In the first set of analyses we examined whether depressive symptoms in pregnancy were associated with adverse birth outcomes. After adjustment for covariates, women with probable depression were over one and a half times more likely to give birth to a preterm baby than non depressed women. In the second set of analyses, we examined whether actual participation in the HSI led to a decrease in adverse birth outcomes among depressed women. Results showed that after adjustment for covariates, rates of babies born low birth weight, preterm and small for gestational age were not significantly different among those depressed women who did or did not participate in the Initiative. This lack of difference held even when we controlled for length of time a woman was enrolled in the initiative. Although we could not reliably measure the amount of services each woman enrolled in the HSI received, if length of time is related to amount of services received, our results suggest they may not be correlated.
In the third set of analyses, we examined whether there was a cohort effect that resulted from community-wide education and dissemination of information and attention that resulted from the Healthy Start Initiative. The use of a comparison group within the same local but at a previous time interval was designed to control for confounding characteristics imposed by the environment (e.g., socioeconomic status, access to care, changes in health insurance). As expected, the rates of preterm delivery were reduced for women giving birth after as compared to before the initiation of the HSI. This suggests a cohort effect as compared to an effect restricted to women who enrolled in HSI.
The issue examined by this evaluation is whether a comprehensive set of services such as that of the HSI which included care coordination and education, can reverse the risk that depression confers on adverse birth outcomes. Our evaluation suggests that there is little immediate benefit derived from participation in the HSI. The risks associated with depressive symptoms and depression in pregnancy may not be reversible for the index pregnancy examined in this evaluation. However, it is possible that increased services may lead to the prevention of adverse birth outcomes in subsequent pregnancies. Our findings may be limited by low power since few subjects in the three cohorts had elevated depressive symptoms.
We did find that the presence of the HSI significantly altered the preterm birth rate of women overall. The fact that women did better in the post HSI cohort could be related to characteristics of women who were enrolled (i.e. healthier people) or to the community awareness that was promoted by the HSI.
Although we found a non-significant effect on birth outcomes for depressed women in Healthy Start, our findings pertain to one Healthy Start program during a specified time period, and are not representative of the other Healthy Start communities nationwide. Our inability to find a programmatic effect for babies born small for gestational age is similar to results reported in an evaluation of the Florida Healthy Start program [18
], although the evaluation did not report specifically on depressed women and reported a robust programmatic effect on birth weight and preterm delivery.
A national evaluation (2000) of 15 Healthy Start programs [19
] found that only 27% of the programs had a statistically significant effect on lowering rates of preterm delivery and only 3 or 20% of the 15 project areas saw reductions in low birth weight and very low birth weight in enrollees as compared to non enrollees. The lack of a uniform programmatic effect found by this national evaluation on birth outcomes could represent the etiological heterogeneity between preterm delivery, small for gestational age and low birth weight. If this was the case, it would suggest that there may be particular programmatic services that differentially influence the dimensions of infant morbidity. Although this national evaluation could not determine a causal relationship between program components and observable outcomes, characteristics of programs that were successful in reducing infant morbidity were most often related to strength in program administration, formation or enhancement of linkages between Healthy Start programs and clinical services, and employment of community members as Healthy Start educators, care managers, or peer counselors. Future evaluative efforts should focus on elucidating the specific mechanisms and program components whereby Healthy Start impacts maternal and child health in general and specifically for depressed women in order to improve the potency of the Healthy Start program and replicability of successful practices nationwide.
Our findings on program effectiveness may have been hampered by a number of issues. First, there were substantial differences in the three groups at baseline. Women enrolled in the HSI were less likely to be white and more likely to be Hispanic and pregnant for the first time than women not enrolled. Women enrolled in the HSI were also less likely to smoke, or use drugs or alcohol in pregnancy (18, 12 and 16%, in Groups 1, 2 and 3 respectively), suggesting a ceiling effect for poor birth outcomes.
Second, we utilized the PRIME-MD BHQ to measure a probable depressive disorder. This instrument measures depressive symptoms rather than an actual depressive disorder. The advantages of screening questionnaires are that they are short, easy to administer and can provide a measure of the severity of symptoms. The disadvantage is that they are not able to diagnose depression specifically but are elevated by general emotional distress, concurrent psychiatric illness or general medical conditions. The BHQ has been used in obstetric-gynecologic settings and correlates highly with the Structured Clinical Interview for DSM-III-R (SCID). The overall reliability between the BHQ and the SCID for major depressive disorder was 92 percent and 89 percent for minor depression [15
Third, rather, than lack of control for confounding it may be that we over controlled for some covariates in our models. For example, if Healthy Start workers referred women to smoking cessation programs and women ceased smoking, but we included smoking in our models, than the program may have had an effect we were unable to detect. However, we do not believe this occurred for smoking, drug, or alcohol use since our unadjusted models pertaining to birth outcomes were non-significant even though smoking and other covariates remained significant.
Fourth the lack of effect of the HSI may also have been due to limited power to examine very preterm, very low birth weight, and small for gestational age in women overall and in a subgroup of depressed women. For example, assuming a type 1 and type 2 error rate of 5 and 20%, respectively, and an adjusted odds ratio of 0.94 for SGA in a Healthy Start vs. non-Healthy Start population, we would require a total sample of 119,904 to detect the differences in SGA between the Healthy Start and non-Healthy Start groups. However, the trend we observed in the reduction of small for gestational age babies after the Healthy Start Initiative is of similar magnitude to the rate observed in both the evaluation of Florida Healthy Start [18
] site and the 15 demonstration sites examined in the National Healthy Start Evaluation [19
]. Moreover, typical risk factors for adverse birth outcomes (race/ethnicity, smoking, age) were consistent in our cohort with the literature [20
Fifth, because external controls in comparable communities were not available, macro-level factors such as global changes in socio-economic status and health insurance and changes in clinical practice and quality of care could have also been responsible for decreases in adverse outcomes. However, the use of the cohort comprised of women who received reproductive health care at the same time as women enrolled in the HSI but who were not enrolled in the program aimed to minimize this possibility.
Finally, it is also possible that the follow up period was not of sufficient duration and/or other potential covariates and mediating variables such as rapid repeat pregnancies, previous preterm deliveries, and utilization of prenatal care were not reliably assessed as part of this evaluation. The evaluation of the overall Healthy Start program conducted by Mathematica did demonstrate an increase in utilization of prenatal care among Healthy Start recipients as compared to non recipients in 8 out of the 15 project areas evaluated [19
]. However, the demonstrated effects of pre-natal care on improvement in birth outcomes are mixed [22
]. Moreover, it may have been that the time during which the evaluation was conducted was too short to detect changes in women’s depression status or pregnancy outcomes. Therefore, the program impact of HSI on birth outcomes of depressed women may not be observed in the pregnancy assessed for this review, but rather the impact may be detected in subsequent pregnancies. Future evaluative efforts should follow Healthy Start enrollees for durations long enough to determine the interrelationship between perinatal depression, pregnancy interval, and birth outcomes in women with subsequent pregnancies.
In 2007, Congress appropriated over $100 million dollars to the Healthy Start Initiative through the Public Health Service Act. This funding has been disbursed by the MCHB to 97 communities nationwide to implement or continue Healthy Start programs. Given the scope, economic investment, and potential for replicability of successful practices, the evaluation of Healthy Start programs across the country provide important mechanisms to demonstrate effectiveness and suggest future directions in maternal and child health programming and for perinatal depression.