Results

The intervention program was fully developed, and all 20 clinic case managers were trained. Both groups were recruited, although not at the projected levels. In all, 154 women were recruited to the comparison group and 142 to the intervention group—72% and 85% consent rates, respectively. The historical cohort design was successful in that the two groups did not differ significantly on the baseline measures (table 1). The average score on the mental health measure (MHI-5)5 was 53 (SD = 22), almost exactly equal to the score of a national sample of women aged 25–34 years seeing mental health providers, and below the average score of women aged 25–34 years in a general US sample (m = 72, SD = 19).6

The proportions lost to follow up before the end of pregnancy were similar—42% of the comparison group and 37% of the intervention group (after reducing the denominators for one maternal and three fetal deaths in the comparison group and three fetal deaths in the intervention group). The main reasons for these rates differed: two thirds of those lost to follow up in the comparison group left the clinic (versus 85% in the intervention group) and almost one third were missed by the clinic for their eighth month measure (versus 10% in the intervention group). Women lost to follow up did not differ from those with eighth month measures, except that completers were more likely to be precontemplators (31% v 20%) and less likely to be in preparation (26% v35%) for pregnancy cessation (p = 0.05).

Exposure to the intervention was limited. Based on clinic visits with and without case management, of those with an eighth month measure, no more than 38% could have received the criterion > 2 motivational interviewing sessions; 44%, 1 session; and 17% no session. Further, observations of the case managers indicated that performance often did not meet our criteria for the motivational interviewing style.7 Thus, no between-group differences were expected at the eighth month, and none were found. Comparing the 55 women in the comparison group and 36 in the intervention group who had both urine samples and questionnaires at the eighth month, and including the 35 in the comparison group and 54 in the intervention group who had not left the clinic but who did not have a urine sample (classified as smokers, regardless of self report), the two groups had the same percentage of cessation—2.2% (both 2/90). The criterion cotinine value was < 80 ng/ml.

Discussion

Development of the intervention components and training was our most significant accomplishment. Systemic barriers and changing health care policies made delivery of the intervention problematic, however. With 25% of the recruited population in the precontemplation stage and in view of the low level of cessation, we believe that the intensive enhanced case management intervention developed for this trial is appropriate and should be tested in an efficacystudy where intervention delivery is assured.

In our enthusiasm to contribute to the SFF initiative and confidence in the perinatal outreach clinic system, we did not foresee how turbulent the transition to managed care for Medicaid beneficiaries would become during the pilot test. We proceeded based on a best case scenario, with insufficient project time for developmental testing of the delivery system and insufficient financial resources to maintain control over implementation when the clinics were forced to reduce case management activities in favour of medical services. On the day the grant was awarded, the medical school based clinic network was given many small rural clinics previously under the direction of the state health department. These additional clinics had to be incorporated without additional funding or personnel, creating a serious burden for clinic staff and budgets. This proved to be a difficult and chaotic period, resulting in high staff turnover.

Another event in the external environment also disrupted clinic operations. Start-up of a Medicaid managed care pilot study in the area gave many clinic patients unprecedented access to private care. In some clinics half of those who initiated prenatal care transferred to a private physician. The clinics responded by reducing the maximum time spent waiting for or receiving services. This made the intervention time unwelcome. Case managers, most of whom are nurses, were reassigned to medical services. In short, the clinics were beset with changes and problems that had a major impact on our ability to recontact and follow recruited women and importantly, on intervention delivery.

Provision of small personal monetary incentives to case managers to recruit study participants and provide the intervention was not effective in the face of major system changes. This intervention should be tested in an efficacy study, with intervention delivery assured by using case managers who are selected, supervised, and paid by the project.

Lastly, from qualitative data we learned that saliva rather than urine is a better choice for a biologic measure to keep the samples from being thrown out in the clinic laboratory and to avoid the suspicion by study participants that urine might be tested for illegal drugs.