Bayesian variable selection in hierarchical difference-in-differences models

A popular method for estimating a causal treatment effect with observational data is the difference-in-differences model. In this work, we consider an extension of the classical difference-in-differences setting to the hierarchical context in which data cannot be matched at the most granular level. Our motivating example is an application to assess the impact of primary care redesign policy on diabetes outcomes in Minnesota, in which the policy is administered at the clinic level and individual outcomes are not matched from pre- to post-intervention. We propose a Bayesian hierarchical difference-in-differences model, which estimates the policy effect by regressing the treatment on a latent variable representing the mean change in group-level outcome. We present theoretical and empirical results showing a hierarchical difference-in-differences model that fails to adjust for a particular class of confounding variables, biases the policy effect estimate. Using a structured Bayesian spike-and-slab model that leverages the temporal structure of the difference-in-differences context, we propose and implement variable selection approaches that target sets of confounding variables leading to unbiased and efficient estimation of the policy effect. We evaluate the methods' properties through simulation, and we use them to assess the impact of primary care redesign of clinics in Minnesota on the management of diabetes outcomes from 2008 to 2017.

Pulmonary ultrasound scoring system for intubated critically ill patients and its association with clinical metrics and mortality: A prospective cohort study

PURPOSE

Pulmonary ultrasound (PU) examination at the point-of-care can rapidly identify the etiology of acute respiratory failure (ARF) and assess treatment response. The often-subjective classification of PU abnormalities makes it difficult to document change over time and communicate findings across providers. The study goal was to develop a simple, PU scoring system that would allow for standardized documentation, have high interprovider agreement, and correlate with clinical metrics.

METHODS

In this prospective study of 250 adults intubated for ARF, a PU examination was performed at intubation, 48-hours later, and at extubation. A total lung score (TLS) was calculated. Clinical metrics and final diagnosis were extracted from the medical record.

RESULTS

TLS correlated positively with mortality (P = .03), ventilator hours (P = .003), intensive care unit, and hospital length of stay (P = .003, P = .008), and decreasing PaO₂ /FiO₂ (P < .001). Agreement of PU findings was very good (kappa = 0.83). Baseline TLS and subscores differed significantly between ARF categories (nonpulmonary, obstructive, and parenchymal disease).

CONCLUSIONS

A quick, scored, PU examination was associated with clinical metrics, including mortality among a diverse population of patients intubated for ARF. In addition to diagnostic and prognostic information at the bedside, a standardized and quantifiable approach to PU provides objectivity in serial assessment and may enhance communication of findings between providers.

Postoperative respiratory failure necessitating transfer to the intensive care unit in orthopedic surgery patients: risk factors, costs, and outcomes

Background

Postoperative pulmonary complications in orthopedic surgery patients have been associated with worse clinical outcomes. Identifying patients with respiratory risk factors requiring enhanced monitoring and management modifications is an important part of postoperative care. Patients with unanticipated respiratory decompensation requiring transfer to the intensive care unit (ICU) have not been studied in sufficient detail.

Methods

A retrospective case-control study of elective orthopedic surgery patients (knee, hip, shoulder, or spine, n = 51) who developed unanticipated respiratory failure (RF) necessitating transfer to the ICU over a 3-year period was conducted. Controls (n = 153) were frequency matched to cases by gender, age, and surgical procedure. Patient and perioperative care factors, clinical outcomes, and cost of care were examined.

Results

Transfer to the ICU occurred within 48 h of surgery in 73 % of the cases, 31 % required non-invasive ventilation, and 18 % required mechanical ventilation. Cases had a higher prevalence of chronic obstructive pulmonary disease (COPD), obstructive sleep apnea (OSA), and regular psychotropic medication use than controls. Cases received more intravenous opioids during the first 24 postoperative hours, were hospitalized 4 days longer, had higher in-hospital mortality, and had excess hospitalization costs of US$26,571. COPD, OSA, preoperative psychotropic medications, and anesthesia time were associated with risk of RF in a multivariate analysis.

Conclusions

Unanticipated RF after orthopedic surgery is associated with extended hospitalization, increased mortality, and higher cost of care. Hospital protocols that include risk factor assessment, enhanced monitoring, and a cautious approach to opioid use in high-risk patients may reduce the frequency of this complication.