Search

BackgroundPatient surveys are the primary tool to measure patient experiences of care. Caution must be taken when analyzing these data, as responses can be influenced by factors that do not reflect the quality of care received.ObjectivesTo provide a practical overview of adjusting patient experience survey results to address bias related to patient case-mix, extreme response tendency, and mode of survey administration.Research designWe discuss options for adjustment for biases in how people respond to patient experience surveys.ResultsCase-mix adjustment (CMA) aims to compare provider performance that would have been observed if all providers had treated the same set of patients by removing the effects of patient characteristics that vary across providers. Extreme response tendency can bias the measurement of the disparities in patient experiences even after typical CMAs, since differences in patients' use of extreme response options may affect patient experience scores when they have a skewed distribution. Survey mode may affect scores for the provider entity being evaluated (eg, hospital) more than CMA if survey mode differs at the provider level.ConclusionsIt is best practice to evaluate known source of bias when analyzing patient experience surveys. Failure to adjust for patient case-mix, extreme response tendency, and survey mode in patient experience surveys may lead to erroneous comparisons of providers.

BackgroundSince early 2016, an increasing number of states passed legislations that limit the duration and/or dosage of initial opioid prescriptions or opioids for acute pain.ObjectiveThe objective of this study was to assess changes in the number of opioid prescriptions covered by Medicaid and received by Medicaid patients associated with state implementation of legislative limits on initial opioid prescriptions.Research designWe explored the natural experiment resulting from the staggered implementation of state legislative limits. The analysis adopted a Difference-in-Differences framework and controlled for other major state policies bearing implications for prescription opioid use. The main analysis included 26 states that implemented limits from early 2016 to late 2018. A secondary analysis included all 50 states and the District of Columbia.MeasuresPopulation-adjusted state-quarter level counts of Schedule II and III opioid prescriptions received by Medicaid patients, based on data from the Medicaid State Drug Utilization Data and state Medicaid enrollment reports for 2013-2018.ResultsImplementation of legislative limits on initial opioid prescriptions was associated with a 7% reduction in the number of opioid prescriptions per 100 Medicaid enrollees. Such reduction was largely attributable to a reduction in Schedule II opioid prescriptions. Secondary analysis by including all jurisdictions and sensitivity checks supported the robustness of results.ConclusionThe recent implementation of state legislative limits on initial opioid prescriptions was associated with meaningful reductions in the volume of Schedule II opioid prescriptions received by Medicaid patients.

BackgroundElectronic health records provide the opportunity to assess system-wide quality measures. Veterans Affairs Pharmacy Benefits Management Center for Medication Safety uses medication use evaluation (MUE) through manual review of the electronic health records.ObjectiveTo compare an electronic MUE approach versus human/manual review for extraction of antibiotic use (choice and duration) and severity metrics.Research designRetrospective.SubjectsHospitalizations for uncomplicated pneumonia occurring during 2013 at 30 Veterans Affairs facilities.MeasuresWe compared summary statistics, individual hospitalization-level agreement, facility-level consistency, and patterns of variation between electronic and manual MUE for initial severity, antibiotic choice, daily clinical stability, and antibiotic duration.ResultsAmong 2004 hospitalizations, electronic and manual abstraction methods showed high individual hospitalization-level agreement for initial severity measures (agreement=86%-98%, κ=0.5-0.82), antibiotic choice (agreement=89%-100%, κ=0.70-0.94), and facility-level consistency for empiric antibiotic choice (anti-MRSA r=0.97, P

BackgroundThe belief that there is inefficiency, or the potential to improve patient health at current levels of spending, is driving the push for greater value in health care. Previous studies demonstrate overuse of a narrow set of services, suggesting provider inefficiency, but existing studies neither quantify inefficiency more broadly nor assess its variation across physician organizations (POs).Data and methodsWe used data on quality of care and total cost of care from 129 California POs participating in a statewide value-based pay-for-performance program. We estimated a production function with quality as the output and cost as the input, using a stochastic frontier model, to develop a measure of relative efficiency for each PO. To validate the efficiency measure, we examined correlations of PO efficiency estimates with indicators representing overuse of services.ResultsThe estimated production function showed that PO quality was positively associated with costs, although there were diminishing marginal returns to spending. A certain minimum level of spending was associated with high quality even among efficient POs. Most strikingly, however, POs had substantial variation in efficiency, producing widely differing levels of quality for the same cost.ConclusionsDifferences among POs in the efficiency with which they produce quality suggest opportunities for improvements in care delivery that increase quality without increasing spending.

BackgroundLittle is known about how often contextual factors such as patient preferences and competing priorities impact prescribing of guideline-recommended medications, or about the extent to which these factors are documented in medical records and available to performance measurement systems.MethodsMixed-methods study of 295 veterans aged 50 years and older in 4 VA health care systems who had systolic heart failure and were not prescribed a β-blocker and/or an angiotensin converting enzyme inhibitor or angiotensin-receptor blocker. Reasons for nontreatment were identified from clinic notes and from interviews with 62 primary care clinicians caring for these patients. These reasons were classified using a published taxonomy.ResultsAmong 295 patients not receiving guideline-recommended drugs for heart failure, chart review identified biomedical reasons for nonprescribing in 42%-58% of patients and contextual reasons in 11%-17%. Clinician interviews identified twice as many reasons for nonprescribing as chart review (mean 1.6 vs. 0.8 reasons per patient, P