1. Methodological Challenges in Developing the National Health Care Quality and Health Care Disparities Reports Julia Holmes NCHS 2004 NCHS Data User’s Conference Washington DC July 12, 2004

2. Presentation Outline 1.Background and Development Ed Kelley (AHRQ) 2. Adapting NCHS and HP2010 measuresSuzanne Proctor (NCHS) for the reportsRichard Klein (NCHS) 3.Methodological ChallengesJulia Holmes (NCHS) in developing the reports 4. Results and Next StepsErnest Moy (AHRQ)

3. Methodological Challenges: Integrating Data from Multiple Data Systems 1. Reporting on race and ethnicity Across data systems Across time 2. Reporting on socioeconomic status (SES) 3. Reporting on clinical guidelines that change over time

4. Reporting on Race and Ethnicity Differs Across Data Systems Data SystemSampling ApproachRace and Ethnicity Reporting Population-based surveysIndividuals in householdsSelf-report and by proxy (parent reporting on child) Establishment-based surveys Records in facilities, providers, establishments Abstracted from provider records Administrative records and claims data Billing records or state discharge systems Same as establishment- based surveys Surveillance systemsRecords of cases that meet selection criteria Generally reflect race and ethnicity by observation, if reported at all Vital StatisticsRecords births and deaths for state registries May be self report (mother for herself); proxy (mother for child); other (funeral director)

5. Reporting on Race and Ethnicity Differs Across Time 1997 OMB Directive mandated a number of changes in how Federal race/ethnicity data are collected & reported Option to select one or more races Split of Asian or Other Pacific Islander into 2 subgroups Several subpopulation category name changes All Federal data systems required to implement by 2003 OMB standards do not affect non-federal data systems Data systems are implementing at different points Data provided to AHRQ for 2003 & 2004 reports used a mix of old and new categories, e.g. - 2000 NIS data use old categories; 2001 use new categories

6. Reporting on Race and Ethnicity in Federal Data Systems Differs Across Time OMB Standard Categories for Race Directive 15 (1977) “Select one…” American Indian or Alaska Native Asian or Pacific Islander Black White Revised Standard (1997) “Select one or more…” American Indian or Alaska Native Asian Native Hawaiian or Other Pacific Islander Black or African American White

7. Effect of Multiple-Race Inclusion … SOURCE: Weed J, Issues in Classification of Race and Ethnicity Data; NCHS Data Users Conference, Summer 2002

8. Multiple Race and “Primary” Race AI/AN/White81.6---16.5---1.8 API/White39.1------48.712.7 Black/White26.750.2------23.2 MR* WhiteBlackAI/ANAPI SOURCE: Sondik EJ, Lucas JW, Madans JH, Smith SS; AJPH; November 2000, Vol. 90, No 10, p1709-1713 * MR = Multiple Race Primary Racial Identification, %

9. New question on Hispanic origin for revised birth and death certificates Decedent of Hispanic origin? Check the box that best describes whether the decedent is Spanish/Hispanic/Latino. Check the “No” box if the decedent is not Spanish/Hispanic/Latino. No, not Spanish/Hispanic/Latino Yes, Mexican, Mexican American, Chicano Yes, Puerto Rican Yes, Cuban Yes, other Spanish/Hispanic/Latino (Specify) _______________________

13. Measures Racial GroupsEthnic Groups Asian[1][1]NHOPIAI/AN>1 RaceHispanic Access to Health Care: Entry into the health care system (21 measures) Data collected100%48%100%33%100% Data collected and able to provide reliable estimate 90%19%52%33%100% Access to Health Care: Structural barriers (9 measures) Data collected100%0%100%33%100% Data collected and able to provide reliable estimate 78%0%44%0%67% Access to Health Care: Patients’ perceptions (23 measures) Data collected100%61%100%0%100% Data collected and able to provide reliable estimate 91%13%48%0%100% Access to Health Care: Utilization of care (34 measures) Data collected100%32%88%50%100% Data collected and able to provide reliable estimate 47%0%38%9%71% Quality of Health Care: Effectiveness (82 measures) Data collected100%50%94%41%100% [1][1] Includes Asian or Pacific Islander

