An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Francisco A. Garcia, MD, MPH, Ada M. Wilkinson-Lee, Patricia M. Herman, ND, PhD, Thomas Ball, MD, Alexandra M. Armenta, BA, Rick A. Rehfeld, MS, and Gery W. Ryan, MA, PhD.
Author Information and AffiliationsWomen from medically underserved minority and poor communities who historically have limited access to preventive health services may not have the basic information needed to fully benefit from these offerings. Such women may lack the necessary tools to effectively identify personal health care needs, take advantage of health services, or participate equitably in health care decision-making. Our study tests the effectiveness of a Medicaid health plan that offers new members the services of community health workers (CHWs) who use a community-developed curriculum and function as navigation aides. The CHWs assist patients to make informed decisions about the utilization of covered preventive health services and engagement with their health care providers.
The following was the overarching question addressed by this study: Does access to a trained CHW improve newly enrolled Medicaid health plan members' engagement in the health care system?
The study participants consisted of women aged 18 to 39 years who had newly enrolled in a Medicaid health plan serving southern Arizona between March 15, 2015, and March 14, 2016. All eligible health plan members were randomized 2:1 to be offered support from a CHW in addition to the standard services of the health plan or to receive only standard health plan services. We used the health plan's call center to contact those randomized to the intervention. We based the primary and secondary outcomes on analysis of 6 months' health care utilization data captured for all eligible members, and our primary outcome was a visit with a primary care provider (PCP). We also assessed patients' informed choice via a baseline survey completed by all members who worked with a CHW and a 6-month survey fielded to all members.
The health plan enrolled 2267 eligible new members during the study; all eligible members were included in the study, and 1521 were randomized to the offer of CHW services. Of those enrolled, 78% were unreachable by the call center to complete the offer and another 5% became unreachable after expressing initial interest and before they could meet with a CHW. Of reached members, 82% said they were interested in working with a CHW and 62% worked with a CHW. Following intent to treat we found no significant difference in our primary outcome (at least 1 PCP visit) between those randomized to the offer (58.2%) and those randomized to control (57.1%). However, although not designed to directly test the effectiveness of CHW services themselves, our results suggest that support provided by the CHW might have a beneficial impact in terms of accessing primary care services. A significantly larger proportion of the group that received the CHW services had 1 or more primary care visits (71.7%) compared with those randomized to the offer who did not work with a CHW (56.6%) or the control group (57.1%).
The study findings do not support the effectiveness of a health plan's offer of CHW services to its members. The direct telephone-based offer of CHW services by a Medicaid health plan to new members is not effective largely due to problems with the reachability of this low-income, highly mobile Medicaid population. The lack of effect may be due not to CHW services, which may themselves have a positive impact, but instead to the way these services were offered.
The main limitation of this study was that the health plan call center was unable to consistently connect with its membership using available member contact information.
Under the provisions of the Patient Protection and Affordable Care Act, many previously uninsured individuals now have access to health care coverage, including through Medicaid. While obtaining health care coverage can improve health outcomes,118,119 the newly insured still face challenges in using the health care system effectively and efficiently. The challenges begin with health care plans' inability to maintain contact with their Medicaid populations; problems often cited include quickly outdated contact information, unstable housing situations, and the reliance on prepaid cell phones that run out of minutes.137 The newly insured face challenges in finding and connecting with a primary care provider, and preventable hospitalizations and emergency department (ED) use have increased, especially in Medicaid populations.120-122 In addition, medically underserved minority and poor communities that historically have had limited access to health care services are largely unprepared to maximally benefit from evidence-based approaches to prevention and wellness.
Many individuals and communities lack the necessary tools to prioritize their own health care needs, take advantage of these health services, or effectively participate in their own health decision-making. For example, low health literacy has been found to be 3 to 5 times more common in minority populations, especially Hispanic adults.1-3 Low health literacy is associated with increased hospitalizations and ED use, lower use of preventive measures (eg, mammography, influenza vaccines), poorer ability to take medications appropriately, and poorer overall health status and higher mortality.1,7,8 Recent immigrants are less likely to have a usual source of care1,2,4 and both Blacks and Hispanics are less likely than Whites to identify a doctor's office as their usual source of care.15,48,49,63 Having a usual source of care, and especially a regular doctor,16,19,20 is associated with better access to care,13,16,20 lower unmet needs,20 and better use of preventive services10-12,14,15,17-19,21 for both parents and their children.125-127 Increasing community-level capacity for effective and efficient health care utilization is essential to narrow the most pressing health disparities and achieve the Healthy People 2020 objectives.128
The stubborn health disparities that impact ethnically diverse and low-income populations are exacerbated by a complex array of social determinants and by fundamentally dysfunctional interactions with the health care system. Even after accounting for educational attainment, communication with health care providers is generally poor, and racial/ethnic and other linguistic minorities experience this phenomenon more acutely than White non-Hispanic individuals.129 Racial, ethnic, and gender discordance may also exacerbate the already uneven power dynamic that is inherent to the doctor-patient relationship. Furthermore, limited access to evidence-based health information that is tailored (in terms of gender, culture, language, and literacy level) to the needs of the individual effectively disenfranchises the communities that may have the most pressing health needs. In total, these factors lead to generally low rates of participation in medical decision-making and low levels of satisfaction among low-income and ethnic minority patients.130
