Ninth Annual Maternal and Child Health Epidemiology Conference / December 10-12, 2003

The Effect of Collocation of WIC at Managed Care Organization Sites: ID of a Simple Referral Process

 

LARISSA BRUNNER:  Thank you.  So today I'll be discussing with you, "The Effect of Collocation of WIC at Managed Care Organization Sites:  Identification of a Simple Referral Process."  So just to give you a brief synopsis of what we're going to be going through here is first looking at the effect of collocation on pregnancy outcomes in Detroit, Michigan and then talking about how the results of this led to the formation of an intervention program to refer other eligible pregnant women for additional services that they might need.  Just some brief background on WIC.  WIC targets low-income nutritionally at-risk populations, including pregnant and lactating women, infants, and children; and it provides supplemental foods, nutrition education, and counseling; and screening and referral to other health, welfare, and social services.  WIC participation has been associated with lower incidents of low birth weight and preterm deliveries, lower costs for newborn medical care, and greater utilization of health care services by children.  With growing health care costs, state Medicaid programs have turned to the use of managed care organizations as a way to provide quality health care while controlling costs. 

The use of managed care organizations by Medicaid, however, has raised some concerns, including whether managed care organization use is appropriate for children because people are concerned that this will limit coordination with public health, education, and social services systems.  A prior evaluation of WIC and managed care organizations in Detroit found improved health outcomes among infants using collocated sites.  Specifically, these infants were less likely to be underweight and were more likely to be up to date for immunizations when recertified for WIC at about 13 months of age.  So the purpose of this study was twofold:  first, to evaluate whether collocation of WIC at managed care organizations affected pregnancy outcomes; and, second, to design an intervention that would enable non-collocated WIC sites to identify all pregnant women potentially in need of additional assistance and refer them to appropriate outreach staff for follow-up. 

The study population consisted of pregnant women certified for WIC in Detroit from March 1996 to December 1998 who gave birth to singleton infants.  Non-African American women were excluded since there were so few of them in the study base, as well as birth weights less than or equal to 700 grams or greater than or equal to 4,500 grams, and women who gave birth at less than or equal to 20 weeks or greater than 44 weeks' gestation.  Thus, a final study population was 37,654 women.  The exposure was based on the managed care and WIC site that a woman used, and this was divided into five different groups.  The first two groups in pink represent the two collocated groups, so women in Group One used Managed Care Organization A and WIC at Managed Care Organization A.  Women in Group Two used Managed Care Organization B and WIC at Managed Care Organization B.  Women in Groups Three and Four used Managed Care Organizations A and B, respectively, but WIC at Detroit Health Department sites, and women in Group Five used neither Managed Care Organization A nor B and WIC at Detroit Health Department sites. 

The outcome was low birth weight defined as less than 2,500 grams or preterm delivery defined as less than 37 weeks' gestation.  Potential confounders considered in the analysis included mother's age; welfare status; mother WIC-certified before July 1, 1997, which was when managed care organization health care was mandated for Medicaid clients in Detroit; mother WIC-certified before 26th week of gestation, first child on WIC; prenatal care during trimester one; and infant's sex.  A multivariate logistic regression was used to obtain adjusted odds ratios in 95-percent confidence intervals, and a backwards elimination approach was used in which all potential confounding factors were entered into the model initially and the variable with the highest P value was then dropped from the model, the model was reassessed, and this was done until the model contained only those variables found to be significant at the P-less-than-.15 level. 

So to get into some of the results, first, if we look at collocated versus collocated, so if you recall, Group Two were the women who used Managed Care Organization B and WIC at Managed Care Organization B, and women in Group One were women who used Managed Care Organization A and WIC at Managed Care Organization A.  So we see that women and collocated Managed Care Organization B had a decreased risk of low birth weight or preterm delivery as compared to those women who used collocated Managed Care Organization A after adjustment for WIC certification before July 1, 1997, welfare status, and first child on WIC, and this decreased risk was statistically significant.  However, when we moved to look at non-collocated versus non-collocated, we see not a decrease in risk but an increase in risk.  And in the case of Group Five, which was the women who used neither Managed Care Organization A nor B versus Group Four, which were the non-collocated Managed Care Organization B women, we see that this increased risk was borderline statistically significant.  The more interesting results perhaps are looking at collocated versus non-collocated. 

