Mary E. Slaughter’s research while affiliated with RAND Corporation and other places

Background Researchers have many options for web-based survey data collection, ranging from access to curated probability-based panels, where individuals are selectively invited to join based on their membership in a representative population, to convenience panels, which are open for anyone to join. The mix of respondents available also varies greatly regarding representation of a population of interest and in motivation to provide thoughtful and accurate responses. Despite the additional dataset-building labor required of the researcher, convenience panels are much less expensive than probability-based panels. However, it is important to understand what may be given up regarding data quality for those cost savings. Objective This study examined the relative costs and data quality of fielding equivalent surveys on Amazon’s Mechanical Turk (MTurk), a convenience panel, and KnowledgePanel, a nationally representative probability-based panel. Methods We administered the same survey measures to MTurk (in 2021) and KnowledgePanel (in 2022) members. We applied several recommended quality assurance steps to enhance the data quality achieved using MTurk. Ipsos, the owner of KnowledgePanel, followed their usual (industry standard) protocols. The survey was designed to support psychometric analyses and included >60 items from the Patient-Reported Outcomes Measurement Information System (PROMIS), demographics, and a list of health conditions. We used 2 fake conditions (“syndomitis” and “chekalism”) to identify those more likely to be honest respondents. We examined the quality of each platform’s data using several recommended metrics (eg, consistency, reliability, representativeness, missing data, and correlations) including and excluding those respondents who had endorsed a fake condition and examined the impact of weighting on representativeness. Results We found that prescreening in the MTurk sample (removing those who endorsed a fake health condition) improved data quality but KnowledgePanel data quality generally remained superior. While MTurk’s unweighted point estimates for demographics exhibited the usual mismatch with national averages (younger, better educated, and lower income), weighted MTurk data matched national estimates. KnowledgePanel’s point estimates better matched national benchmarks even before poststratification weighting. Correlations between PROMIS measures and age and income were similar in MTurk and KnowledgePanel; the mean absolute value of the difference between each platform’s 137 correlations was 0.06, and 92% were <0.15. However, correlations between PROMIS measures and educational level were dramatically different; the mean absolute value of the difference across these 17 correlation pairs was 0.15, the largest difference was 0.29, and the direction of more than half of these relationships in the MTurk sample was the opposite from that expected from theory. Therefore, caution is needed if using MTurk for studies where educational level is a key variable. Conclusions The data quality of our MTurk sample was often inferior to that of the KnowledgePanel sample but possibly not so much as to negate the benefits of its cost savings for some uses. International Registered Report Identifier (IRRID) RR2-10.1186/s12891-020-03696-2

Background. Healthcare provided by bilingual providers or with assistance from qualified interpreters is intended to improve patient-provider communication. Despite federal laws requiring healthcare facilities to provide access to appropriate interpretation language assistance services for patients not proficient in English, many Spanish-preferring patients receive primary care from providers not fluent in Spanish or who regularly use formal interpreters. Methods. Partnering with two urban SafetyNet providers in Southern California, we conducted focus groups in Spanish with Spanish-preferring patients who received care from providers who: 1) were Spanish-qualified, 2) used formal interpreters, and 3) used informal interpreters or other communication strategies. We coded transcripts to identify themes and compared patient experiences across provider types. Subjects. 62 adult Spanish-preferring primary care patients. Results. Spanish-preferring patients reported preference for continuity with their English-speaking providers despite language barriers because of established rapport. Patients receiving care from Spanish-qualified providers reported greater trust, more comprehensive care (i.e., covered more issues with minimal detail), yet with many interactions rushed. Formal interpreters facilitated better understanding and professional communication, however, impersonalized patient-provider interactions. Informal interpreters or ad-hoc strategies led to mixed experiences, often dependent on patient or provider ability to accurately convey medical information. Conclusion. Spanish-preferring patient experiences highlighted the necessity for healthcare systems to support robust language and interpretation services that enhance direct communication, ensure interpreter quality, and maintain long-term patient-provider relationships. Improvements in policy and practice are needed to optimize healthcare communication for Spanish-preferring patients, since patient-provider communication is critical for high-quality health outcomes and experiences in multilingual settings.