The present intervention was designed to address this problem by testing the effectiveness of community health workers (CHWs) trained according to a community-developed curriculum. These CHWs function as personal educational and navigation aides to effectively empower low-income women as consumers of health care and to allow them to take charge of their health. We tested a Medicaid health plan offer of a community-developed targeted intervention that employs CHWs as aides to newly Medicaid-insured women. These women face the challenges of engaging with their health care and of making highly personal and socioculturally embedded decisions about reproductive health-related preventive services (eg, contraception, sexually transmitted infections [STIs], and cancer screening). This CHW-based curriculum has arisen organically from the community and offers individualized education and empowerment-oriented support to new enrollees in a large regional Medicaid health plan serving high-priority (low-income, largely Hispanic, rural, US-Mexican border) populations that have generally poor access to health care services and whose health-seeking behaviors are not well understood. Empowerment helps communities and individuals develop opportunities, capacities, and tools that benefit them, and it can ensure that communities can mobilize targeted populations to obtain needed health resources by fostering awareness of a given problem.136
Rather than focusing on a single disease entity (eg, diabetes) or intervention (eg, mammography), our CHW intervention aimed to improve health care service utilization in general but targeted women's reproductive services for a specific age group. In southern Arizona, specifically low-income, Medicaid-eligible women use routine preventive reproductive health services at lower rates than their more affluent neighbors. Based on the latest Behavioral Risk Factor Surveillance System (BRFSS) data for Pima County, for example, low-income women are significantly less likely to report cervical cancer screening than higher-income women (88% vs 91%), and among those undergoing cervical cancer screening, low-income women are substantially less likely to get the most sensitive type of screening, which includes molecular high-risk HPV (human papillomavirus) testing (28% vs 37%). Likewise, Medicaid-eligible women in southern Arizona are less likely to undergo STI screening than higher-income women (26% vs 32%).135 The effective translation of evidence-based medical recommendations into personal action for low-income individuals from these communities presents important challenges and opportunities for the health of the nation. CHWs are the essential partners who connect this research to the community in a real and accountable way. As community members and leaders, CHWs have an intimate understanding of information gaps, barriers to services, and the sociocultural context of the end user of this intervention. With appropriate technical assistance, CHWs are able to take existing, often complex, health information and tailor it to the target community and to develop strategies and interventions for its dissemination, adapting it to shared cultural and linguistic experiences.131-134 CHWs are critical partners for the tailoring and delivery of individualized usable evidence-based health information that has a real and tangible impact on the health care utilization of individual women and their families. A systematic review of the studies that implemented a CHW approach by Viswanathan and colleagues (2010) found low or moderate strength of evidence suggesting that CHWs can increase health care utilization for some interventions.72 Additionally, this review addressed existing gaps in the literature that the current study aims to fill by focusing on a low-income population, utilizing a health plan as the organization offering the intervention, and not only measuring health care usage but also evaluating changes in participants' knowledge to clarify the processes of change initiated by CHWs.72
The logic model in Figure 1 shows how access to a CHW may improve health care and health outcomes for new health plan members. This model is based on one developed by researchers performing the systematic review of health literacy for the Institute of Medicine1 and draws on several proposed expanded models of health literacy and on an integrated model of behavioral theory.2,3 We split the health behaviors that are the target of the systematic reviewers' model into those related to an individual's healthy lifestyle (eg, diet, exercise, smoking cessation) and those specific to the use of the health care system, since the second is the target of this intervention.
Logic Model for How Access to a CHW Can Affect Health Behaviors and Outcomes.
The overarching question addressed by this study is the following: Does access to a trained CHW improve newly enrolled Medicaid health plan members' engagement in the health care system? The study had 6 specific research questions:
The engagement of patients and stakeholders is an essential and foundational component of this study. Prior to conceptualizing the intervention, the University of Arizona (UA) academic team had already established a 15-year history of engaging diverse women (consumers not actively engaged with the health care system) living in communities throughout Arizona to address pressing health concerns primarily related to issues of reproductive health, cancer prevention, early childhood development, and access to health care services. This study was conceptualized jointly with the REACH Cervical Cancer Prevention Partnership and the Arizona Community Health Outreach Workers (AzCHOW). The REACH Cervical Cancer Prevention Partnership is a CHW-led collaboration with community representatives, organizational stakeholders, public health entities, and clinicians. We invited representatives from the Partnership and AzCHOW to take part in our advisory board. Additionally, the need for formal representation on the part of University Family Care (UFC; our Medicaid health plan partner) administration, providers, and members/patients was essential. UFC designated 2 individuals who represented these categories to participate on the advisory board. It is also important to note that the research team members were also represented in this CHW-led partnership; however, they acted as resources and facilitators for this community-/consumer-oriented group of volunteers.
Representatives from UFC were fully engaged in all stages of the research and intervention design as well as the implementation plans and were active advisory board members. They played central roles in the research design, particularly regarding the recruitment, randomization, curriculum development, enrollment, and data-sharing protocols. The customer care team also assisted in the development of the educational curriculum, providing guidance to the design team on the intervention content related to UFC policies, procedures, and services. The design of the curriculum required an intimate knowledge of the health care plan services and policies. UFC provided valuable input in addition to revising the content and approving the final version of the CHW intervention content and training design. One issue related to patient engagement that proved to be both a challenge and a benefit was being able to synthesize all the key and relevant information in a brief CHW session. Several stakeholders (eg, patients, providers, UFC, university members, and CHWs) reviewed the content. A particular challenge was balancing the comprehensive input from stakeholders in a succinct and informative manner. Content was developed by the UA research team as well as CHWs with extensive experience in community outreach and education.
