Luke Keele

While balancing covariates between groups is central for observational causal inference, selecting which features to balance remains a challenging problem. Kernel balancing is a promising approach that first estimates a kernel that captures similarity across units and then balances a (possibly low‐dimensional) summary of that kernel, indirectly learning important features to balance. In this paper, we propose forest kernel balancing , which leverages the underappreciated fact that tree‐based machine learning models, namely random forests and Bayesian additive regression trees (BART), implicitly estimate a kernel based on the co‐occurrence of observations in the same terminal leaf node. Thus, even though the resulting kernel is solely a function of baseline features, the selected nonlinearities and other interactions are important for predicting the outcome—and therefore are important for addressing confounding. Through simulations and applied illustrations, we show that forest kernel balancing leads to meaningful computational and statistical improvement relative to standard kernel methods, which do not incorporate outcome information when learning features.

In a clustered observational study, treatment is assigned to groups and all units within the group are exposed to the treatment. Here, we use a clustered observational study (COS) design to estimate the effectiveness of Magnet Nursing certificates for emergency surgery patients. Recent research has introduced specialized weighting estimators for the COS design that balance baseline covariates at the unit and cluster level. These methods allow researchers to adjust for observed confounders but are sensitive to unobserved confounding. In this paper, we develop new sensitivity analysis methods tailored to weighting estimators for COS designs. We provide several key contributions. First, we introduce a key bias decomposition, tailored to the specific confounding structure that arises in a COS. Second, we develop a sensitivity framework for weighted COS designs that constrain the error in the underlying weights. We introduce both a marginal sensitivity model and a variance-based sensitivity model, and extend both to accommodate multiple estimands. Finally, we propose amplification and benchmarking methods to better interpret the results. Throughout, we illustrate our proposed methods by analysing the effectiveness of Magnet nursing hospitals.

When units in observational studies are clustered in groups, such as students in schools or patients in hospitals, researchers often address confounding by adjusting for cluster-level covariates or cluster membership. In this paper, we develop a unified weighting framework that clarifies how different estimation methods control two distinct sources of imbalance: global balance (differences between treated and control units across clusters) and local balance (differences within clusters). We show that inverse propensity score weighting (IPW) with a random effects propensity score model -- the current standard in the literature -- targets only global balance and constant level shifts across clusters, but imposes no constraints on local balance. We then present two approaches that target both forms of balance. First, hierarchical balancing weights directly control global and local balance through a constrained optimization problem. Second, building on the recently proposed Generalized Mundlak approach, we develop a novel Mundlak balancing weights estimator that adjusts for cluster-level sufficient statistics rather than cluster indicators; this approach can accommodate small clusters where all units are treated or untreated. Critically, these approaches rest on different assumptions: hierarchical balancing weights require only that treatment is ignorable given covariates and cluster membership, while Mundlak methods additionally require an exponential family structure. We then compare these methods in a simulation study and in two applications in education and health services research that exhibit very different cluster structures.

Background: Nurse managers are central to hospital operations, yet little is known about how organizational factors, such as Magnet Recognition by the American Nurses Credentialing Center (ANCC), shape their assessments of the work environment, care quality, and safety. This study examined whether Magnet status is associated with nurse managers’ assessments of these domains. Methods: This cross‐sectional study used data from the 2024 Penn Nurses4All Survey, the 2023 American Hospital Association Annual Survey, and the ANCC’s list of Magnet‐recognized organizations. The sample included 1362 nurse managers from 771 hospitals across 10 U.S. states (450 in 186 Magnet hospitals and 912 in 585 non‐Magnet hospitals). Outcomes included managers’ assessments of the work environment, nursing care quality, patient safety, and hospital recommendation. Approximate balancing weights were applied to adjust for hospital‐ and manager‐level covariates, and weighted linear probability models estimated average treatment effects. Results: Magnet managers were more likely to hold graduate degrees and to work in large, teaching, and not‐for‐profit hospitals. Compared with non‐Magnet managers, they were 9 percentage points more likely to rate the work environment as excellent or good ( b = 0.09, 95% CI: 0.04–0.16) and to “definitely” recommend their hospital as a workplace ( b = 0.09, 95% CI: 0.02–0.16); 7 percentage points more likely to rate nursing care quality as excellent or good ( b = 0.07, 95% CI: 0.04–0.14); 12 percentage points more likely to assign an excellent or good patient safety grade ( b = 0.12, 95% CI: 0.06–0.19); and 16 percentage points more likely to recommend their hospital to family and friends ( b = 0.16, 95% CI: 0.08–0.24). Conclusion: Magnet status was associated with more favorable assessments of the work environment, quality, and safety among nurse managers. These findings suggest that Magnet structures may strengthen organizational environment and culture across multiple levels of the nursing workforce.

