Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Pathways Community Care Coordination in Low Birth Weight Prevention

Pathways Community Care Coordination in Low Birth Weight Prevention Matern Child Health J (2015) 19:643–650 DOI 10.1007/s10995-014-1554-4 Pathways Community Care Coordination in Low Birth Weight Prevention • • Sarah Redding Elizabeth Conrey • • • Kyle Porter John Paulson Karen Hughes Mark Redding Published online: 20 August 2014 The Author(s) 2014. This article is published with open access at Springerlink.com Abstract The evidence is limited on the effectiveness of Women participating in CHAP and having a live birth in home visiting care coordination in addressing poor birth 2001 through 2004 constituted the intervention group. outcome, including low birth weight (LBW). The Com- Using birth certificate records, each CHAP birth was munity Health Access Project (CHAP) utilizes community matched through propensity score to a control birth from health workers (CHWs) to identify women at risk of having the same census tract and year. Logistic regression was poor birth outcomes, connect them to health and social used to examine the association of CHAP participation services, and track each identified health or social issue to a with LBW while controlling for risk factors for LBW. We measurable completion. CHWs are trained individuals identified 115 CHAP clients and 115 control births. Among from the same highest risk communities. The CHAP the intervention group there were seven LBW births Pathways Model is used to track each maternal health and (6.1 %) compared with 15 (13.0 %) among non-CHAP social service need to resolution and CHWs are paid based clients. The adjusted odds ratio for LBW was 0.35 (95 % upon outcomes. We evaluated the impact of the CHAP confidence interval, 0.12–0.96) among CHAP clients. This Pathways program on LBW in an urban Ohio community. study provides evidence that structured community care coordination coupled with tracking and payment for out- comes may reduce LBW birth among high-risk women. S. Redding  M. Redding (&) Community Health Access Project, Columbus, OH, USA Keywords Low birth weight prevention  Community e-mail: reddingmark@att.net health worker  Community care coordination  Social S. Redding e-mail: Sarah.redding@me.com determinants of health  Pay for performance  Home visiting E. Conrey State Maternal and Child Health Epidemiologist CDC Assignee, Ohio Department of Health, Columbus, OH, USA e-mail: ElizabethJConrey@odh.ohio.gov Introduction K. Porter Infant mortality rates are used as an indicator for the health Ohio State University Center for Biostatistics, Columbus, OH, of a community. To prevent infant deaths, mothers need to USA e-mail: kyle.porter@osumc.edu be healthy, live in a safe environment, and have access to quality care. Reducing low birth weight (LBW) and pre- J. Paulson mature births has been identified as a key strategy to Center for Public Health Statistics and Informatics, Ohio decrease infant mortality [1]. While infant mortality rates Department of Health, Columbus, OH, USA e-mail: John.paulson@odh.ohio.gov in the US have improved over the past decades, they have been stagnant in Ohio. In fact, Ohio ranked second worst K. Hughes for black infant mortality among all states, and fourth worst Division of Family and Community Health Services, Ohio for overall infant mortality in 2010 [2, 3]. Nationally, Department of Health, Columbus, OH, USA despite overall improvements, the 2011 Centers for Disease e-mail: Karen.Hughes@odh.ohio.gov 123 644 Matern Child Health J (2015) 19:643–650 Control (CDC) Health Disparities and Inequalities Report Secondary objectives were a comparison of adequacy of showed that large disparities in infant mortality rates per- prenatal care and a cost savings evaluation. sist [4]. Strategies that incorporate the community and directly reach out to women at greatest risk for poor birth outcomes Methods may help communities move towards health equality. Home visiting services are one strategy used to improve The CHAP Intervention birth outcomes and have received increased attention and focus on providing evidence-based services to vulnerable Initially, 4 years of birth certificate data were used to children and families through the Affordable Care Act and identify where the LBW births were occurring in Richland the Maternal, Infant, and Early Childhood Home Visiting County. Eligibility for participation in CHAP was based on (MIECHV) program [5]. Although home visiting has been residence in a census tract with high LBW and poverty shown to be effective in impacting parent behaviors, child rates. Seven census tracts comprised the program-eligible cognitive outcomes and maternal life course, the impact on communities; two of these census tracts (6 and 7) repre- birth outcomes is not as clearly evident [6, 7]. sented only six percent of the county population, but The Community Health Access Project (CHAP) is a almost thirty percent of all county LBW births. nonprofit, community based organization that has been The CHWs that provided home visiting services here providing care coordination services in Richland County, were hired from the program-eligible communities and Ohio since 1999. CHAP utilizes community health workers trained at the local community college. CHAP developed (CHWs) to identify women at risk of having poor birth an extensive CHW-specific training curriculum that was outcomes, connect them to health and social services, and delivered for college credit. CHWs were supervised by track each identified issue to a measurable completion. either a registered nurse or physician. CHAP’s intensive home visiting model uses an account- Community health workers (CHWs) functioned as ability tool called Pathways [8, 9]. A Pathway addresses community care coordinators, not providers of direct ser- clearly defined actions towards problem resolution and is vices, and assisted participants to overcome barriers faced not considered complete until a measurable outcome is in obtaining necessary health or social services. CHAP achieved. One participant may be assigned to many dif- developed checklists to be used at each face-to-face home ferent Pathways depending on the problems identified visit encounter between the client and the CHW. A ‘‘yes’’ during the initial interview and subsequent home visits answer to certain questions triggered the initiation of a [10]. As in most communities, Richland County had geo- defined Pathway. For example, if a client answered ‘‘yes’’ to the question—‘‘Do you need a medical home?’’—then a graphic areas of health inequality. CHAP used a mapping strategy to determine the census tracts where the unfavor- Medical Home Pathway was initiated. able birth outcomes were disproportionately occurring. The Pathways are tools to track each identified health or infant mortality rates in Richland County from 2001 to social issue through to a measurable completion or out- 2005 were 6.7 infant deaths per 1,000 live births for white come; typically confirmation that the client actually women, and 17.3 for African-American women [2]. received the medical or social service is required. The The impact of CHWs has been difficult to document. Medical Home Pathway tracks the participant’s connection The Agency for Healthcare Research and Quality (AHRQ) to an ongoing source of primary care and is not docu- released a report on the outcomes of CHW interventions in mented as complete until the CHW confirms that the client 2009, based on 15 different programs, which showed has a medical home. If the client does not connect with a minimal impact on birth outcomes [11]. The CHAP model medical home, then the Pathway is closed as ‘‘finished differs from those programs previously studied in that an incomplete’’; recording that the desired outcome was not accountability measurement tool—Pathways—was used to achieved. In a similar fashion, the Pregnancy Pathway track each health or social issue a pregnant client faced confirms the connection to and maintenance of prenatal through to a measurable completion. Additionally, con- care and is not complete until delivery of a viable normal tracts were developed with funders to pay for completed birth weight infant (Fig. 1). A full description of the model Pathways or outcomes [8, 9]. can be found in the Agency for Healthcare Research and We evaluated if LBW would be reduced when women at Quality ‘‘Connecting Those at Risk to Care’’ publications risk of having a LBW infant were provided with intensive [8, 9]. home visiting and community based care coordination by Contracts were developed between funders and CHAP CHWs, and Pathways were used to document outcomes. with payment tied to specific Pathway benchmarks and The primary objective was to compare the adjusted odds of Pathway completions. In addition, the CHWs received LBW between CHAP recipients and non-CHAP recipients. incentive payments if they completed a