In addition, representatives from the UFC team identified eligible health plan members and invited them to participate in the advisory board by phone and by mail. The UFC chief medical officer, customer care manager (a bilingual Hispanic woman who also fit our study age criteria), and customer care call center staff (3 Hispanic women who also fit our study age criteria and 2 who were previously enrolled in Medicaid) were all involved in the development and design of these protocols. They provided assistance and collaborated on the development of the recruitment/invitation letter and script, and they contributed valuable information on effective incentives for members and potential barriers to participation. In addition, they interacted with the Arizona Health Care Cost Containment System (AHCCCS), the Medicaid program in the state of Arizona, to receive approval for all recruitment and intervention materials and content. Participation in our advisory board was completely voluntary. Most of our partners and stakeholders held full-time jobs, and participation required a commitment outside of their work hours. We were always mindful of this and tried to make sure that we held our meetings at convenient times for members and that we made complete use of their time and effort. We held meetings quarterly in the early evenings and at a central location. Openness and transparency were key factors when interacting with patients and/or stakeholders on the advisory board. Members were updated on the progress of each stage of the study. They were looked to for advice and guidance when challenges arose with the project. This type of dialogue and interaction between the board members and research team built a sense of trust and respect.
Through our partners and stakeholders we learned which topics were at the forefront of concern for the community. During the pilot phase of the study, patient members participated in focus groups and provided meaningful feedback that contributed to the educational content of this study. Those members provided us with valuable information and helped us narrow down which specific topics should be highlighted during the CHW educational meetings with members. We also used input from the focus groups to develop the informed patient choice survey (a survey measuring respondent knowledge and empowerment to make their health care decisions), which was administered at baseline (pre) and at 6 months (post) to the group that accepted the invitation to work with a CHW. These groups, along with the interaction with the health care plan, provided additional resources to include in the CHW training materials.
Two major issues faced throughout this project were the ability to make initial contact and to retain contact with study participants. Our advisory board played an instrumental role in helping overcome some of these obstacles. In attempting to contact eligible members, we had a hard time reaching participants due to certain characteristics of our target population. New Medicaid health plan enrollees face important economic and social challenges. These individuals frequently have inconsistent cellular and landline telephone services, and they have relatively poor domicile stability. The board provided feedback and potential tactics on how to contact hard-to-reach participants. Based on 1 suggestion from the advisory board when the project team contacted a potential participant, we verified and updated the participant's contact information to maximize retention of the subject. We also began collecting secondary contact information from those who worked with a CHW should their primary contact information ever become invalid. We implemented some of these ideas and suggestions into our study protocols. Having members of the UFC outreach committee participate in our board was extremely helpful. They were able to share some lessons learned when dealing with this population.
UFC has indicated that the engagement of this PCORI project with its patients and other stakeholders to generate strategies for engaging hard-to-reach populations beneficially impacts the health care plan. The AHCCCS mandates that all of the health care plans have an active mode of gathering information from their stakeholders on strategies to improve the dissemination and contact process. The findings generated by this project will assist UFC from a market segmentation perspective so it may effectively use its resources for its target population. Market segmentation is the process of dividing a market of potential customers into groups, or segments, based on different characteristics. The segments created are composed of consumers who will respond similarly to marketing strategies and who share traits such as similar interests, needs, or locations. Market segmentation also reduces the risk of an unsuccessful or ineffective marketing campaign. When marketers divide a market based on key characteristics and personalize their strategies based on that information, the chance of success is much higher than if they were to create a generic campaign and try to implement it across all segments.
This study used a randomization-to-invitation design to test the offer by a Medicaid health plan to provide CHW services to women newly enrolled in this plan. These CHW services were offered in addition to their standard health plan and were community developed and targeted. This study is not an effectiveness study for CHWs; it is a test of the effectiveness of a health plan offering these CHW services to all newly enrolled female members aged 18 to 39 years—that is, results also depend on the delivery mechanism.
Randomization to invitation offered the following benefits to this study:4 (1) It generated a comparable control group (ie, a control group as identical as possible to the intervention group based on income eligibility, enrollment timeline, recommended screening guidelines for reproductive age preventive services, etc; also, health care utilization data was available for all group participants at 2 time points); (2) it allowed a measure of CHW services acceptance not confounded with agreement to be randomized; and (3) it allowed all new health plan members an equal chance to access a CHW. Since we expected some nonacceptance, we used adjustable biased-coin randomization (starting ratio of 2:1) to ensure a sufficient invited-accepted group and to allow for ethical proportional adjustment as needed during the study.5
This study was uniquely inclusive in that all women aged 18 to 39 years who were newly enrolled in one of Arizona's large Medicaid health plans (UFC, a partner in this study) and who were ambulatory and community dwelling were eligible to participate and were included in the study. Income eligibility requirements for these Medicaid enrollees fall at or below 138% of the federal poverty level for nonpregnant women. According to the health plan, this target population has been uninsured for about a year on average.
Our partner health plan (UFC) serves a geographically vast area encompassing 7 southern Arizona counties, including 4 on the US-Mexican border, and the fifth poorest metropolitan area in the United States (Tucson). The in-person meeting with the CHW for those who accepted the offer of CHW services was delivered in a community setting of the new member's choosing.
