If you’re a Blue Cross Blue Shield of Massachusetts member, you can help improve the accuracy of these data right now, by updating your race, ethnicity, and language preferences here.
Race/ethnicity data: The 2023 member race and ethnicity data underlying Blue Cross’ analyses (race/ethnicity data version 5) were a mix of self-reported data (approximately 28% of Blue Cross Blue Shield of Massachusetts members included in this report) and imputed data, which is a commonly used approach when self-reported data are incomplete. In data version 5, race and ethnicity were imputed using a multinomial implementation of Bayesian Additive Regression Trees (BART). This approach flexibly combines data from the RAND Bayesian Improved First Name, Surname, and Geocoding (BIFSG) method, the Massachusetts Immunization Information System (MIIS), and other member-level information. Relative to traditional multinomial logistic regression-based approaches, BART improves imputation accuracy by more flexibly learning the shapes of relationships (e.g. linear, quadratic, logarithmic) and the interactions between data sources. More information about the RAND BIFSG method can be requested from: https://www.rand.org/health-care/tools-methods/bisg.html. More information about the MIIS data can be requested from: https://www.mass.gov/massachusetts-immunization-information-system-miis.
Analyses that include imputed data might overestimate or underestimate the true magnitude of inequities (i.e., the magnitude of inequities that would be calculated if 100% self-reported race and ethnicity data were available). For this reason, Blue Cross is currently engaged in a major effort to collect self-reported race and ethnicity data from members directly. Future versions of these analyses will incorporate more member self-reported race and ethnicity data as it becomes available.
We assessed the accuracy of race/ethnicity data version 5 by first comparing the imputed observations to the values reported by in-state members who have shared their self-reported race and ethnicity with Blue Cross. We then took the weighted average of the accuracy of imputed data and 100% (representing accuracy of the self-reported data), with weights equal to the proportion of members who self-report, accounting for non-random missingness in the self-reported data. For members self-identifying as Asian, race/ethnicity data version 5 have sensitivity: 96.6%, specificity: 99.1%, positive predictive value (PPV): 93.6%, negative predictive value (NPV): 99.5%. Among members self-identifying as Black, these data have sensitivity: 87.6%, specificity: 99.4%, PPV: 87.1%, NPV: 99.5%. For members self-identifying as Hispanic, these data have sensitivity: 86.0%, specificity: 99.4%, PPV: 91.7%, NPV: 98.9%. Among White non-Hispanic members, these data have sensitivity: 99.1%, specificity: 89.0%, PPV: 95.9%, NPV: 97.3%. We currently lack sufficient data to assess the accuracy of imputed data for members who self-identify as American Indian/Alaska Native, Native Hawaiian/Other Pacific Islander, Other, and Multiracial. To generate the accuracy calculations reported in this paragraph, we categorized the output probabilities from the imputation model described above using a plurality rule (i.e., assigning the category with the highest probability). However, in the health equity report, the output probabilities from race/ethnicity data version 5 (when self-reported data were missing) were used directly.
*Indicates when the inequity between minoritized racial and ethnic group (Asian, Black, Hispanic) members and White members is statistically significant (p < 0.05). .
“Insufficient Data” indicates that there were fewer than 90 members of the indicated race and ethnicity who were included in the measure denominator (i.e., who had a condition or health event that caused the measure to apply to them).
Learn more about the corresponding NCQA measures.
The measure specification for Controlling High Blood Pressure is now based on the HEDIS definition that was first implemented for measurement year 2020.
The Severe Maternal Morbidity rate among delivery hospitalizations is based on the CDC measure specification.
The MHQP Patient Experience Survey is a statewide survey of commercially insured patients in Massachusetts. This survey asked patients about experiences with their providers or their children’s providers, and other staff in the providers’ office. The survey also asked members for their self-reported race and ethnicity, with >95% response rates to these survey items among members who returned the survey. To assess care delivered in 2023, MHQP collected survey data in the spring of 2024 from over 33,000 patients statewide, including nearly 18,000 BCBSMA members. The Patient Experience Survey data displayed above reflect care delivered to these nearly 18,000 BCBSMA members, using self-reported race and ethnicity data (and no imputed data) given the high response rate to the race and ethnicity survey items. More information about the MHQP survey can be found at: https://www.mhqp.org/learn-about-the-survey/.
This report is based on Blue Cross’ race/ethnicity data version 5, as described above. To access earlier versions of these race/ethnicity data and earlier performance years, you can download these files:
- Our 2019 equity report based on race/ethnicity data version 1 (as published in September 2021) here
- Our 2019 equity report based on race/ethnicity data version 2 (as published in October 2022) here
- Our 2019 equity report based on race/ethnicity data version 3 here
- Our 2019 equity report based on race/ethnicity data version 4 here
- Our 2019 equity report based on race/ethnicity data version 5 here
- Our 2020 equity report based on race/ethnicity data version 2 (as published in October 2022) here, and
- Our 2020 equity report based on race/ethnicity data version 3 here
- Our 2020 equity report based on race/ethnicity data version 4 here
- Our 2020 equity report based on race/ethnicity data version 5 here
- Our 2021 equity report based on race/ethnicity data version 3 (as published in May 2023) here
- Our 2021 equity report based on race/ethnicity data version 4 here
- Our 2021 equity report based on race/ethnicity data version 5 here
- Our 2022 equity report based on race/ethnicity data version 4 (as published in July 2024) here
- Our 2022 equity report based on race/ethnicity data version 5 here
