear, nonmonotonic functions, the DR-learner had lower integrated bias than splines and fractional polynomials, with values of 141.3, 251.7, and 209.0, respectively. Our findings suggest comparable performance between nonparametric and correctly specified parametric models in evaluating effect modification.

Applying Machine Learning in Distributed Data Networks for Pharmacoepidemiologic and Pharmacovigilance Studies: Opportunities, Challenges, and Considerations

Increasing availability of electronic health databases capturing real-world experiences with medical products has garnered much interest in their use for pharmacoepidemiologic and pharmacovigilance studies. The traditional practice of having numerous groups use single databases to accomplish similar tasks and address common questions about medical products can be made more efficient through well-coordinated multi-database studies, greatly facilitated through distributed data network (DDN) architectures. Access to larger amounts of electronic health data within DDNs has created a growing interest in using data-adaptive machine learning (ML) techniques that can automatically model complex associations in high-dimensional data with minimal human guidance. However, the siloed storage and diverse nature of the databases in DDNs create unique challenges for using ML. In this paper, we discuss opportunities, challenges, and considerations for applying ML in DDNs for pharmacoepidemiologic and pharmacovigilance studies. We first discuss major types of activities performed by DDNs and how ML may be used. Next, we discuss practical data-related factors influencing how DDNs work in practice. We then combine these discussions and jointly consider how opportunities for ML are affected by practical data-related factors for DDNs, leading to several challenges. We present different approaches for addressing these challenges and highlight efforts that real-world DDNs have taken or are currently taking to help mitigate them. Despite these challenges, the time is ripe for the emerging interest to use ML in DDNs, and the utility of these data-adaptive modeling techniques in pharmacoepidemiologic and pharmacovigilance studies will likely continue to increase in the coming years.

AIPW: An R Package for Augmented Inverse Probability-Weighted Estimation of Average Causal Effects

An increasing number of recent studies have suggested that doubly robust estimators with cross-fitting should be used when estimating causal effects with machine learning methods. However, not all existing programs that implement doubly robust estimators support machine learning methods and cross-fitting, or provide estimates on multiplicative scales. To address these needs, we developed AIPW, a software package implementing augmented inverse probability weighting (AIPW) estimation of average causal effects in R (R Foundation for Statistical Computing, Vienna, Austria). Key features of the AIPW package include cross-fitting and flexible covariate adjustment for observational studies and randomized controlled trials (RCTs). In this paper, we use a simulated RCT to illustrate implementation of the AIPW estimator. We also perform a simulation study to evaluate the performance of the AIPW package compared with other doubly robust implementations, including CausalGAM, npcausal, tmle, and tmle3. Our simulation showed that the AIPW package yields performance comparable to that of other programs. Furthermore, we also found that cross-fitting substantively decreases the bias and improves the confidence interval coverage for doubly robust estimators fitted with machine learning algorithms. Our findings suggest that the AIPW package can be a useful tool for estimating average causal effects with machine learning methods in RCTs and observational studies.

Doubly robust nonparametric instrumental variable estimators for survival outcomes

Instrumental variable (IV) methods allow us the opportunity to address unmeasured confounding in causal inference. However, most IV methods are only applicable to discrete or continuous outcomes with very few IV methods for censored survival outcomes. In this article, we propose nonparametric estimators for the local average treatment effect on survival probabilities under both covariate-dependent and outcome-dependent censoring. We provide an efficient influence function-based estimator and a simple estimation procedure when the IV is either binary or continuous. The proposed estimators possess double-robustness properties and can easily incorporate nonparametric estimation using machine learning tools. In simulation studies, we demonstrate the flexibility and double robustness of our proposed estimators under various plausible scenarios. We apply our method to the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial for estimating the causal effect of screening on survival probabilities and investigate the causal contrasts between the two interventions under different censoring assumptions.

Causal inference under interference with prognostic scores for dynamic group therapy studies

Group therapy is a common treatment modality for behavioral health conditions. Patients often enter and exit groups on an ongoing basis, leading to dynamic therapy groups. Examining the effect of high versus low session attendance on patient outcomes is a research question of interest. However, there are several challenges to identifying causal effects in this setting, including the lack of randomization, interference among patients, and the interrelatedness of patient participation. Dynamic therapy groups motivate a unique causal inference scenario, as the treatment statuses are completely defined by the patient attendance record for the therapy session, which is also the structure inducing interference. We adopt the Rubin causal model framework to define the causal effect of high versus low session attendance of group therapy at both the individual patient and peer levels. We propose a strategy to identify individual, peer, and total effects of high attendance versus low attendance on patient outcomes by the prognostic score stratification. We examine performance of our approach via simulation and apply it to data from a group cognitive behavioral therapy trial for treating depression among patients in a substance use disorders treatment setting.

Estimating the Marginal Causal Effect and Potential Impact of Waterpipe Smoking on Risk of Multiple Sclerosis Using the Targeted Maximum Likelihood Estimation Method: A Large, Population-Based Incident Case-Control Study

There are few if any reports regarding the role of lifetime waterpipe smoking in the etiology of multiple sclerosis (MS). In a population-based incident case-control study conducted in Tehran, Iran, we investigated the association between waterpipe smoking and MS, adjusted for confounders. Cases (n = 547) were patients aged 15-50 years identified from the Iranian Multiple Sclerosis Society between 2013 and 2015. Population-based controls (n = 1,057) were persons aged 15-50 years recruited through random digit telephone dialing. A doubly robust estimation method, the targeted maximum likelihood estimator (TMLE), was used to estimate the marginal risk ratio and odds ratio for the association between waterpipe smoking and MS. The estimated risk ratio and odds ratio were both 1.70 (95% confidence interval: 1.34, 2.17). The population attributable fraction was 21.4% (95% confidence interval: 4.0, 38.8). Subject to the limitations of case-control studies in interpreting associations causally, these results suggest that waterpipe use, or strongly related but undetermined factors, increases the risk of MS. Further epidemiologic studies, including nested case-control studies, are needed to confirm these findings.