The intervention group (IG) in this study received an offer to access a CHW's services. Eligible new Medicaid members randomized to this invitation could choose to accept the offer or not. Therefore, the results indicate the likely impact of a real-world offer of these services by a Medicaid health plan. Our study examined how all 3 groups (IG, IG2, and control group [CG]) fared, because we had access to health care utilization data at 2 time points for all eligible members regardless of the group to which they were assigned. The study evaluated those who chose not to partake in the intervention fare by comparing time point 1 data to time point 2 data based on the health care utilization data available for all IG, IG2, and CG participants.4 All eligible members received services per current standard health plan practice, and the CG received only these services, which included a series of welcome mail communications designed to apprise new members of covered benefits, facilitate selection of a primary care provider, and provide general preventive health resources and information; access to the health plan's online resources and customer care line; and a nurse follow-up call. The health plan made the offer of CHW services to those randomized to the IG through outgoing calls from its call center. The health plan's usual mode of member contact was through mailings, with its call center fielding member-initiated inquiries. However, for this study, and because of past experiences with substantial returned mail, the health plan used this enhanced form of member contact: outgoing calls by the call center.
Up to 3 invitation call attempts were made at varying weekday, evening, and weekend times. When contact was made, a brief verbal overview of the intervention was provided. If the member agreed, their name was immediately sent to the UA study team and the assigned CHW attempted contact to set up an individual confidential appointment. The CHWs made up to 3 weekday call attempts followed by up to 2 evening calls and then up to 2 weekend call attempts before the member was denoted as unreachable.
We designed the CHW curriculum to focus on the services that new health plan members want and need to successfully navigate and utilize the health care system. The CHWs attended a 2-day training program and mandatory, quarterly refresher training sessions9 to teach this curriculum. In particular, in a 30-minute one-on-one presentation and tailored planning session (in English or Spanish), new members were given information about women's reproductive health-related preventive services and tools to assist their navigation of the health care system (eg, a tip sheet for interacting with the health plan; a guide to identifying/connecting with a primary care provider; a summary of covered benefits; telephone numbers and contact information for the health plan, CHW, and other health-related resources). Two booster telephone interactions were performed at 2 and 4 months. At each contact the CHW emphasized her availability to address any questions that might arise.
Self-report patient-informed choice data were captured through a 6-month telephone survey, and we captured health care utilization over 6 months from the health plan's records.
The primary outcome in this study was whether new Medicaid health plan members identify and utilize (ie, have at least 1 visit with) a primary care provider (PCP; RQ1). We chose this outcome because use of a PCP is an important target outcome for the health plan, and access to (and use of) a PCP is a gateway to successful health care system utilization, including preventive services.10-21
The primary and RQ2 and RQ3 secondary outcomes are based on health care utilization data using place of service indicators, vendor categories, Current Procedural Terminology (CPT) codes, and ICD-9 or ICD-10 procedure and diagnosis codes. We determined new members to have identified and utilized a primary care provider (RQ1) if they had at least 1 claim during their first 6 months that met any of several primary care criteria. Similarly, we determined uses of recommended preventive services or immunizations (RQ2) as members having at least 1 claim that met any of the preventive visit criteria. We measured avoidable ED visits and hospitalizations22 as the number of ED visits and hospitalizations over a 6 month period and as the proportion of those visits and hospitalizations that were avoidable. We defined avoidable ED visits following an Oregon Health Authority report on the topic.23,24 The Agency for Healthcare Research and Quality's (AHRQ) Prevention Quality Indicators version 6 software was used for avoidable hospitalizations for Ambulatory Care Sensitive Conditions.25
For RQ4, we examined measures of members' health-related knowledge and whether they were engaged and participated in informed decision-making regarding their health care and health.26-28 Although a number of measures of related concepts such as health literacy8,29 and decision quality30,31 exist, each has limitations regarding patient-informed choice for preventive services and general health care. Therefore, we used related existing measures and added to these, as needed, with items developed through 3 focus groups (8-10 members each) with newly enrolled and previously enrolled engaged and nonengaged health plan members.32,33
The included existing measures were from the International Patient Decision Aids Standards (IPDAS) Collaboration's recommendations for measuring the quality of the health care decisions,30 the single-item Control Preference Scale34 (patients' preferred level of decision involvement), and the 5-item patient information and 4-item patient participation in decision-making subscales of the modified Patients' Perceived Involvement in Care Scale (M-PICS).35,36 The Single-item Literacy Screener (“How often do you need to have someone help you when you read instructions, pamphlets, or other written material from your doctor or pharmacy?”)37 and the short form of the Patient Activation Measure (PAM), a measure of patient engagement in health care, were also included.38-47 Following the IPDAS general categories through the focus groups. we added items on members' knowledge, confidence, and actions. Questionnaires were available in Spanish and English.
Data on number and types of members that accept the CHW offer (RQ5) came from the health plan's demographic data, and the resources used to provide CHW services (RQ6) came from study records.
We obtained health care utilization data (RQ1-RQ3 and RQ5) from UFC. We performed 2 “test” pulls to determine data quality, prebuild our systems and algorithms, and test data linkages. Due to project delays our final data included 6 (not 12) months of utilization for each member in our sample. Since the health plan has an incentive to avoid missing data, and to correct spurious claims, these data are rarely missing.
Data for the patient-informed choice outcomes (RQ4) came mainly from a preassessment and a 6-month postassessment of the invited-accepted (IG1) group. The baseline survey was administered to IG1 in person, after informed consent, when the member first met with the CHW; thus, no baseline data were missing. This same group was again surveyed by phone at 6 months, and nonresponse was minimized by capturing detailed multicontact information during the CHW visit, encouraging ongoing participation through the CHW-member relationship and planned booster sessions at 2 and 4 months, and offering a gift card at survey completion. The health plan's call center administered a similar 6-month survey by phone to a sample of the group randomized to invitation that did not accept the invitation (IG2; n = 62) and to a randomized group not invited (CG; n = 49). These calls were made using callers experienced with this population and with achieving completion. In all cases, the interview format ensured negligible item nonresponse.