designated number 123 Matern Child Health J (2015) 19:643–650 645 Fig. 1 Pregnancy pathway of Pathways. This strategy improved the accuracy of factors for LBW than the general population within each Pathway tracking within the agency, because monitoring census tract, propensity score matching was performed to was occurring both programmatically and operationally. select a comparison group with a similar distribution of risk factors from Ohio vital statistics records [12, 13] The Study Population and Data Sources matching process consisted of estimating propensity scores using a logistic regression model, then matching CHAP The study was limited to census tracts in which at least five clients to controls with similar propensity scores. The women received CHAP care coordination and gave birth in logistic regression model was fit to the data from eligible the time period 2001–2004 (tracts 3, 4, 5, 6, 7, 8, 10 in mothers, with CHAP client (yes/no) as the dependent Richland County, Ohio). Only singleton births were variable. Predictors of CHAP enrollment in this model included in the analysis. CHAP medical records were included mother’s age (\16, 16–18, [18), race (African- identified for all women meeting the study criteria and all American or white), education (if [18 years old: less than were successfully matched to an Ohio live birth record. high school, high school graduate, one or more years of Data on the mother’s trimester of enrollment into CHAP college), marital status, census tract, and delivery year. All and the number of Pathways initiated were extracted from two-way interactions were tested; none were statistically CHAP records. All other study data were from Ohio vital significant and all were dropped from the model. From this statistics records. Because CHAP clients had more risk logistic regression model, a score reflecting the probability 123 646 Matern Child Health J (2015) 19:643–650 of CHAP enrollment was estimated for each eligible Table 1 Characteristics of community health access project (CHAP) clients, all non-CHAP mothers* identified from birth certificates, and mother. matched controls Next, the propensity score was used in an optimal matching algorithm to match each CHAP recipient to one CHAP Matched All non-CHAP* clients controls Births (pre- control. Optimal matching is known to be superior to (n = 115) (n = 115) matching)* nearest-neighbor or ‘‘greedy’’ matching [14]. Exact mat- (n = 1,443) ches for county and delivery year were required. Age This study was exempted by the Ohio Department of \16 16 (13.9 %) 10 (8.7 %) 36 (2.5 %) Health Institutional Review Board and conducted in accord 16–18 13 (11.3 %) 13 (11.3 %) 122 (8.5 %) with prevailing ethical principles. [18 86 (74.8 %) 92 (80.0 %) 1,285 (89.0 %) Race Analysis African- 78 (67.8 %) 80 (69.6 %) 325 (22.5 %) American To evaluate the CHAP program’s impact on LBW, logistic White 37 (32.2 %) 35 (30.4 %) 1,118 (77.5 %) regression models were fit to the LBW outcome. First, the Education unadjusted LBW odds ratio for CHAP mothers versus non- Less than HS 28 (32.6 %) 29 (31.5 %) 220 (17.1 %) CHAP mothers was calculated. Then, two multivariate High school 36 (41.9 %) 40 (43.5 %) 628 (48.9 %) logistic regression models were fit, the primary with only graduate non-modifiable risk factors and a secondary also including Any college 22 (25.6 %) 23 (25.0 %) 436 (34.0 %) factors modifiable by the CHAP program. Multivariable Marital status adjustment was also appropriate, as propensity score Married 17 (14.8 %) 19 (16.5 %) 661 (45.8 %) matching and multivariable adjustment are often used in Not married 98 (85.2 %) 96 (83.5 %) 782 (52.2 %) combination to reduce potential bias [15]. The primary Census tract model was ‘‘non-modifiable only’’ because it is less likely 3 18 (15.7 %) 20 (17.4 %) 110 (7.6 %) to over adjust for the mediating effects of CHAP inter- 4 8 (7.0 %) 5 (4.4 %) 188 (13.0 %) vention. Covariates included in the primary model were the 5 20 (17.4 %) 17 (14.8 %) 211 (14.6 %) propensity score matching variables (mother’s age, race, 6 51 (21.7 %) 26 (22.6 %) 226 (15.7 %) education, marital status, census tract, and delivery year), 7 31 (27.0 %) 34 (29.6 %) 159 (11.0 %) previous preterm or LBW delivery and tobacco use during 8 5 (4.4 %) 6 (5.2 %) 159 (11.0 %) pregnancy (none vs. any throughout pregnancy, thus non- 10 8 (7.0 %) 7 (6.1 %) 390 (27.0 %) modifiable). Other risk factors considered for inclusion in the secondary model were hypertension (chronic or preg- Year of birth nancy-associated), eclampsia, incompetent cervix, renal 2001 44 (38.3 %) 44 (38.3 %) 383 (26.5 %) disease, and uterine bleeding. However, only hypertension 2002 34 (29.6 %) 34 (29.6 %) 347 (24.1 %) was added to the secondary model because there were very 2003 26 (22.6 %) 26 (22.6 %) 354 (24.5 %) few occurrences of the other conditions. 2004 11 (9.6 %) 11 (9.6 %) 359 (24.9 %) To evaluate the secondary objective, the CHAP pro- Tobacco use 45 (39.1 %) 43 (37.4 %) 528 (36.6 %) gram’s impact on the adequacy of prenatal visits, an Previous preterm 3 (2.6 %) 2 (1.7 %) 11 (0.8 %) or LBW ordinal logistic regression model was fit to adequate pre- delivery natal visits versus less than adequate prenatal visits based Hypertension 2 (1.7 %) 4 (3.5 %) 43 (3.0 %) on the Kotelchuck index [16]. A logistic regression model Eclampsia 2 (1.7 %) 2 (1.7 %) 16 (1.1 %) was also fit to first trimester prenatal care versus other than first trimester prenatal care. * Single birth from census tract 3, 4, 5, 6, 7, 8, or 10 The number of LBW births prevented was estimated by Among mothers [18 years of age subtracting the observed number of LBW deliveries from Defined as any tobacco use during pregnancy reported on birth the number expected in the study population if there had certificate been no CHAP intervention. The calculation required the Chronic or pregnancy-related relative risk, for which the odds ratio was considered a sufficient estimate (unadjusted relative risk = 0.43 and 0:5 þ OR  0:5 unadjusted odds ratio = 0.47). The estimate was taken from the model adjusting for both hypertension (modifi- which is the fraction of study women in the non-CHAP able) and non-modifiable risk-factors. First, the fraction of group ? CHAP risk relative to non-CHAP (CHAP odds LBW births not prevented by CHAP was calculated as ratio) multiplied by the fraction in the CHAP group. Next, 123 Matern Child Health J (2015) 19:643–650 647 the observed number of LBW births was divided by this common non-medical Pathways initiated were Employ- fraction and rounded to the nearest integer. This method ment (52 %), Adult Education (50 %), Smoking Cessation was repeated using the lower and upper confidence limits (39 %), Food Security (30 %), and Housing (27 %). Two of the odds ratio to obtain the confidence interval. This major barriers that were identified to completion of Path- method is equivalent to multiplying the preventable frac- ways included transportation and limited community tion (1—odds ratio) by the fraction treated, subtracting that resources for non-medical issues. from one and multiplying the reciprocal by the number of Women enrolled in CHAP care coordination from 2001 observed events [17]. through 2004 had significantly lower adjusted odds of To estimate the potential cost savings of the CHAP experiencing a low-birth weight delivery than non-CHAP program, we first estimated the number of LBW births women [adjusted odds ratio = 0.36, 95 % CI (0.12, 0.96)] avoided using the method described above. We then esti- (Table 2). There were no significant differences between mated the average cost of delivering the CHAP interven- the adjusted odds of the adequacy of prenatal visits or the tion per client by evaluating the cost per Pathway, cost per timing of the first prenatal visit between CHAP participants client, and the amount paid to CHAP per number of and non-CHAP mothers. This finding is different from pregnant clients within grant and service contracts. The other home visiting studies that have shown a dosage effect greatest cost of the program was time spent by a CHW to of prenatal home visiting in at-risk women [19, 20]. provide care coordination and the amount of time spent by Fifty-six percent of clients in this study entered CHAP a CHW was primarily driven by trimester of entry into in the first trimester of pregnancy, 20 % in the second CHAP. trimester and 24 % in the third trimester. The estimated To evaluate cost savings from LBW births averted by cost to provide Pathways community care coordination by CHAP participation, we applied the average excess LBW CHAP in the time period studied averaged $751 per costs provided in the 2006 Institute of Medicine (IOM) pregnant client. An estimated 10 LBW births (1 prevented report [18] to our estimate of LBW births averted. Per per 11.5 participants) were prevented by participation in IOM, in the first year of life, excess medical expenses per the CHAP program from 2001 through 2004 (95 % LBW infant are $29,000 and long term costs (including CI = 1, 17). The cost savings in the first year of life, for maternal costs, early intervention, special education and each dollar invested in Pathways based community care lost household and labor market productivity) are $48,275. coordination was $3.36, and the long term cost savings was The dollars saved per dollar invested was calculated by $5.59 for each dollar invested. dividing the total cost savings for one prevented LBW infant by the total cost to serve enough pregnant women with Pathways focused care coordination. Discussion Pregnant women who participated in CHAP, a structured Results community-based care coordination program provided by CHWs and coupled with Pathways tracking and payment Characteristics of CHAP participants and non-participant for outcomes, had a significantly lower probability of controls are summarized in Table 1. The CHAP and non- delivering a LBW infant. CHAP participants living in the CHAP groups did not differ significantly (p \ 0.05) in any targeted census tracts were at an increased risk for poor of the propensity score variables; the groups are within birth outcomes compared to the general population— 2.6 % points for all levels of all propensity score variables 67.8 % African-American, 25.2 % age 18 or younger, with the exception of age, which had a 5.2 % point dif- 85.2 % unmarried, and 39.1 % tobacco users. A challenge ference. There were no reported cases of incompetent to determining the effectiveness of CHW interventions has cervix, uterine bleeding, or renal disease in either group. been identifying a valid control group that effectively A total of 653 Pathways were initiated for the CHAP accounts for social determinants and their impact on out- participants, and all 115 women in this study finished a comes [21, 22]. Use of an optimal matching algorithm Pregnancy Pathway (7 were finished incomplete due to using propensity scores allowed each CHAP recipient to be LBW). Including the Pregnancy Pathway, CHAP partici- matched with one control and supported estimation of the pants had an average of 5.6 Pathways tracked for health number of LBW births prevented. and social issues that were identified during the pregnancy Areas of health inequalities—whether related to birth and postpartum period. 102 Postpartum and Family Plan- outcomes or chronic diseases—can be easily mapped in ning Pathways were completed for participants, confirming communities. This study demonstrates the value of iden- that 89 % of women attended their postpartum appoint- tifying communities with disparately poor health outcomes ments and were using a family planning method. The most and directly reaching out to individuals within those 123 648 Matern Child Health J (2015) 19:643–650 Table 2 Odds ratios and 95 % a Variable Unadjusted Primary model: adjusts for non- Secondary model: adjusts for confidence intervals for pre- b c modifiable risk-factor covariates all risk-factor covariates term birth CHAP versus 0.43 (0.16, 1.07) 0.36 (0.12, 0.96) 0.37 (0.12, 1.02) non-CHAP Age \16 versus [18 1.58 (0.40, 6.28) 1.17 (0.42, 6.70) Census tract comparisons 16–18 versus [18 2.13 (0.66, 6.85) 2.11 (0.65, 6.84) excluded African-American 1.13 (0.35, 3.70) 0.93 (0.28, 3.09) Mother’s age (\16, 16–18, versus White [18), race (African-American, Not married 3.06 (0.87, 10.0) 4.11 (1.06, 15.92) white), marital status, census versus married tract, previous preterm or LBW delivery, tobacco use at any Previous preterm 3.06 (0.50, 18.52) 3.44 (0.55, 21.43) time during pregnancy (y/n) or LBW delivery All from primary model and additionally hypertension Tobacco use 4.76 (1.92, 11.84) 5.09 (2.01, 12.87) (chronic and/or pregnancy- Hypertension 6.25 (0.91, 43.16) associated) communities, engaging them through care coordination, important part of the care plan, documentation, and connecting them to health and social service interventions, reporting in this study. and measuring the results through an accountable mea- There were several limitations in this study. First, surement tool. although data was collected over a 4-year time period, the Community health workers perform their work by total number of women in the CHAP intervention group approaching the whole person—and take into consideration was small, reflecting the size of program enrollment within their social, environmental, psychological and health needs the targeted census tracts over the time period studied. A in order to impact health outcomes. This is evidenced by larger sample size would have provided more precise the additional Pathways initiated by CHWs in this study for estimates of odds ratios and more power to detect signifi- issues related to food security, housing, transportation, cant differences in all models. Second, there was no ran- employment, and education. These additional Pathways dom assignment to CHAP intervention or control. had to be addressed in coordination with preventive health Although we attempted to control for bias as much as care needs and consideration of the client’s priorities of possible through propensity score matching and covariate care. Health and social service siloes exist in communities, adjustment, some selection bias may remain. Additional and individuals living in poverty often face barriers in evaluations, with randomized group assignments, larger accessing these critical services. The community-based numbers of participants, and in different locations are care coordinator serves an important role on the healthcare needed to replicate and confirm our findings. Third, the team because of their trusted relationship with the client. evaluation was limited by the vital statistics records on They are able to identify key non-medical issues and are what cofounders and outcomes we could study. For skilled in navigating the fragmented health and social example, prenatal smoking is potentially modifiable service systems. through CHAP with a Pathway that included specific Some social determinants of health can be addressed at education and support to help patients reduce or quit the population level—such as safe drinking water, smoking smoking; however smoking status by trimester was not in public places, elimination of food deserts and safe standard documentation on the Ohio birth certificate. sidewalks—but individually addressable social determi- Future work should control for first trimester smoking nants also represent a significant intervention opportunity. status and other factors related to low birth weight. Finally, Housing, education, employment, food security, and many the evaluation was limited by the quality of birth certificate other critical issues can be identified and addressed with data, which is shown to generally be specific, but not effective and accountable care coordination to improve sensitive, as a source of maternal complications [23, 24]. In individual progress, reduce stress, and improve health for contrast, birth weight data from the birth certificate has those individuals at greatest risk. been shown to be more reliable [25]. The CHAP Pathways Model provided the measurement CHAP may reduce LBW delivery among high risk tool to monitor successful connections to both health and women through multiple mechanisms. As there were no social services. Pathways were developed as the pay-for- differences in prenatal care initiation between groups, performance model for CHAP’s contracts and were an improvement in early prenatal care does not appear to be 123 Matern Child Health J (2015) 19:643–650 649 one, and this finding is consistent with other studies [26]. programs’’ [1]. This study shows that structured commu- However, factors besides medical care are known to impact nity-based care coordination coupled with standardized and health outcomes and models of care that address both accountable tracking tools and payment for outcomes may medical and social factors show promise in reducing LBW reduce LBW delivery among high-risk pregnant women. [27–30]. The Pathways Model allows for targeting the diversity of This study represents our initial experience with using needs across racial, ethnic and other sociodemographic the Pathways Model to quantify and track care coordina- distinctions. Identifying communities with disparately poor tion provided to high risk pregnant women. Since the health outcomes and ensuring the connection of residents model’s inception, effort has been placed on refining the to health and social programs can potentially reduce per- measurement and tracking process of the Pathways. It was sistent inequalities in health. not possible in this study to identify which Pathways spe- Acknowledgments We would like to acknowledge the CHAP cifically led to improved birth outcomes. Newer technology community health workers—their wisdom and understanding of the for Pathway tracking has remedied that and can support community has always led the way; The Osteopathic Heritage future research. CHAP participants were initially identified Foundation who supported the initial development and pilot of the as being at increased risk by where they lived (identified Pathways Model; Dan Wertenberger the Executive Director of CHAP; Celia Flinn, MD who continues to be a key partner in census tracts), but now we have the capability to monitor designing the intervention and supporting the program; Kathryn risk throughout the care coordination period. Our pre- Meagly, MPH who served as a volunteer collecting data; Cynthia liminary study can be incorporated into the larger move- Shellhaas, MD, MPH, who’s clinical expertise provided guidance ment to create a national home visiting research network during the analytic phase; and Wallace L. Alward, MD who provided editorial work and statistical review. We would also like to thank the that works to promote the translation of research findings Centers for Disease Control and Prevention, Maternal and Child into policy and practice [31]. Health Epidemiology Team for their critical assistance in the devel- Starting from an American Academy of Pediatrics— opment of this article. The findings and conclusions in this report are Community Access to Child Health (CATCH) Grant in those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. 