While balancing covariates between groups is central for observational causal inference, selecting which features to balance remains a challenging problem. Kernel balancing is a promising approach that first estimates a kernel that captures similarity across units and then balances a (possibly low-dimensional) summary of that kernel, indirectly learning important features to balance. In this paper, we propose forest kernel balancing, which leverages the underappreciated fact that tree-based machine learning models, namely random forests and Bayesian additive regression trees (BART), implicitly estimate a kernel based on the co-occurrence of observations in the same terminal leaf node. Thus, even though the resulting kernel is solely a function of baseline features, the selected nonlinearities and other interactions are important for predicting the outcome -- and therefore are important for addressing confounding. Through simulations and applied illustrations, we show that forest kernel balancing leads to meaningful computational and statistical improvement relative to standard kernel methods, which do not incorporate outcome information when learning features.

Key Points: Among patients with kidney failure treated with dialysis, 37% underwent parathyroidectomy in the outpatient setting. Significant differences exist between patients selected for inpatient versus outpatient parathyroidectomy. Risk-adjusted readmission rates are similar after inpatient or outpatient parathyroidectomy for secondary hyperparathyroidism. Background: Outpatient parathyroidectomy for secondary hyperparathyroidism of kidney disease is increasing. Reduced length of stay is often associated with higher rates of readmission. The safety of outpatient parathyroidectomy in this population must be evaluated given the morbidity and higher risk of post-parathyroidectomy hungry bone syndrome associated with kidney failure. In this study, we compare parathyroidectomy outcomes and costs by admission status in dialysis-dependent patients with secondary hyperparathyroidism. Methods: Adult patients on dialysis who underwent elective parathyroidectomy for secondary hyperparathyroidism were identified in the Healthcare Cost and Utilization Project State Inpatient and State Ambulatory Surgery Databases from five geographically diverse states (2013–2019). Patients were classified by admission status: inpatient and outpatient (overnight or same-day). The primary outcome was 30-day readmission. Secondary outcomes were parathyroidectomy-related complications and cost. Outcomes were compared by admission status using chi-squared tests and costs using median quantile regression. Balancing weights were used for adjustment. Forward selection with logistic regression was used to identify factors associated with readmission among outpatients. Results: Among 1267 study participants, 63% were inpatients. Among outpatients, 40% underwent overnight stay and 42% underwent same-day discharge. Inpatients were more likely to be younger, Black race, have nonprivate insurance, and increased comorbidities than outpatients ( P < 0.05). Overall readmission rate was 19%. Readmission was higher among inpatients (Inpatient: 22% versus Outpatient: 15%; P = 0.002). After balancing weights, adjusted 30-day readmission rate was similar between inpatients and outpatients (14% versus 15%; P = 0.79). Readmission was similar between overnight stay and same-day discharge (Unadjusted: 12% versus 15%; P = 0.47. Adjusted: 15% versus 15%; P = 0.97). Outpatients with parathyroid autotransplantation, anxiety disorder, autoimmune conditions, congestive heart failure, and moderate/severe liver failure had significantly increased odds of readmission after parathyroidectomy. Adjusted median costs varied significantly (Adjusted Inpatient: $9615 [Interquartile range (IQR), $6463–12,963] versus Outpatient: $6357 [IQR, $4966–8153]; P < 0.001; Adjusted Overnight: $6453 [IQR, $5551–7850] versus Same-day: $5526 [IQR, $4346–7187]; P < 0.001). Conclusions: Select patients with secondary hyperparathyroidism may undergo outpatient parathyroidectomy safely with reduced costs.

When an exposure of interest is confounded by unmeasured factors, an instrumental variable (IV) can be used to identify and estimate certain causal contrasts. Identification of the marginal average treatment effect (ATE) from IVs relies on strong untestable structural assumptions. When one is unwilling to assert such structure, IVs can nonetheless be used to construct bounds on the ATE. Famously, Alexander Balke and Judea Pearl proved tight bounds on the ATE for a binary outcome, in a randomized trial with noncompliance and no covariate information. We demonstrate how these bounds remain useful in observational settings with baseline confounders of the IV, as well as randomized trials with measured baseline covariates. The resulting bounds on the ATE are nonsmooth functionals, and thus standard nonparametric efficiency theory is not immediately applicable. To remedy this, we propose (1) under a novel margin condition, influence function-based estimators of the bounds that can attain parametric convergence rates when the nuisance functions are modelled flexibly, and (2) estimators of smooth approximations of these bounds. We propose extensions to continuous outcomes, explore finite sample properties in simulations, and illustrate the proposed estimators in an observational study targeting the effect of higher education on wages.

In a clustered observational study, treatment is assigned to groups and all units within the group are exposed to the treatment. Here, we use a clustered observational study (COS) design to estimate the effectiveness of Magnet Nursing certificates for emergency surgery patients. Recent research has introduced specialized weighting estimators for the COS design that balance baseline covariates at the unit and cluster level. These methods allow researchers to adjust for observed confounders, but are sensitive to unobserved confounding. In this paper, we develop new sensitivity analysis methods tailored to weighting estimators for COS designs. We provide several key contributions. First, we introduce a key bias decomposition, tailored to the specific confounding structure that arises in a COS. Second, we develop a sensitivity framework for weighted COS designs that constrain the error in the underlying weights. We introduce both a marginal sensitivity model and a variance-based sensitivity model, and extend both to accommodate multiple estimands. Finally, we propose amplification and benchmarking methods to better interpret the results. Throughout, we illustrate our proposed methods by analyzing the effectiveness of Magnet nursing hospitals.