The main comparisons in this study were between the randomized groups (IG and CG). However, functionally the IG was composed of individuals who accepted the invitation to work with a CHW (IG1) and those who did not (IG2). The comparison between IG1 and IG2 on baseline values described those who accepted CHW services when they were offered (RQ5). Finally, RQ6 required a simple accounting of time and costs.
UFC expected to enroll 2100 new eligible members (women aged 18-39 years) annually. Assuming a 2:1 randomization ratio and that 30% of this population lacks a usual source of care,48,49 detection of a 10 percentage-point reduction in those lacking a usual source of care—assuming an α of .05 and 80% power—would require a sample size of 675 (450 IG, 225 CG). For the pre-postassessment survey, the sample size for continuous variables (paired t test), assuming a small effect size (d = 0.2), an α of .05, and 80% power, was 199. The sample size required for dichotomous variables (McNemar test), assuming a 5-point difference in proportions, an α of .05, and power of 80%, was 77. The current study's random sample size of CG participants was 746 (n = 671 were unreachable for the 6-month survey; n = 49 completed the 6-month survey; n = 26 refused to complete the 6-month survey, which addressed RQ4). It should be noted that health care utilization data were available for the entire sample size of CG participants, which were analyzed and addressed RQ1, RQ2, and RQ3.
Comparisons across the randomized groups and subgroups used χ2 for dichotomous (eg, identification of a PCP) and categorical variables and Student's t test for variables that demonstrate a normal (or generally normal) distribution (eg, age). We compared baseline characteristics for those in IG1 who did not complete the 6-month survey with those who did, and when we found no significant differences, we made the assumption of missing completely at random yielding unbiased results in our pre-post analysis. Comparisons across randomized groups (IG and CG) for highly skewed variables (eg, number of avoidable ED visits) utilized nonparametric bias-corrected and accelerated bootstrap confidence intervals.50,51 Logistic regression was used to analyze predictors (eg, age, comorbidities) of our primary outcome.
All women aged 18 to 39 years newly enrolled in one of Arizona's Medicaid health plans (UFC) who were ambulatory and community dwelling were identified by the health plan on a monthly basis as eligible for this study. The study team provided 2:1 (invited: not invited) randomized lists based on Excel's random number generator to 1 individual at UFC with no member contact. We applied this randomized allocation consecutively to the lists of new members. Members randomly assigned to IG were contacted by the health plan's call center. However, call center personnel had no information on those not invited.
Of the potential participants who remained reachable after indicating an interest in the study, 62% (n = 159) accepted and received the services of the CHW. At the appointed time the CHW met with the subject, completed the informed consent process and the baseline patient informed choice survey (via touch tablet device), and conducted the individualized 30-minute education session. The entire CHW visit ranged from 50 minutes to an hour and 15 minutes depending on the number of questions a participant asked about the consent process or any follow-up questions regarding the educational session. If members did not show for their appointments, they were re-contacted; however, 2 no-shows ended contact attempts. At the conclusion of this appointment the member was encouraged to reach out to the CHW with any subsequent questions or concerns, appointments for the 2- and 4-month booster telephone calls were scheduled, and the member was given a $40 gift card. Following the delivery of the intervention, a range of participants (n = 114, 2-month booster; n = 117, 4-month booster) had additional contact with the CHW to get guidance on navigating the system and other problems. The 2-month booster telephone call rate was 78.6% (n = 114/145; 14 participants were no longer with the health plan when contacted for the 2-month booster call) and the 4-month booster telephone call rate was 80.7% (n = 117/145; 14 participants were no longer with the health plan when contacted).
At 6 months postrandomization, all eligible members (IG and CG) were called for a patient-informed choice survey identical to the IG1 group's baseline survey. Initial study plans were to use the health plan's regular member satisfaction survey to capture up to 10 informed patient choice items from the non-IG1 groups. However, the fact that the satisfaction survey was anonymous made linking these data with health care utilization data unfeasible. Fortunately, the health plan's call center agreed to field 6-month surveys (with up to 5 call attempts) to all members of the IG2 and CGs. The IG1 group's 6-month survey was conducted by UA research staff following the same 3-2-2 call attempt protocol used for CHW scheduling. A $40 gift card incentive was provided to the IG1 group and a $30 gift card to the IG2 and CGs for survey completion. Protocol approval came from both the UA's IRB and RAND's Human Subjects Protection Committee. No adverse events were reported.
As shown in Figure 2 and the tables in Appendix A, only 22% (331/1521) of the group randomized to invitation were reachable by the UFC call center to be offered the intervention, and only 17% ([198 + 58]/1521) remained reachable after they were transferred to the UA study team to be connected with a CHW. The issue of reachability in this population resulted in our having insufficient numbers in IG1 to detect our primary outcome when comparing the IG and CGs. Note, however, that when these new members were reached, they tended to accept the offer to work with a CHW. Of those who were contacted by the UFC call center and offered the services of a CHW, 82% (273/331) said that they were interested. This list of individuals was then passed on to the UA team to schedule a meeting with a CHW. Unfortunately, in that pass-through, even though the elapsed time was less than 5 days, 27% (75/273) of those initially interested became unreachable and another 14% (39/273) later refused the CHW offer. Overall, of the members who remained reachable until they could either decline (58 + 39) or accept the CHW offer (159), 62% (159/[159 + 58 + 39]) accepted and received the services of a CHW. This percentage is remarkably consistent with the percentages of those reachable for the 6-month survey who completed that survey: 65% (62/95) for the group randomized to the offer of a CHW who did not accept that offer and 65% (49/75) for the control group—ie, once someone was reached, almost two-thirds also accepted the offer.
