Making Better Decisions About Health Disparity Measurement: Slope Index of Inequality Confidence Interval Width Depends More on Number of Groups Than Overall Sample Size

• Activity Number: 358
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Stuart Gansky* and Nancy F. Cheng and Gloria C. Mejia and Wael Sabbah and Eduardo Bernabé
• Companies: UC San Francisco and UC San Francisco and Australian Research Centre for Population Oral Health and King's College London Dental Institute and King's College London Dental Institute
• Keywords: health inequality ; confidence interval ; power ; half-width ; socioeconomic status ; health inequity

Abstract

Methods and software to estimate health disparity indices (HDIs) and 95% confidence intervals (CI) for complex surveys are available (Cheng et al. 2008; US NCI HD*Calc 2009). Slope Index of Inequality (SII) is an often used HDI to assess health inequity among groups, such as socioeconomic position (SEP), along an ordered gradient. Since 2002, over 80 papers in PubMed reported SII to characterize an ordinal explanatory variable's relationship to health differences. We used the 2004-5 California Oral Health Needs Assessment of Children (total N=21,399 children) as a basis for initial simulations with a probit model and a fixed prevalence difference among SEP group levels. We varied overall disease prevalence (0.10-0.90), sample size (2156; 21,399; 42,798), and number of SEP groups (3,4,5,10) with 50 replicates and estimated SII. Varying sample size, by even 20 times more, affected SII CI width little. However, varying number of SEP groups had a dramatic effect: 3-4 groups had much wider CIs than 5-10 groups. Future health research should consider the number of SEP groups when using SII to measure health disparities. Support: UCSF Pilot Grant for King's College London Collaboration.