2001—the Pathways Model was further developed to embrace multiple care coordination agencies within a ser- vice region. The Pathways Community HUB Model is Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, dis- designed to identify the most at-risk individuals in a tribution, and reproduction in any medium, provided the original community, connect them to evidence-based interventions, author(s) and the source are credited. and measure the results [8, 10]. The HUB Model was developed and piloted by CHAP in 2004 in Richland County, Ohio based on the success of these initial findings. References The model was recognized by the Agency for Healthcare Research and Quality (AHRQ) Innovations Exchange and a 1. McCormick, M. C. (1985). The contribution of low birth weight to infant mortality and childhood morbidity. New England learning network was established to further study the Journal of Medicine, 312(2), 82–90. model [8, 9]. The Pathways Community HUB does not 2. Ohio Department of Health. Latest Ohio Infant Mortality Data. directly provide care coordination services, but subcon- http://www.odh.ohio.gov/odhprograms/cfhs/octpim/latestoimd. tracts with care coordination agencies serving vulnerable aspx. 3. MacDorman, M. F., & Mathews, T. H. (2011). Understanding populations in the community. The community HUB works racial and ethnic disparities in U.S. infant mortality rates. NCHS to coordinate and track progress for all of the agencies Data Brief, 74, 1–8. within a community providing care coordination. The HUB 4. CDC Health Disparities and Inequalities Report. (2011). Sup- serves to register and collect focused data on each client plement (p. 2011). United States: Centers for Disease Control. 5. Wakefield, M. K., & Sheldon, G. H. (2013). Special supplement served using common Pathways to track quality and out- on the maternal, infant and early childhood home Visiting (MI- comes. This model eliminates duplication of care coordi- ECHV) program. Foreword. Pediatrics, 132(Suppl 2), S57–S58. nation and provides standard quality measurements, 6. Filene, J. H., Kaminski, J. W., Valle, L. A., & Cachat, P. (2013). allowing care coordination agencies to focus on the most Components associated with home visiting program outcomes: A meta-analysis. Pediatrics, 132, S100–S109. vulnerable community members and strive towards 7. Avellar, S. A., & Supplee, L. H. (2013). Effectiveness of home improving overall health outcomes. The Kresge Founda- visiting in improving child health and reducing child maltreat- tion has recently supported an initiative to develop a ment. Pediatrics, 132, S90–S99. standard approach for certification of communities utilizing 8. Connecting Those at Risk to Care—The Quick Start Guide to Developing Community Care Coordination Pathways. http:// the Pathways Community HUB Model to assure consistent www.innovations.ahrq.gov/guide/quickstartguide/commhub_ quality of care coordination. quickstart.pdf. As stated by CDC, health disparities ‘‘must be addressed 9. Connecting Those at Risk to Care—A guide to building a com- with intervention strategies related to both health and social munity ‘‘HUB’’ to promote a system of collaboration, 123 650 Matern Child Health J (2015) 19:643–650 accountability, and improved outcomes. http://www.innovations. birth outcomes. JAMA Pediatrics. doi:10.1001/jamapediatrics. ahrq.gov/guide/HUBManual/CommunityHUBManual.pdf. 2013.4347. Published online January 6, 2014. 10. Pathways Manual, Community Health Access Project. http:// 21. Livingood, W. C., Brady, C., Pierce, K., Atrash, H., Hou, T., & chap-ohio.net/press/wp-content/uploads/2010/09/PathwaysMan Bryant, T., I. I. I. (2010). Impact of pre-conception health care: ual1.pdf. evaluation of a social determinants focused intervention. Mater- 11. Viswanathan M KJ, Nishikawa B, Morgan LC, Thieda P, Hon- nal and Child Health Journal, 14, 382–391. eycutt A, Lohr KN, Jonas D. Outcomes of Community Health 22. Salihu, H. M., Mbah, A. K., Jeffers, D., Alio, A. P., & Berry, L. Worker Interventions Evidence Report/Technology Assessment (2009). Healthy start program and feto-infant morbidity out- No. 181. Rockville MD: Agency for Healthcare Research and comes: evaluation of program effectiveness. Maternal and Child Quality; 2009 June 2009. Health Journal, 13, 56–65. 12. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the 23. Bailit, J. L. (2010). Rates of labor induction without medical propensity score in observational studies for causal effects. Bio- indication are overestimated when derived from birth certificate metrika, 70, 41–55. data. Am J Obstetrics and Gynecol, 203(3), 269.e1–269.e3. 13. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a control- 24. Piper, J., Mitchel, E., Snowden, M., et al. (1993). Validation of group using multivariate matched sampling methods that incor- 1989 Tennessee birth certificates using maternal and newborn porate the propensity score. American Statistician, 39, 33–38. hospital records. American Journal of Epidemiology, 137, 14. Rosenbaum, P. R. (1989). Optimal matching for observational 758–768. studies. Journal of American Statistical Association, 84, 25. Northam, S., & Knapp, T. R. (2006). The reliability and validity 1024–1032. of birth certificates. JOGNN, 35, 3–12. 15. Rubin, D. B., & Thomas, N. (2000). Combining propensity score 26. Lu, M. C., et al. (2003). Preventing low birth weight: Is prenatal matching with additional adjustments for prognostic covariates. care the answer? Journal of Maternal-Fetal and Neonatal Med- Journal of American Statistical Association, 95, 573–585. icine, 13, 362–380. 16. Kotelchuck, M. (1994). An evaluation of the Kessner adequacy of 27. Krans, E. E., & Davis, M. M. (2012). Preventing low birth prenatal care index and a proposed adequacy of prenatal care weight: 25 years, prenatal risk, and the failure to reinvent pre- utilization index. American Journal of Public Health, 84, natal care. American Journal of Obstetrics and Gynecology, 1414–1420. 206(5), 398–403. 17. Meddings, D. R., Ronald, L. A., Marion, S., Pinera, J. F., & 28. Lee, E., et al. (2009). Reducing low birth weight through home Oppliger, A. (2004). Cost effectiveness of a latrine revision visitation: A randomized controlled trial. American Journal of programme in Kabul, Afghanistan. Bulletin of the World Health Preventive Medicine, 36, 154–160. Organization, 82, 281–289. 29. Ickovics, J. R., et al. (2003). Group prenatal care and preterm 18. Butler, A. S., & Behrman, R. E. (2007). Preterm birth: Causes, birth weight: Results from a matched cohort study at public consequences, and prevention. Washington DC: Institute of clinics. Obstetrics and Gynecology, 102, 1051–1057. Medicine, National Academy Press. 30. Ickovics, J. R., Kershaw, T. S., Westdahl, C., et al. (2007). Group 19. Goyal, N. K., Hall, E. S., Meinzen-Derr, J. K., Kahn, R. S., Short, prenatal care and perinatal outcomes: A randomized controlled J. A., Van Ginkel, J. B., et al. (2013). Dosage effect of prenatal trial. Obstetrics and Gynecology, 110, 330–339. home visiting on pregnancy outcomes in at-risk first-time moth- 31. Duggan, A., Minkovitz, C. S., Chaffin, M., et al. (2013). Creating ers. Pediatrics, 132, S118–S125. a national home visiting research network. Pediatrics, 132(Sup- 20. Roman, L., Raffo, J. E., Zhu, Q., & Meghea, C. (2014). A plement 2), S82–S89. statewide medicaid enhanced prenatal care program impact on http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Maternal and Child Health Journal Pubmed Central

Pathways Community Care Coordination in Low Birth Weight Prevention

Maternal and Child Health Journal , Volume 19 (3) – Aug 20, 2014

Loading next page...