Member Flow Through the Project.
Overall the IG1 group members reported at 6 months that they were very satisfied with their CHW services. Table 1 shows the baseline characteristics of the 2 randomized groups, IG and CG. As expected, according to the data available from the health plan, these groups are remarkably similar. However, we found substantial differences between those who accepted the offer of a CHW (IG1) and those who did not (IG2). This is discussed further under RQ5.
Baseline Characteristics Across Groups.
Table 2, Table 3, and Table 4 show the results for RQ1 to RQ3 using several measures for each outcome. The primary outcome is 1 or more visits with a primary care provider according to any of our measures and is shown in bold at the bottom of Table 2a. The main outcomes in the other 2 tables are also indicated in bold. The first 2 columns in each table show the results by the randomized groups. These comparisons indicate the impact of the offer of a CHW to this population using our member contact methods. As can be seen by comparing the first 2 columns in all 3 tables, we found no real difference between those randomized to be offered the intervention and those who were not offered.
Percentage of New Members Who Had at Least 1 Visit With Their Primary Care Provider During Their First 6 Months of Enrollment.
Statistical Analysis for Identification and Utilization of PCP.
Percentage of New Members Who Had at Least 1 Preventive Visit of These Different Types During Their First 6 Months of Enrollment.
Statistical Analysis For the Impact of the Intervention on the Utilization of Women's Preventive Services and Other Preventive Services.
New Members' Avoidable Use of ED Visits and Hospital Days During Their First 6 Months of Enrollment.
Statistical Analysis for the Impact of the Intervention on the Utilization of Avoidable ED Visits.
Statistical Analysis for the Impact of the Intervention on Preventable Hospital Stays.
The last 2 columns in each table break down the intervention group (the group that was randomized to be offered the services of a CHW) into those who accepted the offer (and worked with a CHW; IG1) and those who did not accept the offer (IG2). The group that actually worked with a CHW tended to have better outcomes.
Of the 159 members who completed the baseline survey and worked with a CHW (IG1), 101 (64%) also completed the 6-month follow-up survey. When we compared the baseline values of those who completed the 6-month survey (n = 101) with those who did not complete this survey (n = 58), we found few statistically significant differences (Table 6). In fact, only 3 of the almost 50 comparisons made, including across baseline survey responses, had P < .2. Similarly, Table 5 also shows that those who did not work with a CHW (IG2 group and CG) who were reachable and completed a 6-month survey were not really different from those who did not complete this survey. The main purpose of the 6-month survey data for those who did not work with a CHW was to provide context for the changes seen between baseline and 6 months in the IG1 group—ie, to help determine whether the changes seen in the IG1 group over the 6 months could be due to the experience of 6 months' enrollment rather than to working with a CHW. The IG2 + CG group that completed 6-month surveys were also generally similar to the IG1 group that completed this survey; they differed only by the number who declined to answer the race and general health questions.
Responses to the Knowledge/Information Informed Patient Choice Questions.
Demographic Characteristics of the Groups Based on Survey and Health Plan Demographics.
Table 6, Table 7, and Table 8 show the changes in the responses to the informed patient choice survey between baseline and 6 months for those who worked with a CHW (IG1) and the responses seen at 6 months for those who did not work with a CHW (IG2 + CG) for 3 sets of items: information/knowledge (Table 6; part of IPDAS decision quality52); patient engagement/empowerment (Table 7a; “I am confident in my health care decisions”26,27,53); and take action (Table 8;; part of IPDAS quality of the decision-making process52). Across the 3 tables we first compare baseline with 6-month responses for the IG1 group and then compare 6-month responses for the IG1 group and the IG2 + CG group to see if the changes seen pre-postassessment in the IG1 group could be mainly because of 6 months' enrollment in the health plan.
Responses to Taking Action Informed Patient Choice Questions.
Responses to the Patient Engagement Informed Patient Choice Questions.
Statistical Analysis for Informed Decision-Making Regarding the Health Care System and Their Health.
Regarding knowledge/information (Table 6), it appears that the IG1 group generally improved over time, in terms of knowing which doctors they could use as their primary care providers. They also improved in terms of knowing where to go if they were very sick or injured and could not wait but the illness or injury was not life threatening vs if they had an emergency or life-threatening illness or injury. They may also have worsened over time in knowing the preventive actions for their age range. However, the case that these changes are due to working with a CHW is best made (ie, the 6-month values for the IG1 group are significantly different from the 6-month values for the IG2 + CG group) for knowing where to go when very sick or injured and unable to wait but the illness or injury is not life threatening.
Regarding patient engagement/empowerment (Table 7a), across the 6 months we found no differences in the IG1 groups' PAM scores, the number of members who attained the higher levels of patient engagement, and their likelihood of endorsing as likely a number of statements of confidence in navigating the health care system, with the exception of 1—knowing where to go if sick or injured and need to see a doctor—which was statistically significant. However, the results seen at 6 months in the IG2 + CG group are similar.
