|Home | About | Journals | Submit | Contact Us | Français|
OBJECTIVE. To evaluate whether adjusting the Health Plan Employer Data and Information Set (HEDIS) low birth weight (LBW) measure for maternal risk factors is feasible and improves its validity as a quality indicator. DATA SOURCE: The Washington State Birth Event Record Data for calendar years 1989 and 1990, including birth certificate data matched with mothers' and infants' hospital discharge records, with 5,837 records of singlet on infants identified as LBW (< 2,500 g) and a 25 percent sample ( n = 31,570) of the normal-weight births ( </= 2,500 g). STUDY DESIGN: We reviewed literature on factors associated with birth weight and identified factors for risk adjustment that are associated with LBW and th at are not modifiable by the health plan . We used vit al records Data to develop and test possible risk adjustment strategies. Finally, because feasibility is important for a HEDIS measure, we assessed health plan readiness to produce a risk-adjusted measure. PRINCIPAL FINDINGS: An LBW indicator that is adjusted for maternal risks represents health plan performance better than the unadjusted rate. In the most parsimonious risk adjustment model LBW risk was higher for mothers with a history of prior preterm birth , LBW, or fet al death . Risk was also high er for primiparas or mothers with high parity, mothers less than 19 years of age, and primiparas over age 35. In a model adding race to these obstetric factors, black, Asian/Pacific Islander, or other non-white, non-Hispanic race were also significantly associated with higher LBW risk. While adjusting for maternal risk improved the LBW measure's validity, the rate adjustment magnitude was small (0.17 percentage points) for the most plausible model. Th is may not be mean in gf ul clinically or for measuring differences in quality. The costs and data collection requirements of risk adjustment could be substantial for health plans lacking access to State birth records data. CONCLUSIONS Selection of risk adjusters for quality measures depends on judgments of their effect, legitimacy, and feasibility. A comprehensive examination of validity and feasibility is needed to understand to what extent outcome measures represent quality and how their value compares to their cost of collection .