BACKGROUND Researchers have many options for web-based survey data collection, ranging from access to curated probability-based panels, where individuals are selectively invited to join based on their membership in a representative population, to convenience panels, which are open for anyone to join. The mix of respondents available also varies greatly regarding representation of a population of interest and in motivation to provide thoughtful and accurate responses. Despite the additional dataset-building labor required of the researcher, convenience panels are much less expensive than probability-based panels. However, it is important to understand what may be given up regarding data quality for those cost savings. OBJECTIVE This study examined the relative costs and data quality of fielding equivalent surveys on Amazon’s Mechanical Turk (MTurk), a convenience panel, and KnowledgePanel, a nationally representative probability-based panel. METHODS We administered the same survey measures to MTurk (in 2021) and KnowledgePanel (in 2022) members. We applied several recommended quality assurance steps to enhance the data quality achieved using MTurk. Ipsos, the owner of KnowledgePanel, followed their usual (industry standard) protocols. The survey was designed to support psychometric analyses and included >60 items from the Patient-Reported Outcomes Measurement Information System (PROMIS), demographics, and a list of health conditions. We used 2 fake conditions (“syndomitis” and “chekalism”) to identify those more likely to be honest respondents. We examined the quality of each platform’s data using several recommended metrics (eg, consistency, reliability, representativeness, missing data, and correlations) including and excluding those respondents who had endorsed a fake condition and examined the impact of weighting on representativeness. RESULTS We found that prescreening in the MTurk sample (removing those who endorsed a fake health condition) improved data quality but KnowledgePanel data quality generally remained superior. While MTurk’s unweighted point estimates for demographics exhibited the usual mismatch with national averages (younger, better educated, and lower income), weighted MTurk data matched national estimates. KnowledgePanel’s point estimates better matched national benchmarks even before poststratification weighting. Correlations between PROMIS measures and age and income were similar in MTurk and KnowledgePanel; the mean absolute value of the difference between each platform’s 137 correlations was 0.06, and 92% were <0.15. However, correlations between PROMIS measures and educational level were dramatically different; the mean absolute value of the difference across these 17 correlation pairs was 0.15, the largest difference was 0.29, and the direction of more than half of these relationships in the MTurk sample was the opposite from that expected from theory. Therefore, caution is needed if using MTurk for studies where educational level is a key variable. CONCLUSIONS The data quality of our MTurk sample was often inferior to that of the KnowledgePanel sample but possibly not so much as to negate the benefits of its cost savings for some uses. INTERNATIONAL REGISTERED REPORT RR2-10.1186/s12891-020-03696-2

Purpose The Patient-Reported Outcomes Measurement Information System® (PROMIS)-16 assesses the same multi-item domains but does not include the pain intensity item in the PROMIS-29. We evaluate how well physical and mental health summary scores estimated from the PROMIS-16 reproduce those estimated using the PROMIS-29. Methods An evaluation of data collected from 4130 respondents from the KnowledgePanel. Analyses include confirmatory factor analysis to assess physical and mental health latent variables based on PROMIS-16 scores, reliability estimates for the PROMIS measures, mean differences and correlations of scores estimated by the PROMIS-16 with those estimated by the PROMIS-29, and associations between differences in corresponding PROMIS-16 and PROMIS-29 scores by sociodemographic characteristics. Results A two-factor (physical and mental health) model adequately fits the PROMIS-16 scores. Reliability estimates for the PROMIS-16 measures were slightly lower than for the PROMIS-29 measures. There were minimal differences between PROMIS physical and mental health summary scores estimated using the PROMIS-16 or the PROMIS-29. PROMIS-16 and PROMIS-29 score differences by sociodemographic characteristics were small. Using the PROMIS pain intensity item when scoring the PROMIS-16 produced similar estimates of physical and mental health summary scores. Conclusion The PROMIS-16 provides similar estimates of the PROMIS-29 physical and mental health summary scores. The high reliability of these scores indicates they are accurate enough for use with individual patients.

Purpose Adults’ comments on patient experience surveys explain variation in provider ratings, with negative comments providing more actionable information than positive comments. We investigate if narrative comments on the Consumer Assessment of Healthcare Providers and Systems (CAHPS®) survey of inpatient pediatric care (Child HCAHPS) account for global perceptions of the hospital beyond that explained by reports about specific aspects of care. Methods We analyzed 545 comments from 927 Child HCAHPS surveys completed by parents and guardians of hospitalized children with at least a 24-h hospital stay from July 2017 to December 2020 at an urban children’s hospital. Comments were coded for valence (positive/negative/mixed) and actionability and used to predict Overall Hospital Rating and Willingness to Recommend the Hospital along with Child HCAHPS composite scores. Results Comments were provided more often by White and more educated respondents. Negative comments and greater actionability of comments were significantly associated with Child HCAHPS global rating measures, controlling for responses to closed-ended questions, and child and respondent characteristics. Each explained an additional 8% of the variance in respondents’ overall hospital ratings and an additional 5% in their willingness to recommend the hospital. Conclusions Child HCAHPS narrative comment data provide significant additional information about what is important to parents and guardians during inpatient pediatric care beyond closed-ended composites. Practice implications Quality improvement efforts should include a review of narrative comments alongside closed-ended responses to help identify ways to improve inpatient care experiences. To promote health equity, comments should be encouraged for racial-and-ethnic minority patients and those with less educational attainment.