 
/lp/pubmed-central/pathways-community-care-coordination-in-low-birth-weight-prevention-P7kyg3AaAZ

References (40)

Publisher
Pubmed Central
Copyright
© The Author(s) 2014
ISSN
1092-7875
eISSN
1573-6628
DOI
10.1007/s10995-014-1554-4
Publisher site
See Article on Publisher Site

Abstract

Matern Child Health J (2015) 19:643–650 DOI 10.1007/s10995-014-1554-4 Pathways Community Care Coordination in Low Birth Weight Prevention • • Sarah Redding Elizabeth Conrey • • • Kyle Porter John Paulson Karen Hughes Mark Redding Published online: 20 August 2014 The Author(s) 2014. This article is published with open access at Springerlink.com Abstract The evidence is limited on the effectiveness of Women participating in CHAP and having a live birth in home visiting care coordination in addressing poor birth 2001 through 2004 constituted the intervention group. outcome, including low birth weight (LBW). The Com- Using birth certificate records, each CHAP birth was munity Health Access Project (CHAP) utilizes community matched through propensity score to a control birth from health workers (CHWs) to identify women at risk of having the same census tract and year. Logistic regression was poor birth outcomes, connect them to health and social used to examine the association of CHAP participation services, and track each identified health or social issue to a with LBW while controlling for risk factors for LBW. We measurable completion. CHWs are trained individuals identified 115 CHAP clients and 115 control births. Among from the same highest risk communities. The CHAP the intervention group there were seven LBW births Pathways Model is used to track each maternal health and (6.1 %) compared with 15 (13.0 %) among non-CHAP social service need to resolution and CHWs are paid based clients. The adjusted odds ratio for LBW was 0.35 (95 % upon outcomes. We evaluated the impact of the CHAP confidence interval, 0.12–0.96) among CHAP clients. This Pathways program on LBW in an urban Ohio community. study provides evidence that structured community care coordination coupled with tracking and payment for out- comes may reduce LBW birth among high-risk women. S. Redding  M. Redding (&) Community Health Access Project, Columbus, OH, USA Keywords Low birth weight prevention  Community e-mail: reddingmark@att.net health worker  Community care coordination  Social S. Redding e-mail: Sarah.redding@me.com determinants of health  Pay for performance  Home visiting E. Conrey State Maternal and Child Health Epidemiologist CDC Assignee, Ohio Department of Health, Columbus, OH, USA e-mail: ElizabethJConrey@odh.ohio.gov Introduction K. Porter Infant mortality rates are used as an indicator for the health Ohio State University Center for Biostatistics, Columbus, OH, of a community. To prevent infant deaths, mothers need to USA e-mail: kyle.porter@osumc.edu be healthy, live in a safe environment, and have access to quality care. Reducing low birth weight (LBW) and pre- J. Paulson mature births has been identified as a key strategy to Center for Public Health Statistics and Informatics, Ohio decrease infant mortality [1]. While infant mortality rates Department of Health, Columbus, OH, USA e-mail: John.paulson@odh.ohio.gov in the US have improved over the past decades, they have been stagnant in Ohio. In fact, Ohio ranked second worst K. Hughes for black infant mortality among all states, and fourth worst Division of Family and Community Health Services, Ohio for overall infant mortality in 2010 [2, 3]. Nationally, Department of Health, Columbus, OH, USA despite overall improvements, the 2011 Centers for Disease e-mail: Karen.Hughes@odh.ohio.gov 123 644 Matern Child Health J (2015) 19:643–650 Control (CDC) Health Disparities and Inequalities Report Secondary objectives were a comparison of adequacy of showed that large disparities in infant mortality rates per- prenatal care and a cost savings evaluation. sist [4]. Strategies that incorporate the community and directly reach out to women at greatest risk for poor birth outcomes Methods may help communities move towards health equality. Home visiting services are one strategy used to improve The CHAP Intervention birth outcomes and have received increased attention and focus on providing evidence-based services to vulnerable Initially, 4 years of birth certificate data were used to children and families through the Affordable Care Act and identify where the LBW births were occurring in Richland the Maternal, Infant, and Early Childhood Home Visiting County. Eligibility for participation in CHAP was based on (MIECHV) program [5]. Although home visiting has been residence in a census tract with high LBW and poverty shown to be effective in impacting parent behaviors, child rates. Seven census tracts comprised the program-eligible cognitive outcomes and maternal life course, the impact on communities; two of these census tracts (6 and 7) repre- birth outcomes is not as clearly evident [6, 7]. sented only six percent of the county population, but The Community Health Access Project (CHAP) is a almost thirty percent of all county LBW births. nonprofit, community based organization that has been The CHWs that provided home visiting services here providing care coordination services in Richland County, were hired from the program-eligible communities and Ohio since 1999. CHAP utilizes community health workers trained at the local community college. CHAP developed (CHWs) to identify women at risk of having poor birth an extensive CHW-specific training curriculum that was outcomes, connect them to health and social services, and delivered for college credit. CHWs were supervised by track each identified issue to a measurable completion. either a registered nurse or physician. CHAP’s intensive home visiting model uses an account- Community health workers (CHWs) functioned as ability tool called Pathways [8, 9]. A Pathway addresses community care coordinators, not providers of direct ser- clearly defined actions towards problem resolution and is vices, and assisted participants to overcome barriers faced not considered complete until a measurable outcome is in obtaining necessary health or social services. CHAP achieved. One participant may be assigned to many dif- developed checklists to be used at each face-to-face home ferent Pathways depending on the problems identified visit encounter between the client and the CHW. A ‘‘yes’’ during the initial interview and subsequent home visits answer to certain questions triggered the initiation of a [10]. As in most communities, Richland County had geo- defined Pathway. For example, if a client answered ‘‘yes’’ to the question—‘‘Do you need a medical home?’’—then a graphic areas of health inequality. CHAP used a mapping strategy to determine the census tracts where the unfavor- Medical Home Pathway was initiated. able birth outcomes were disproportionately occurring. The Pathways are tools to track each identified health or infant mortality rates in Richland County from 2001 to social issue through to a measurable completion or out- 2005 were 6.7 infant deaths per 1,000 live births for white come; typically confirmation that the client actually women, and 17.3 for African-American women [2]. received the medical or social service is required. The The impact of CHWs has been difficult to document. Medical Home Pathway tracks the participant’s connection The Agency for Healthcare Research and Quality (AHRQ) to an ongoing source of primary care and is not docu- released a report on the outcomes of CHW interventions in mented as complete until the CHW confirms that the client 2009, based on 15 different programs, which showed has a medical home. If the client does not connect with a minimal impact on birth outcomes [11]. The CHAP model medical home, then the Pathway is closed as ‘‘finished differs from those programs previously studied in that an incomplete’’; recording that the desired outcome was not accountability measurement tool—Pathways—was used to achieved. In a similar fashion, the Pregnancy Pathway track each health or social issue a pregnant client faced confirms the connection to and maintenance of prenatal through to a measurable completion. Additionally, con- care and is not complete until delivery of a viable normal tracts were developed with funders to pay for completed birth weight infant (Fig. 1). A full description of the model Pathways or outcomes [8, 9]. can be found in the Agency for Healthcare Research and We evaluated if LBW would be reduced when women at Quality ‘‘Connecting Those at Risk to Care’’ publications risk of having a LBW infant were provided with intensive [8, 9]. home visiting and community based care coordination by Contracts were developed between funders and CHAP CHWs, and Pathways were used to document outcomes. with payment tied to specific Pathway benchmarks and The primary objective was to compare the adjusted odds of Pathway completions. In addition, the