Regarding taking action (Table 8), the IG1 group members improved over the 6 months on the following tasks: making sure that they received and read and/or asked questions about their health plan welcome package; knowing appropriate strategies for changing doctors; requesting clinic paperwork ahead of visit; arranging for transportation if needed; and making the final health care decision after seriously considering their doctors' opinions. We also saw the same results for all but the latter at 6 months in the IG2 + CG group. Over the 6 months, the IG1 group also had significantly fewer respondents reporting that they never or rarely needed someone's help to read instructions, pamphlets, or other written material from their doctor or pharmacy. We did not see similar results in the IG2 + CG group.
At the end of the 6-month survey, all respondents answered a series of open-ended questions regarding any problems with understanding the health plan's materials and procedures, deciding where to go to get help, changing providers, scheduling an appointment, or preparing for an appointment. Between 62% and 75% of respondents responded to each question that they had no problems. Those who were in the invited group but did not take the CHW offer were also asked why they declined. The most common answers were that they did not feel that they needed the help or that they were too busy and/or did not have childcare.
One of the benefits of the randomization-to-invitation design was that it allowed an accurate picture (ie, one not confounded with consent to be randomized) of the types of members who accepted the invitation when offered. Table 9 shows the results of this analysis.
Characteristics of Those Who Accepted the CHW Offer, Those Who Declined the Offer, and Those Who Were Unreachable for the Offer.
Because of the challenge in reaching this population, we first compared those who were reachable (IG1 + IG2a) with those who were not (IG2b). It seems that those who were reachable were older (P = .005); more likely to be Hispanic or Latina and less likely to be “other” (P = .038); less likely to be located in a small metro area and more likely to live in a large fringe metro, medium metro, or micropolitan area (P = .003); and more likely to live closer to where the CHWs were based (P = .003 and P = .009), which was at the UA in Tucson. Comparing among those who were reachable between those who accepted or declined the offer of CHW services, we those who accepted were statistically significantly older (P = .004), were more likely to be Hispanic or Latina and less likely to be White or “other” (P = .002); were more likely to live closer to where the CHWs were based (P = .004 and P = .005); and were more likely to give Spanish as their language preference (P = .017). Overall the IG1 group members reported at 6 months that they were very satisfied with their CHW services (Table 10).
Responses at 6 Months From IG1 Regarding Their Experience of Working With a CHW.
Table 11 shows the resource use and cost for the CHWs in this study. This estimate includes the time and travel costs for the CHWs themselves. It does not include supervisor time or time spent by the call center to contact the members who were eligible for CHW services and to schedule the time with the CHW. It took an average of 1.7 hours per member served for the CHWs to be trained and another day of CHW time to deliver those services to each member who received them (n = 159). Note that the number of hours per member for training could decrease if CHW time were fully utilized. The CHWs also drove 1074 miles in total (average of 6.8 miles per member) and for the longer trips incurred rental car, hotel, and per diem costs to meet with members. Using an average salary plus fringe for these CHWs of $15 per hour and mileage reimbursement of 44.5 cents per mile (the rates paid), and including $3845.40 for rental cars and $1197.78 for hotel and per diem, the total cost of receiving CHW services was about $181 per member who accepted ([1551 × $15 + 1074 × $0.445 + $3845.40 + $1197.78]/159). This cost should be compared by the health plan with the changes in health care utilization seen in this group to determine if offering CHWs is financially viable.
Time Expended by CHWs to Recruit and Work With New Members Who Accept Their Services (n = 159).
Given the large number of new Medicaid members who were unreachable in this study, we conducted exploratory analyses using 2 logistic regression models to examine the separate and combined effects of working with a CHW and of just being reachable—ie, being someone who has a working number and who answers their phone. We also used these models to explore other potential determinants of primary care use (our primary outcome). Our dependent variable for both models was our primary outcome: having a visit with a primary care provider. The 2 logistic regression models, 1 for each of our available measures of reachability, had 3 possible last sets of variables. All models included available variables for need, enabling factors, and predisposing factors (the base models), and then we included working with a CHW alone (the first option), the reachability variables alone (the second option), and then both (the third option) to see which of these variables added significant explanatory power to the base models.
We entered blocks of explanatory variables available from the health plan into each model sequentially according to a defined causal theory.54 The work by Andersen et al provided the basis for the models.55-58 Each block contained a set of related available variables believed to define the causal mechanism in question: need, enabling factors, and predisposing factors. The blocks were ordered from those hypothesized to be most proximal to those most distal in explaining an individual's use of their PCP. The exceptions to this scheme were that working with a CHW and reachability were included separately and then in combination as the last variables in the models to determine whether these variables were explanatory even after all other variables had been taken into account. Sequential entry of these blocks allowed the incremental explanatory power of each block to be measured and tested in a logical order.
This analysis was limited to the variables available from the health plan. Available variables related to need were age and the Charlson/Deyo index.49,59-64 Enabling factors included living in a more urban environment (access to health care),49,63,65 distance from the UA (located in a major metropolitan area, site of one of the major health care centers, and base for the CHWs), and English language preference.63,65-69 Predisposing factors were race/ethnicity.63,66
Reachability was defined by codes (see tables in Appendix A) given for each member after their call attempts were completed and subdivided into 4 types: (1) hard unreachable (“dead air”/no sound, “fast” busy, recording saying number is not working or disconnected, or no phone number available); (2) soft nonhuman unreachable (constant ring, constant busy signal, or left message on answering machine with no response); (3) soft human unreachable (respondent requests callback or that interview be conducted in person but connection never re-established, language other than English or Spanish); and (4) no longer at UFC (self-report of no longer being a health plan member).