Robust Inference for Mediated Effects in Partially Linear Models

We consider mediated effects of an exposure, X on an outcome, Y, via a mediator, M, under no unmeasured confounding assumptions in the setting where models for the conditional expectation of the mediator and outcome are partially linear. We propose G-estimators for the direct and indirect effects and demonstrate consistent asymptotic normality for indirect effects when models for the conditional means of M, or X and Y are correctly specified, and for direct effects, when models for the conditional means of Y, or X and M are correct. This marks an improvement, in this particular setting, over previous 'triple' robust methods, which do not assume partially linear mean models. Testing of the no-mediation hypothesis is inherently problematic due to the composite nature of the test (either X has no effect on M or M no effect on Y), leading to low power when both effect sizes are small. We use generalized methods of moments (GMM) results to construct a new score testing framework, which includes as special cases the no-mediation and the no-direct-effect hypotheses. The proposed tests rely on an orthogonal estimation strategy for estimating nuisance parameters. Simulations show that the GMM-based tests perform better in terms of power and small sample performance compared with traditional tests in the partially linear setting, with drastic improvement under model misspecification. New methods are illustrated in a mediation analysis of data from the COPERS trial, a randomized trial investigating the effect of a non-pharmacological intervention of patients suffering from chronic pain. An accompanying R package implementing these methods can be found at github.com/ohines/plmed.

Longitudinal 5-year prediction of cognitive impairment among men with HIV disease

BACKGROUND

Although combination antiretroviral therapy reduced the prevalence of HIV-associated dementia, milder syndromes persist. Our goals were to predict cognitive impairment of the Multicenter AIDS Cohort Study (MACS) participants 5 years ahead and from a large pool of factors, select the ones that mostly contributed to our predictions.

DESIGN

Longitudinal, natural and treated history of HIV infection among MSM.

METHODS

The MACS is a longitudinal study of the natural and treated history of HIV disease in MSM; the neuropsychological substudy aims to characterize cognitive disorders in men with HIV disease.

RESULTS

We modeled on an annual basis the risk of cognitive impairment 5 years in the future. We were able to predict cognitive impairment at individual level with high precision and overperform default methods. We found that while a diagnosis of AIDS is a critical risk factor, HIV infection per se does not necessarily convey additional risk. Other infectious processes, most notably hepatitis B and C, are independently associated with increased risk of impairment. The relative importance of an AIDS diagnosis diminished across calendar time.

CONCLUSION

Our prediction models are a powerful tool to help clinicians address dementia in early stages for MACS paticipants. The strongest predictors of future cognitive impairment included the presence of clinical AIDS and hepatitis B or C infection. The fact that the pattern of predictive power differs by calendar year suggests a clinically critical change to the face of the epidemic.

Causal inference for recurrent event data using pseudo-observations

Recurrent event data are commonly encountered in observational studies where each subject may experience a particular event repeatedly over time. In this article, we aim to compare cumulative rate functions (CRFs) of two groups when treatment assignment may depend on the unbalanced distribution of confounders. Several estimators based on pseudo-observations are proposed to adjust for the confounding effects, namely inverse probability of treatment weighting estimator, regression model-based estimators, and doubly robust estimators. The proposed marginal regression estimator and doubly robust estimators based on pseudo-observations are shown to be consistent and asymptotically normal. A bootstrap approach is proposed for the variance estimation of the proposed estimators. Model diagnostic plots of residuals are presented to assess the goodness-of-fit for the proposed regression models. A family of adjusted two-sample pseudo-score tests is proposed to compare two CRFs. Simulation studies are conducted to assess finite sample performance of the proposed method. The proposed technique is demonstrated through an application to a hospital readmission data set.

Minimal dispersion approximately balancing weights: asymptotic properties and practical considerations

Weighting methods are widely used to adjust for covariates in observational studies, sample surveys, and regression settings. In this paper, we study a class of recently proposed weighting methods, which find the weights of minimum dispersion that approximately balance the covariates. We call these weights ‘minimal weights’ and study them under a common optimization framework. Our key observation is that finding weights which achieve approximate covariate balance is equivalent to performing shrinkage estimation of the inverse propensity score. This connection leads to both theoretical and practical developments. From a theoretical standpoint, we characterize the asymptotic properties of minimal weights and show that, under standard smoothness conditions on the propensity score function, minimal weights are consistent estimates of the true inverse probability weights. In addition, we show that the resulting weighting estimator is consistent, asymptotically normal and semiparametrically efficient. From a practical standpoint, we give a finite-sample oracle inequality that bounds the loss incurred by balancing more functions of the covariates than strictly needed. This inequality shows that minimal weights implicitly bound the number of active covariate balance constraints. Finally, we provide a tuning algorithm for choosing the degree of approximate balance in minimal weights. The paper concludes with an empirical study which suggests that approximate balance is preferable to exact balance, especially when there is limited overlap in covariate distributions. Further studies show that the root mean squared error of the weighting estimator can be reduced by as much as a half with approximate balance.