CHWs received LBW between CHAP recipients and non-CHAP recipients. incentive payments if they completed a designated number 123 Matern Child Health J (2015) 19:643–650 645 Fig. 1 Pregnancy pathway of Pathways. This strategy improved the accuracy of factors for LBW than the general population within each Pathway tracking within the agency, because monitoring census tract, propensity score matching was performed to was occurring both programmatically and operationally. select a comparison group with a similar distribution of risk factors from Ohio vital statistics records [12, 13] The Study Population and Data Sources matching process consisted of estimating propensity scores using a logistic regression model, then matching CHAP The study was limited to census tracts in which at least five clients to controls with similar propensity scores. The women received CHAP care coordination and gave birth in logistic regression model was fit to the data from eligible the time period 2001–2004 (tracts 3, 4, 5, 6, 7, 8, 10 in mothers, with CHAP client (yes/no) as the dependent Richland County, Ohio). Only singleton births were variable. Predictors of CHAP enrollment in this model included in the analysis. CHAP medical records were included mother’s age (\16, 16–18, [18), race (African- identified for all women meeting the study criteria and all American or white), education (if [18 years old: less than were successfully matched to an Ohio live birth record. high school, high school graduate, one or more years of Data on the mother’s trimester of enrollment into CHAP college), marital status, census tract, and delivery year. All and the number of Pathways initiated were extracted from two-way interactions were tested; none were statistically CHAP records. All other study data were from Ohio vital significant and all were dropped from the model. From this statistics records. Because CHAP clients had more risk logistic regression model, a score reflecting the probability 123 646 Matern Child Health J (2015) 19:643–650 of CHAP enrollment was estimated for each eligible Table 1 Characteristics of community health access project (CHAP) clients, all non-CHAP mothers* identified from birth certificates, and mother. matched controls Next, the propensity score was used in an optimal matching algorithm to match each CHAP recipient to one CHAP Matched All non-CHAP* clients controls Births (pre- control. Optimal matching is known to be superior to (n = 115) (n = 115) matching)* nearest-neighbor or ‘‘greedy’’ matching [14]. Exact mat- (n = 1,443) ches for county and delivery year were required. Age This study was exempted by the Ohio Department of \16 16 (13.9 %) 10 (8.7 %) 36 (2.5 %) Health Institutional Review Board and conducted in accord 16–18 13 (11.3 %) 13 (11.3 %) 122 (8.5 %) with prevailing ethical principles. [18 86 (74.8 %) 92 (80.0 %) 1,285 (89.0 %) Race Analysis African- 78 (67.8 %) 80 (69.6 %) 325 (22.5 %) American To evaluate the CHAP program’s impact on LBW, logistic White 37 (32.2 %) 35 (30.4 %) 1,118 (77.5 %) regression models were fit to the LBW outcome. First, the Education unadjusted LBW odds ratio for CHAP mothers versus non- Less than HS 28 (32.6 %) 29 (31.5 %) 220 (17.1 %) CHAP mothers was calculated. Then, two multivariate High school 36 (41.9 %) 40 (43.5 %) 628 (48.9 %) logistic regression models were fit, the primary with only graduate non-modifiable risk factors and a secondary also including Any college 22 (25.6 %) 23 (25.0 %) 436 (34.0 %) factors modifiable by the CHAP program. Multivariable Marital status adjustment was also appropriate, as propensity score Married 17 (14.8 %) 19 (16.5 %) 661 (45.8 %) matching and multivariable adjustment are often used in Not married 98 (85.2 %) 96 (83.5 %) 782 (52.2 %) combination to reduce potential bias [15]. The primary Census tract model was ‘‘non-modifiable only’’ because it is less likely 3 18 (15.7 %) 20 (17.4 %) 110 (7.6 %) to over adjust for the mediating effects of CHAP inter- 4 8 (7.0 %) 5 (4.4 %) 188 (13.0 %) vention. Covariates included in the primary model were the 5 20 (17.4 %) 17 (14.8 %) 211 (14.6 %) propensity score matching variables (mother’s age, race, 6 51 (21.7 %) 26 (22.6 %) 226 (15.7 %) education, marital status, census tract, and delivery year), 7 31 (27.0 %) 34 (29.6 %) 159 (11.0 %) previous preterm or LBW delivery and tobacco use during 8 5 (4.4 %) 6 (5.2 %) 159 (11.0 %) pregnancy (none vs. any throughout pregnancy, thus non- 10 8 (7.0 %) 7 (6.1 %) 390 (27.0 %) modifiable). Other risk factors considered for inclusion in the secondary model were hypertension (chronic or preg- Year of birth nancy-associated), eclampsia, incompetent cervix, renal 2001 44 (38.3 %) 44 (38.3 %) 383 (26.5 %) disease, and uterine bleeding. However, only hypertension 2002 34 (29.6 %) 34 (29.6 %) 347 (24.1 %) was added to the secondary model because there were very 2003 26 (22.6 %) 26 (22.6 %) 354 (24.5 %) few occurrences of the other conditions. 2004 11 (9.6 %) 11 (9.6 %) 359 (24.9 %) To evaluate the secondary objective, the CHAP pro- Tobacco use 45 (39.1 %) 43 (37.4 %) 528 (36.6 %) gram’s impact on the adequacy of prenatal visits, an Previous preterm 3 (2.6 %) 2 (1.7 %) 11 (0.8 %) or LBW ordinal logistic regression model was fit to adequate pre- delivery natal visits versus less than adequate prenatal visits based Hypertension 2 (1.7 %) 4 (3.5 %) 43 (3.0 %) on the Kotelchuck index [16]. A logistic regression model Eclampsia 2 (1.7 %) 2 (1.7 %) 16 (1.1 %) was also fit to first trimester prenatal care versus other than first trimester prenatal care. * Single birth from census tract 3, 4, 5, 6, 7, 8, or 10 The number of LBW births prevented was estimated by Among mothers [18 years of age subtracting the observed number of LBW deliveries from Defined as any tobacco use during pregnancy reported on birth the number expected in the study population if there had certificate been no CHAP intervention. The calculation required the Chronic or pregnancy-related relative risk, for which the odds ratio was considered a sufficient estimate (unadjusted relative risk = 0.43 and 0:5 þ OR  0:5 unadjusted odds ratio = 0.47). The estimate was taken from the model adjusting for both hypertension (modifi- which is the fraction of study women in the non-CHAP able) and non-modifiable risk-factors. First, the fraction of group ? CHAP risk relative to non-CHAP (CHAP odds LBW births not prevented by CHAP was calculated as ratio) multiplied by the fraction in the CHAP group. Next, 123 Matern Child Health J (2015) 19:643–650 647 the observed number of LBW births was divided by this common non-medical Pathways initiated were Employ- fraction and rounded to the nearest integer. This method ment (52 %), Adult Education (50 %), Smoking Cessation was repeated using the lower and upper confidence limits (39 %), Food Security (30 %), and Housing (27 %). Two of the odds ratio to obtain the confidence interval. This major barriers that were identified to completion of Path- method is equivalent to multiplying the preventable frac- ways included transportation and limited community tion (1—odds ratio) by the fraction treated, subtracting that resources for non-medical issues. from one and multiplying the reciprocal by the number of Women enrolled in CHAP care coordination from 2001 observed events [17]. through 2004 had significantly lower adjusted odds of To estimate the potential cost savings of the CHAP experiencing a low-birth weight delivery than non-CHAP program, we first estimated the number of LBW births women [adjusted odds ratio = 0.36, 95 % CI (0.12, 0.96)] avoided using the method described above. We then esti- (Table 2). There were no significant differences between mated the average cost of delivering the CHAP interven- the adjusted odds of the adequacy of prenatal visits or the tion per client by evaluating the cost per Pathway, cost per timing of the first prenatal visit between CHAP participants client, and the amount paid to CHAP per number of and non-CHAP mothers. This finding is different from pregnant clients within grant and service contracts. The other home visiting studies that have shown a dosage effect greatest cost of the program was time spent by a CHW to of prenatal home visiting in at-risk women [19, 20]. provide care coordination and the amount of time spent by Fifty-six percent of clients in this study entered CHAP a CHW was primarily driven by trimester of entry into in the first trimester of pregnancy, 20 % in the second CHAP. trimester and 24 % in the third trimester. The estimated To evaluate cost savings from LBW births averted by cost to provide Pathways community care coordination by CHAP participation, we applied the average excess LBW CHAP in the time period studied averaged $751 per costs provided in the 2006 Institute of Medicine (IOM) pregnant client. An estimated 10 LBW births (1 prevented report [18] to our estimate of LBW births averted. Per per 11.5 participants) were prevented