So recall again that Group Two are the collocated Managed Care Organization B women and Group One are the collocated Managed Care Organization A women.  So we see that for the Group Two women, when they were compared to all the other non-collocated groups, in every instance, they had a decreased risk of low birth weight or preterm delivery, and in most cases, this decreased risk was statistically significant.  However, we don't see this same result when we look at Group One collocated women.  We don't see that decrease in risk there.  So why is it that the effect of group on low birth weight or preterm delivery doesn't apply to just collocation in general but specifically just to collocated Managed Care Organization B?  Well, staff interviews were completed, and it was found that the WIC and managed care organization staff at Managed Care Organization A did not engage in special interactions on behalf of their clients, whereas at Managed Care Organization B, the nutritional staff providing WIC used an internal referral form to inform their supervisor of clients with special needs.  And this supervisor, a senior nutritionist, in turn followed up with evaluation, extra nutritional counseling, and collaboration with the managed care organization's case management coordinator to obtain other needed services for the client.  So non-collocated Managed Care Organization B clients did have the same services available to them, but it appears that without detection and intervention by managed care organization staff during WIC service delivery, high-risk clients did not obtain these other services. 

So now to switch back to the intervention program, following this first portion of the study the following criteria were developed for referral of pregnant women by WIC:  lack of Medicaid or other health insurance, lack of health care provider, previous poor pregnancy outcome, and other WIC risk factors corresponding to the state's Medicaid-defined criteria for eligibility to be screened or enrolled in maternal support services.  And these other risk factors include things such as nutritional weight-related concerns, severe demographic or social disadvantages, transportation problems, and concerns with the adequacy of prenatal care.  So a one-page referral form was created to screen every pregnant woman and refer her if necessary.  Clients meeting the referral criteria were asked for verbal approval to release information about them to the Detroit Health Department outreach staff.  And forms were collected weekly at the Detroit Health Department WIC sites and distributed to Detroit Health Department outreach programs. 

This intervention was implemented from October 2000 to September 2001 at 14 Detroit Health Department WIC sites.  And quarterly information obtained from WIC was used to establish a number of pregnant who were clients of the 14 sites, and this information was used to calculate unadjusted rates of referral per total load of pregnant women per quarter.  So of the completed referral forms, 99 percent of the clients accepted referral to an outreach program, 80 percent of these women were identified as lacking Medicaid, other insurance, and/or had no health care provider, while the remaining 20 percent had a risk factor, making them eligible for maternal support services.  The referral rate increased over time.  At the beginning of the intervention, the referral rate was less than five percent of total pregnant clients for 10 of 14 sites, and by the end of the intervention, this referral rate increased to greater than 10 percent of total pregnant clients for seven of 14 sites, and this referral rate ranged from 14 to 38 percent. 

So why was there a difference in referral patterns?  Well, it may have been because of the attitudes of WIC staff.  During the course of this intervention, it was observed that staff at some of the sites weren't too enthusiastic about doing this intervention.  There may have also been differences in populations served by the clinics, which seems like a distinct possibility for two clinics that specialize in high-risk pregnancies.  And since these women were already receiving support through the clinics, it may have been unnecessary to then refer them for further support services.  Otherwise, there was no reason to believe that the populations differed from the average WIC population in Detroit.  So the use of a standard form helped WIC staff initiate an active referral process.  The outreach program was notified of 250 at-risk women per quarter from high-referral sites, and approximately one-third of these referrals received service benefits from the Detroit Health Department and were enrolled in Medicaid with the health care provider or in the Maternal Support Services program.  If other sites had increased their referral rates, approximately 600 women per year could have benefited directly by enrollment in Medicaid, Maternal Support Services, or with the health care provider.  Though the cost of enacting the referral system was not known, it appears that this simple process assisted the Detroit Health Department in identifying pregnant women in need of its services. 

So in conclusion, information provided during WIC certification visits enables WIC to provide an important surveillance mechanism that identifies clients in need of assistance.  Mere proximity to resources doesn't mean that a woman will get the help and support that she needs.  Rather, active recruitment, communication, and coordination between staff is needed for clients to obtain the timely health care and support services they need.  And just to acknowledge some other people who were instrumental throughout this study.