Background Burnout among providers negatively impacts patient care experiences and safety. Providers at Federally Qualified Health Centers (FQHC) are at high risk for burnout due to high patient volumes; inadequate staffing; and balancing the demands of patients, families, and team members. Objective Examine associations of provider burnout with their perspectives on quality improvement (QI), patient experience measurement, clinic culture, and job satisfaction. Design We conducted a cross-sectional provider survey about their perspectives including the single-item burnout measure. We fit separate regression models, controlling for provider type, gender, being multilingual, and fixed effects for clinic predicting outcome measures from burnout. Participants Seventy-four providers from 44 clinics in large, urban FQHC (52% response rate; n = 174). Main Measures Survey included a single-item, self-defined burnout measure adapted from the Physician Worklife Survey, and measures from the RAND AMA Study survey, Heath Tracking Physician survey, TransforMed Clinician and Staff Questionnaire, Physician Worklife Survey, Minimizing Errors Maximizing Outcomes survey, and surveys by Friedberg et al. ³¹ and Walling et al. ³² Results Thirty percent of providers reported burnout. Providers in clinics with more facilitative leadership reported not being burned out (compared to those reporting burnout; p -values < 0.05). More pressures related to patient care and lower job satisfaction were associated with burnout ( p -values < 0.05). Conclusions Creating provider-team relationships and environments where providers have the time and space necessary to discuss changes to improve care, ideas are shared, leadership supports QI, and QI is monitored and discussed were related to not being burned out. Reducing time pressures and improving support needed for providers to address the high-need levels of FQHC patients can also decrease burnout. Such leadership and support to improving care may be a separate protective factor against burnout. Research is needed to further examine which aspects of leadership drive down burnout and increase provider involvement in change efforts and improving care.

Background In 2014, the National Institute of Health Pain Consortium’s research task force (RTF) on research standards for chronic low back pain (CLBP) proposed the Impact Stratification Score (ISS) as a patient-reported outcome measure that could stratify patients by the impact CLBP has on their lives. This work compares three newly developed ISS-based classifications to the RTF’s original to provide an optimal recommendation. Methods The online sample included 1226 individuals from Amazon’s Mechanical Turk who indicated having non-specific CLBP, average age of 40, 49% female, and 67% White. Participants completed the PROMIS-29 v2.1 profile survey that contains the 9 ISS items as well the Roland-Morris Disability Questionnaire (RMDQ) and Graded Chronic Pain Scale (GCPS). Other items included high-impact chronic pain; not working due to health problems; overall health; and number of healthcare visits for back pain in the past 6 months. Three new classifications were created using quartiles (Classification 2), latent profile analysis (Classification 3), and one modeled after the GCPS (Classification 4). Classifications were subsequently compared to the RTF-proposed classification (Classification 1) on several concurrent and prognostic criteria. Results Classification 1 had three CLBP severity groups, four in Classification 2, three in Classification 3, and four in Classification 4. All novel classifications improved upon the original. Classification 2 performed best at minimizing the classification of those with negative outcomes into the lowest severity groups at baseline (e.g., 11% with RMDQ ≥ 7) and 6 months (e.g., 8.2% had fair/poor health). Classification 4 performed best at maximizing classification of those with negative outcomes into the most severe group concurrently (e.g., 100% had GCPS grade ≥ 2) and at 6 months (e.g., 100% with RMDQ ≥ 7). Conclusions We developed three ISS-based classification schemes and tested them against several outcomes. All three improved upon the original scheme. While appearing more optimal than other classifications in the lowest severity groups, Classification 2 presents some considerations and limitations. Given that Classification 4 was an improvement at the lowest end of severity and was the best at the highest end, it is our tentative recommendation that this approach be adopted to classify individuals with non-specific CLBP.

Background Shadow coaching, a type of one-on-one provider counseling by trained peers, is an effective strategy for improving provider behaviors and patient interactions, but its effects on improving patient experience for English- and Spanish-preferring patients is unknown. Objective Assess effects of shadow coaching on patient experience for English- and for Spanish-preferring patients. Design We analyzed 2012–2019 Clinician and Group Consumer Assessment of Healthcare Providers and Systems (CG-CAHPS) data ( n =46,089) from an urban Federally Qualified Health Center with 44 primary care practices and 320 providers. One-third ( n =14,631) were Spanish-preferring patients. We fit mixed-effects regression models with random effects for provider (the level of treatment assignment) and fixed effects for time (a linear spline for time with a knot and “jump” at coaching date), patient characteristics, and site indicators, stratified by preferred language. Participants The 74 providers who had a 6-month average top-box score on the CAHPS overall provider rating below 90 (on a 100-point scale) were shadow coached. Similar percentages of English-preferring (45%) and Spanish-preferring patients (43%) were seen by coached providers. Intervention Trained providers observed patient care by colleagues and provided suggestions for improvement. Verbal feedback was provided immediately after the observation and the participant received a written report summarizing the comments and recommendations from the coaching session. Main Measures CG-CAHPS Visit Survey 2.0 provider communication composite and overall provider rating (0–100 scoring). Key Results We found a statistically significant 2-point (small) jump in CAHPS provider communication and overall provider rating among English-preferring patients of coached providers. There was no evidence of a coaching effect on patient experience for Spanish-preferring patients. Conclusions Coaching improved care experiences for English-preferring patients but may not have improved patient experience for Spanish-preferring patients. Selection and training of providers to communicate effectively with Spanish-preferring patients is needed to extend the benefits of shadow coaching to Spanish-preferring patients.