by participation in IOM, in the first year of life, excess medical expenses per the CHAP program from 2001 through 2004 (95 % LBW infant are $29,000 and long term costs (including CI = 1, 17). The cost savings in the first year of life, for maternal costs, early intervention, special education and each dollar invested in Pathways based community care lost household and labor market productivity) are $48,275. coordination was $3.36, and the long term cost savings was The dollars saved per dollar invested was calculated by $5.59 for each dollar invested. dividing the total cost savings for one prevented LBW infant by the total cost to serve enough pregnant women with Pathways focused care coordination. Discussion Pregnant women who participated in CHAP, a structured Results community-based care coordination program provided by CHWs and coupled with Pathways tracking and payment Characteristics of CHAP participants and non-participant for outcomes, had a significantly lower probability of controls are summarized in Table 1. The CHAP and non- delivering a LBW infant. CHAP participants living in the CHAP groups did not differ significantly (p \ 0.05) in any targeted census tracts were at an increased risk for poor of the propensity score variables; the groups are within birth outcomes compared to the general population— 2.6 % points for all levels of all propensity score variables 67.8 % African-American, 25.2 % age 18 or younger, with the exception of age, which had a 5.2 % point dif- 85.2 % unmarried, and 39.1 % tobacco users. A challenge ference. There were no reported cases of incompetent to determining the effectiveness of CHW interventions has cervix, uterine bleeding, or renal disease in either group. been identifying a valid control group that effectively A total of 653 Pathways were initiated for the CHAP accounts for social determinants and their impact on out- participants, and all 115 women in this study finished a comes [21, 22]. Use of an optimal matching algorithm Pregnancy Pathway (7 were finished incomplete due to using propensity scores allowed each CHAP recipient to be LBW). Including the Pregnancy Pathway, CHAP partici- matched with one control and supported estimation of the pants had an average of 5.6 Pathways tracked for health number of LBW births prevented. and social issues that were identified during the pregnancy Areas of health inequalities—whether related to birth and postpartum period. 102 Postpartum and Family Plan- outcomes or chronic diseases—can be easily mapped in ning Pathways were completed for participants, confirming communities. This study demonstrates the value of iden- that 89 % of women attended their postpartum appoint- tifying communities with disparately poor health outcomes ments and were using a family planning method. The most and directly reaching out to individuals within those 123 648 Matern Child Health J (2015) 19:643–650 Table 2 Odds ratios and 95 % a Variable Unadjusted Primary model: adjusts for non- Secondary model: adjusts for confidence intervals for pre- b c modifiable risk-factor covariates all risk-factor covariates term birth CHAP versus 0.43 (0.16, 1.07) 0.36 (0.12, 0.96) 0.37 (0.12, 1.02) non-CHAP Age \16 versus [18 1.58 (0.40, 6.28) 1.17 (0.42, 6.70) Census tract comparisons 16–18 versus [18 2.13 (0.66, 6.85) 2.11 (0.65, 6.84) excluded African-American 1.13 (0.35, 3.70) 0.93 (0.28, 3.09) Mother’s age (\16, 16–18, versus White [18), race (African-American, Not married 3.06 (0.87, 10.0) 4.11 (1.06, 15.92) white), marital status, census versus married tract, previous preterm or LBW delivery, tobacco use at any Previous preterm 3.06 (0.50, 18.52) 3.44 (0.55, 21.43) time during pregnancy (y/n) or LBW delivery All from primary model and additionally hypertension Tobacco use 4.76 (1.92, 11.84) 5.09 (2.01, 12.87) (chronic and/or pregnancy- Hypertension 6.25 (0.91, 43.16) associated) communities, engaging them through care coordination, important part of the care plan, documentation, and connecting them to health and social service interventions, reporting in this study. and measuring the results through an accountable mea- There were several limitations in this study. First, surement tool. although data was collected over a 4-year time period, the Community health workers perform their work by total number of women in the CHAP intervention group approaching the whole person—and take into consideration was small, reflecting the size of program enrollment within their social, environmental, psychological and health needs the targeted census tracts over the time period studied. A in order to impact health outcomes. This is evidenced by larger sample size would have provided more precise the additional Pathways initiated by CHWs in this study for estimates of odds ratios and more power to detect signifi- issues related to food security, housing, transportation, cant differences in all models. Second, there was no ran- employment, and education. These additional Pathways dom assignment to CHAP intervention or control. had to be addressed in coordination with preventive health Although we attempted to control for bias as much as care needs and consideration of the client’s priorities of possible through propensity score matching and covariate care. Health and social service siloes exist in communities, adjustment, some selection bias may remain. Additional and individuals living in poverty often face barriers in evaluations, with randomized group assignments, larger accessing these critical services. The community-based numbers of participants, and in different locations are care coordinator serves an important role on the healthcare needed to replicate and confirm our findings. Third, the team because of their trusted relationship with the client. evaluation was limited by the vital statistics records on They are able to identify key non-medical issues and are what cofounders and outcomes we could study. For skilled in navigating the fragmented health and social example, prenatal smoking is potentially modifiable service systems. through CHAP with a Pathway that included specific Some social determinants of health can be addressed at education and support to help patients reduce or quit the population level—such as safe drinking water, smoking smoking; however smoking status by trimester was not in public places, elimination of food deserts and safe standard documentation on the Ohio birth certificate. sidewalks—but individually addressable social determi- Future work should control for first trimester smoking nants also represent a significant intervention opportunity. status and other factors related to low birth weight. Finally, Housing, education, employment, food security, and many the evaluation was limited by the quality of birth certificate other critical issues can be identified and addressed with data, which is shown to generally be specific, but not effective and accountable care coordination to improve sensitive, as a source of maternal complications [23, 24]. In individual progress, reduce stress, and improve health for contrast, birth weight data from the birth certificate has those individuals at greatest risk. been shown to be more reliable [25]. The CHAP Pathways Model provided the measurement CHAP may reduce LBW delivery among high risk tool to monitor successful connections to both health and women through multiple mechanisms. As there were no social services. Pathways were developed as the pay-for- differences in prenatal care initiation between groups, performance model for CHAP’s contracts and were an improvement in early prenatal care does not appear to be 123 Matern Child Health J (2015) 19:643–650 649 one, and this finding is consistent with other studies [26]. programs’’ [1]. This study shows that structured commu- However, factors besides medical care are known to impact nity-based care coordination coupled with standardized and health outcomes and models of care that address both accountable tracking tools and payment for outcomes may medical and social factors show promise in reducing LBW reduce LBW delivery among high-risk pregnant women. [27–30]. The Pathways Model allows for targeting the diversity of This study represents our initial experience with using needs across racial, ethnic and other sociodemographic the Pathways Model to quantify and track care coordina- distinctions. Identifying communities with disparately poor tion provided to high risk pregnant women. Since the health outcomes and ensuring the connection of residents model’s inception, effort has been placed on refining the to health and social programs can potentially reduce per- measurement and tracking process of the Pathways. It was sistent inequalities in health. not possible in this study to identify which Pathways spe- Acknowledgments We would like to acknowledge the CHAP cifically led to improved birth outcomes. Newer technology community health workers—their wisdom and