Quality dashboards: Uncovering potential health disparities.

person Kevin Juan Zhang Beth Israel Deaconess Medical Center, Boston, MA info_outline Kevin Juan Zhang, Aya Sato-DiLorenzo, Melis Celmen, Rosemary Chude-Sokei, Tenzin Dechen, Ashley L. O’Donoghue, Jessica Ann Zerillo

Authors person Kevin Juan Zhang Beth Israel Deaconess Medical Center, Boston, MA info_outline Kevin Juan Zhang, Aya Sato-DiLorenzo, Melis Celmen, Rosemary Chude-Sokei, Tenzin Dechen, Ashley L. O’Donoghue, Jessica Ann Zerillo Organizations Beth Israel Deaconess Medical Center, Boston, MA, BIDMC, Boston, MA Abstract Disclosures Research Funding No funding received None. Background: A number of demographic characteristics and social determinants of health have been demonstrated to impact cancer outcomes. It is essential to assess quality metrics stratified by these characteristics so that potential disparities may be identified and addressed amongst marginalized populations. Methods: We examined our Cancer Center’s quality dashboard’s quality metrics and identified which could be stratified by race, ethnicity, language preference, sexual identity, and gender. Databases included Vizient, Press Ganey, Tumor Registry, and our institution’s performance reports. Where data was available, we conducted cross-group comparisons to assess for differences in outcomes. ANOVA and Fisher’s exact tests were performed to compare groups. A P-value of < 0.05 was considered statistically significant. Results: Of the 13 metrics, six (46%) could be stratified by demographic characteristics (likelihood to recommend the practice, mortality index, length of stay index, 30-day readmission, return to the operating room, adequate lymph node sampling). Six (46%) could not be stratified due to a lack of demographic information (new patient access, medication harm events, distress screening, preventable admissions, preventable emergency department visits, survivorship care plan). One metric was not at the patient-level, and thus did not apply to our investigation (non-template chemotherapy orders). More White patients (86.1%) were very likely to recommend the practice compared to Black (72.1%) or Asian patients (60.1%) (P < 0.001). Additionally, significantly more English speakers were likely to recommend the practice than non-English speakers (84% vs. 65%; P < 0.001). A significant difference was found in the 30-day readmission rate when stratified by race (White [8.2%] vs. Black [8.8%] vs. Asian [10.6%] vs. other [14.5%] vs. unknown [1.0%]; P = 0.02). No significant group differences were found in mortality index, length of stay, the proportion of adequate lymph node sampling, or returning to the operating room. Notably, the proportion of patients with “unknown” race and ethnicity ranging from 3.9%-17.5% for five metrics, represented the second largest group behind White race and non-Hispanic ethnicity. Conclusions: Less than half of the quality dashboard’s core metrics could be stratified by demographic characteristics. Even when data was available, the high frequency of missing data identifies a need for improved data collection accuracy so that more meaningful assessments can be made. Existing data suggests a disparity exists for patients of limited English proficiency having a lower satisfaction with the practice. This warrants further exploration in order to close gaps in cancer quality outcomes.