Moving Toward Evidence-Based Federal Healthy Start Program Evaluations: Accounting for Bias in Birth Outcomes Studies Cristian I. Meghea, PhD, Jennifer E. Raffo, MA, Peggy VanderMeulen, RN, MSN, and Lee Anne Roman, PhD, MSN We used administrative and screening data from 2009 to 2010 to determine if Healthy Start (HS), an enhanced prenatal services pro- gram, is reaching the most vulner- able African American women in Kent County, Michigan. Women in HS are at higher risk of key predic- tors of birth outcomes compared with other women. To advance to- ward evidence-based HS program evaluations in the absence of ran- domized controlled trials, future stud- ies using comparison groups need to appropriately establish baseline equivalence on a variety of risk fac- tors related to birth outcomes. (Am J Public Health. 2014;104:S25–S27. doi: 10.2105/AJPH.2013.301276) For more than 20 years, the federal Healthy Start (HS) program has worked to reduce disparities in maternal and infant health using a core set of interventions tailored to high-risk communities. 1---3 Strong Beginnings, the Grand Rapids, Michigan, HS program, uses a commu- nity collaborative model that builds on the state Medicaid-sponsored enhanced prenatal ser- vices program, the Michigan Maternal and Infant Health Program (MIHP). MIHP services are available to all Medicaid-insured pregnant women and infants and include case manage- ment through office or home visits provided by nurses or social workers employed in multiple community agencies. 4 All African American pregnant women in Kent County are eligible for Strong Beginnings HS, and the program relies on outreach, referrals, and other strate- gies to enroll women at greater risk for adverse birth outcomes. The HS program pairs MIHP professionals with community health workers and mental health coordinators to provide more intensive services. 5 HS evaluations, with 1 exception, received low federal evidence-based ratings because they used quasi-experimental designs that did not establish the study groups equiva- lence at baseline. 6---9 One randomized trial HS evaluation compared high intensity to typical HS home visits, and therefore, was not considered evidence of effectiveness. 10 Previous HS studies were unable to properly account for potential bias introduced by differences between HS participants, women in other enhanced prenatal programs, and nonparticipants, along with risk factors known to be predictors of adverse birth outcomes. 6---9,11,12 To fill this research gap, we aimed to de- termine if the Strong Beginnings HS program is reaching African American women who are at greater risk than those in traditional MIHP or those in Medicaid. We presented risk factors previously not included in HS effec- tiveness studies. This would inform future evidence-based HS evaluations on the extent of the differences and the need to establish comparison group equivalence on specific characteristics. METHODS The study sample consisted of all African American women in Kent County, Michigan, who were Medicaid eligible and delivered a singleton baby between January 1, 2009, and December 31, 2010. Administrative data on all women included sociodemographics, pre- natal care adequacy, 13 smoking, drug use, pregnancy history, and birth outcomes, mea- sured on the infant birth records. We used linked pregnancy Medicaid claims to define depression diagnoses and treatment during pregnancy, as well as chronic disease. We used a detailed MIHP psychosocial and health pre- natal risk screener to measure a variety of risks for the HS and MIHP participants. The prenatal screener has well-established measures (e.g., Edinburgh Depression Scale 14 ) for specific risk factors, and Medicaid state policy mandates its use at MIHP enrollment. Relevant to HS and birth outcomes, it also includes history of adverse birth outcomes, whether the preg- nancy was unwanted, depressive symptoms, perceived stress, and mental health history. Details on the MIHP screener are provided elsewhere. 4 We used the v 2 test for the comparisons reported in Tables 1 and 2. RESULTS Based on administrative data (Table 1), HS participants were less likely to be married (8% vs 18%) and to have had previous pregnancies, and more likely to have incomes less than 33% of the federal poverty level, used drugs during pregnancy (17% vs 8%), and had a depression diagnosis during pregnancy (13% vs 5%) compared with women not participating in either HS or in MIHP. There were no statistically significant differences be- tween HS and MIHP participants, except on marital status (8% vs 16% married). Using the MIHP psychosocial and health risk screener data (Table 2), compared with other MIHP participants, pregnant women in HS were more likely to have a history of adverse pregnancy outcomes (39% vs 28%), smoke during pregnancy, live in a household in which somebody consumed illegal drugs (18% vs 11%), use drugs themselves, and have depres- sive symptoms, a history of mental health problems (41% vs 29%), and a history of domestic violence (43% vs 31%). All the dif- ferences were statistically significant (P < .05). DISCUSSION The administrative data suggested that Strong Beginnings HS reached higher risk women than those not enrolled in either Strong Beginnings HS or MIHP, and that Strong Beginnings HS women were at a similar risk compared with those in MIHP. However, the administrative data had limitations, including the fact that mental health and unhealthy behaviors were identified in medical claims only if diagnosed or treated. Among the women enrolled in either HS or MIHP, risk screener data showed that HS participants were at significantly higher risk on key pre- dictors of adverse birth outcomes compared RESEARCH AND PRACTICE Supplement 1, 2014, Vol 104, No. S1 | American Journal of Public Health Meghea et al. | Peer Reviewed | Research and Practice | S25