understanding of the for Pathway tracking has remedied that and can support community has always led the way; The Osteopathic Heritage future research. CHAP participants were initially identified Foundation who supported the initial development and pilot of the as being at increased risk by where they lived (identified Pathways Model; Dan Wertenberger the Executive Director of CHAP; Celia Flinn, MD who continues to be a key partner in census tracts), but now we have the capability to monitor designing the intervention and supporting the program; Kathryn risk throughout the care coordination period. Our pre- Meagly, MPH who served as a volunteer collecting data; Cynthia liminary study can be incorporated into the larger move- Shellhaas, MD, MPH, who’s clinical expertise provided guidance ment to create a national home visiting research network during the analytic phase; and Wallace L. Alward, MD who provided editorial work and statistical review. We would also like to thank the that works to promote the translation of research findings Centers for Disease Control and Prevention, Maternal and Child into policy and practice [31]. Health Epidemiology Team for their critical assistance in the devel- Starting from an American Academy of Pediatrics— opment of this article. The findings and conclusions in this report are Community Access to Child Health (CATCH) Grant in those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. 2001—the Pathways Model was further developed to embrace multiple care coordination agencies within a ser- vice region. The Pathways Community HUB Model is Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, dis- designed to identify the most at-risk individuals in a tribution, and reproduction in any medium, provided the original community, connect them to evidence-based interventions, author(s) and the source are credited. and measure the results [8, 10]. The HUB Model was developed and piloted by CHAP in 2004 in Richland County, Ohio based on the success of these initial findings. References The model was recognized by the Agency for Healthcare Research and Quality (AHRQ) Innovations Exchange and a 1. McCormick, M. C. (1985). The contribution of low birth weight to infant mortality and childhood morbidity. New England learning network was established to further study the Journal of Medicine, 312(2), 82–90. model [8, 9]. The Pathways Community HUB does not 2. Ohio Department of Health. Latest Ohio Infant Mortality Data. directly provide care coordination services, but subcon- http://www.odh.ohio.gov/odhprograms/cfhs/octpim/latestoimd. tracts with care coordination agencies serving vulnerable aspx. 3. MacDorman, M. F., & Mathews, T. H. (2011). Understanding populations in the community. The community HUB works racial and ethnic disparities in U.S. infant mortality rates. NCHS to coordinate and track progress for all of the agencies Data Brief, 74, 1–8. within a community providing care coordination. The HUB 4. CDC Health Disparities and Inequalities Report. (2011). Sup- serves to register and collect focused data on each client plement (p. 2011). United States: Centers for Disease Control. 5. Wakefield, M. K., & Sheldon, G. H. (2013). Special supplement served using common Pathways to track quality and out- on the maternal, infant and early childhood home Visiting (MI- comes. This model eliminates duplication of care coordi- ECHV) program. Foreword. Pediatrics, 132(Suppl 2), S57–S58. nation and provides standard quality measurements, 6. Filene, J. H., Kaminski, J. W., Valle, L. A., & Cachat, P. (2013). allowing care coordination agencies to focus on the most Components associated with home visiting program outcomes: A meta-analysis. Pediatrics, 132, S100–S109. vulnerable community members and strive towards 7. Avellar, S. A., & Supplee, L. H. (2013). Effectiveness of home improving overall health outcomes. The Kresge Founda- visiting in improving child health and reducing child maltreat- tion has recently supported an initiative to develop a ment. Pediatrics, 132, S90–S99. standard approach for certification of communities utilizing 8. Connecting Those at Risk to Care—The Quick Start Guide to Developing Community Care Coordination Pathways. http:// the Pathways Community HUB Model to assure consistent www.innovations.ahrq.gov/guide/quickstartguide/commhub_ quality of care coordination. quickstart.pdf. As stated by CDC, health disparities ‘‘must be addressed 9. Connecting Those at Risk to Care—A guide to building a com- with intervention strategies related to both health and social munity ‘‘HUB’’ to promote a system of collaboration, 123 650 Matern Child Health J (2015) 19:643–650 accountability, and improved outcomes. http://www.innovations. birth outcomes. JAMA Pediatrics. doi:10.1001/jamapediatrics. ahrq.gov/guide/HUBManual/CommunityHUBManual.pdf. 2013.4347. Published online January 6, 2014. 10. Pathways Manual, Community Health Access Project. http:// 21. Livingood, W. C., Brady, C., Pierce, K., Atrash, H., Hou, T., & chap-ohio.net/press/wp-content/uploads/2010/09/PathwaysMan Bryant, T., I. I. I. (2010). Impact of pre-conception health care: ual1.pdf. evaluation of a social determinants focused intervention. Mater- 11. Viswanathan M KJ, Nishikawa B, Morgan LC, Thieda P, Hon- nal and Child Health Journal, 14, 382–391. eycutt A, Lohr KN, Jonas D. Outcomes of Community Health 22. Salihu, H. M., Mbah, A. K., Jeffers, D., Alio, A. P., & Berry, L. Worker Interventions Evidence Report/Technology Assessment (2009). Healthy start program and feto-infant morbidity out- No. 181. Rockville MD: Agency for Healthcare Research and comes: evaluation of program effectiveness. Maternal and Child Quality; 2009 June 2009. Health Journal, 13, 56–65. 12. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the 23. Bailit, J. L. (2010). Rates of labor induction without medical propensity score in observational studies for causal effects. Bio- indication are overestimated when derived from birth certificate metrika, 70, 41–55. data. Am J Obstetrics and Gynecol, 203(3), 269.e1–269.e3. 13. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a control- 24. Piper, J., Mitchel, E., Snowden, M., et al. (1993). Validation of group using multivariate matched sampling methods that incor- 1989 Tennessee birth certificates using maternal and newborn porate the propensity score. American Statistician, 39, 33–38. hospital records. American Journal of Epidemiology, 137, 14. Rosenbaum, P. R. (1989). Optimal matching for observational 758–768. studies. Journal of American Statistical Association, 84, 25. Northam, S., & Knapp, T. R. (2006). The reliability and validity 1024–1032. of birth certificates. JOGNN, 35, 3–12. 15. Rubin, D. B., & Thomas, N. (2000). Combining propensity score 26. Lu, M. C., et al. (2003). Preventing low birth weight: Is prenatal matching with additional adjustments for prognostic covariates. care the answer? Journal of Maternal-Fetal and Neonatal Med- Journal of American Statistical Association, 95, 573–585. icine, 13, 362–380. 16. Kotelchuck, M. (1994). An evaluation of the Kessner adequacy of 27. Krans, E. E., & Davis, M. M. (2012). Preventing low birth prenatal care index and a proposed adequacy of prenatal care weight: 25 years, prenatal risk, and the failure to reinvent pre- utilization index. American Journal of Public Health, 84, natal care. American Journal of Obstetrics and Gynecology, 1414–1420. 206(5), 398–403. 17. Meddings, D. R., Ronald, L. A., Marion, S., Pinera, J. F., & 28. Lee, E., et al. (2009). Reducing low birth weight through home Oppliger, A. (2004). Cost effectiveness of a latrine revision visitation: A randomized controlled trial. American Journal of programme in Kabul, Afghanistan. Bulletin of the World Health Preventive Medicine, 36, 154–160. Organization, 82, 281–289. 29. Ickovics, J. R., et al. (2003). Group prenatal care and preterm 18. Butler, A. S., & Behrman, R. E. (2007). Preterm birth: Causes, birth weight: Results from a matched cohort study at public consequences, and prevention. Washington DC: Institute of clinics. Obstetrics and Gynecology, 102, 1051–1057. Medicine, National Academy Press. 30. Ickovics, J. R., Kershaw, T. S., Westdahl, C., et al. (2007). Group 19. Goyal, N. K., Hall, E. S., Meinzen-Derr, J. K., Kahn, R. S., Short, prenatal care and perinatal outcomes: A randomized controlled J. A., Van Ginkel, J. B., et al. (2013). Dosage effect of prenatal trial. Obstetrics and Gynecology, 110, 330–339. home visiting on pregnancy outcomes in at-risk first-time moth- 31. Duggan, A., Minkovitz, C. S., Chaffin, M., et al. (2013). Creating ers. Pediatrics, 132, S118–S125. a national home visiting research network. Pediatrics, 132(Sup- 20. Roman, L., Raffo, J. E., Zhu, Q., & Meghea, C. (2014). A plement 2), S82–S89. statewide medicaid enhanced prenatal care program impact on

Journal

Maternal and Child Health JournalPubmed Central

Published: Aug 20, 2014

There are no references for this article.