14. Measures Black or African American WomenChildrenElderlyLow Income Rural Access to Health Care: Entry into the health care system Data collected and able to provide reliable estimate 95%57%48%95%86% Access to Health Care: Structural barriers Data collected and able to provide reliable estimate 100% 80%70% Access to Health Care: Patients’ perceptions Data collected and able to provide reliable estimate 100%39%83%96%87% Access to Health Care: Utilization of care Data collected and able to provide reliable estimate 97%41%59%21%41% Quality of Health Care: Effectiveness Data collected and able to provide reliable estimate 74%58%55%32%30% Quality of Health Care: Safety Data collected and able to provide reliable estimate 95% 79%0%100% Quality of Health Care: Timeliness Data collected and able to provide reliable estimate 91%82%73%64%82%

15. Population Surveys AHRQ Medical Expenditure Panel Survey CDC-NCHS National Health Interview Survey SAMHSA National Survey on Drug Abuse Socioeconomic Measures Income, Education, and Insurance Facility or establishment Surveys CDC-NCHS National Ambulatory Medical Care Survey 1999/2000 CDC-NCHS National Nursing Home Survey Insurance Collection of Socioeconomic Status in the first NHDR

16. Surveillance and Vital Statistics CDC-NCHS National Vital Statistics System, 2000 Socioeconomic Measures Education Administrative Databases AHRQ Healthcare Cost and Utilization Project State Inpatient Database, 2000 Median Income of zipcode of patient residence Insurance Collection of Socioeconomic Status in the first NHDR

17. Disadvantages of Selected Socioeconomic Measures Income Respondents often refuse to report Often underreported Varies over time Different meaning depending on a person’s wealth and access to insurance Education Not applicable to children and young adults who have not completed education Often overestimated on death certificates, especially among elderly Different meaning across generations Insurance (Expected Source of Payment) Multiple payers Varies over time and with condition Some groups very heterogeneous Area-based or geocoded income Varies over time More precise at census track level but matching addresses a larger problem May be related to service availability rather than individual income

18. Clinical Guidelines Sometimes Change Over Time Change in HP2010 definition 5-12 (modified). Percent of adults (18 years and over) with diabetes who had a hemoglobin A1c test at least once in the past year Original HP2010 measure was at least once (National Diabetes Quality Improvement Alliance and the National Quality Forum) HP2010 Diabetes Workgroup changed the definition to at least twice to reflect changes in clinical recommendations AHRQ decided to report a hemoglobin A1c test at least once in the past year in the first report

19. Illustrative Example From the National Health And Nutrition Examination Survey, 1999-2000 : Clinical Guidelines Sometimes Differ Based on How the Guidelines Are Used % adult diabetics with acceptable hemoglobin A1c level Percent of Adult Diabetics Whose Diabetes is Under Control

20. American Diabetes Association (ADA) provides treatment guidelines (goals) for individual diabetics: HbA1c< 7.0 National Diabetes Quality Improvement Alliance provides treatment guidelines (goals) for public reporting on a population of diabetics HbA1c < 9.5 Reporting on Clinical Guidelines: Which Guidelines to Use in a National Report

21. Reporting on Clinical Guidelines Why Are There Different Treatment Goals? Individual goal can vary based on severity of diabetes and presence of other comorbid conditions in an individual patient Population goal for a group of diabetic patients does not control for illness severity or comorbid conditions that may make it difficult to attain the lower (Risk adjustment mechanism lacking)

22. HgA1c <9.586.5% HgA1c <9.079.1% HgA1c <7.037.0% Reporting on Clinical Guidelines; Percent of Adult Diabetics Whose Diabetes in Under Control National Health and Nutrition Examination Survey, 1999-2000

23. Methodological Challenges Resolved? Not Really Reporting on Race and Ethnicity  Expand surveys to collect more data on racial groups  Encourage reporting of race and ethnicity on all data collection systems Reporting on Socioeconomic Status  Support research on optimal measures of socioeconomic status Reporting on Clinical Guidelines  Institute mechanism in the report development process that monitors changes in clinical guidelines and performance standards