The first model includes only the IG group and uses the measures of reachability captured when attempts were made to contact these members to offer CHW services. The second model includes all eligible members and uses the measures of reachability from attempts made to conduct the 6-month survey. Note that for both sets of call attempts, more than half (54%) of those who could not be contacted fell into the category of hard unreachable and about 30% were soft nonhuman unreachable.
Since the adjusted odds ratios (aORs) for the first 3 blocks of variables (the base model—ie, all variables other than working with a CHW and reachability) remained fairly constant across all versions of the model, in Table 12 aORs for the first 3 blocks from the models ending with the CHW variable are reported. Health status based on the Charlson/Deyo index was highly predictive in both models. The enabling factors of living in a county designated as small metro (metropolitan areas with populations of 50 000-250 000) and living close to the UA also helped predict use of a PCP. However, no other available need, enabling, or predisposing factor variables added much to the models. When we added “working with a CHW” to these 3 groups of variables, it explained a significant amount of variance, and its aORs were statistically significant. When we added the reachability variables instead of the CHW variable, they also explained a significant amount of variance, and the aORs for all but the soft human unreachables in the full cohort model were statistically significant. Finally, when we entered both working with a CHW and reachability, as expected the combination explained a significant amount of variance, but the CHW and several of the reachability variables' aORs were no longer statistically significant. This indicates that the PCP benefits associated with working with a CHW may actually be at least partially explained by the characteristics that make a new member reachable or unreachable.
Estimated aORs for Models Predicting a Primary Care Visit Within the First 6 Months of Enrollment for New Medicaid Members Based on 2 Samples of Members, 2 Measures of Unreachability, and 3 Alternatives for the Last Block of Variables.
This study examined whether the offer by a Medicaid health plan of CHW services to new members would improve their use of the health care system and increase their rate of informed decision-making. Overall this offer was not effective, mainly because the health plan was unable to connect with these new members to make the offer. Following usual practice, the health plan offered these services to new members via its telephone call center. Of the new members randomized to be offered CHW services, 78% were unreachable. In the short (less than 1 week) step from indicating interest in working with a CHW to having the appointment with the CHW, another 5% became unreachable. Therefore, a health plan interested in offering CHW services to its members should consider another method of member contact. One suggestion discussed below is to cover the services of CHWs embedded at local health care centers.
Note that this study did not test the effectiveness of the CHW services themselves, other than to measure their impact on improving new members' informed patient choices. However, those who received the CHW services were highly satisfied.
CHWs are a core component of recommendations to reduce health disparities,70 and systematic reviews71-73 and a number of newer studies74-85 attest to their effectiveness in improving health care utilization for many conditions. However, little attention has been given to how these interventions are delivered or who should deliver them. According to our review of the literature, ours is the first study to examine the offer of CHW services by a health plan. The offer was made by the health plan's telephone call center, its usual mode of member contact, to new Medicaid members—in this case to new female members aged 18 to 39 years—with the purpose of improving their health care utilization and health. The offer by a health plan also has the advantage of providing complete aggregate data of this vulnerable patient population, which allows the full effect of this real-world offer to be tracked. However, although numerous attempts were made, a large proportion of the eligible new members were unreachable, and this offer was ineffective.
Note that, although not explicitly stated, the participant recruitment strategies used in the studies of CHW effectiveness imply a delivery organization—ie, an organization that would be capable of offering CHW services through the recruitment channels used in the study. In some cases this implied organization is a health care provider of some type—eg, for studies in which participants were identified from medical records80,86-90 or captured in person during a medical visit.91-94 In other cases the organization could be some sort of social services organization—eg, for studies in which participants were identified through church attendance,78,79,82,95-97 the social network of the CHWs involved,84,98,99 households,85,100,101 or in the community at large.82,102-104 Note that none of these channels are directly available to a health care payer or health plan. However, the success of provider organization-based methods and the ready connection between those organizations and health plans could signal that reimbursement of CHW services based in the provider organization might have been a more effective delivery mechanism.80
Through the call result codes documented for each eligible member (see Appendix A), we have information on what happened during the health plan's call attempts but can only hypothesize as to why this population was so unreachable. Since all members qualified for Medicaid, we know that they were low income, and thus, lacked economic security,105 but they could also have been facing housing instability106,107 or other life challenges.108 Caller ID and the use of inexpensive cell phones with minute limits and disposable prepaid phones could also be contributors.
Nevertheless, although the CHW offer was attractive (82% of those contacted expressed interest in working with a CHW, and 62% successfully met with a CHW) and CHW services were themselves likely effective, because of low reachability the proportion of the group randomized to the offer of CHW services that actually received these services was so small that the group as a whole did not show statistically significant benefits over the control group. Although this study did not support the effectiveness of a health plan's offer of CHW services, and although it was not designed to directly test the effectiveness of these services themselves, our results show some indications of CHW effectiveness. A major goal for the health plan was to increase new members' use of a PCP, and a statistically larger proportion of the group that received the CHW services (IG1) had 1 or more primary care visits (our primary outcome; 71.7%) and well woman visits (15.4%) than did those randomized to the offer who did not work with a CHW (IG2; 56.6% and 9.8%, respectively) or the control group (CG; 57.1% and 10.1%, respectively). Although improved access to care is one of the main outcomes hypothesized for CHWs,70 we could find only 1 other study in which primary care access was measured, and it was a small study of a nurse-CHW team in which patient access to primary care improved.109 Many more studies have shown improvements in the use of specific preventive services, including cervical cancer screening.75,77-79,83-90,95-101,104 However, in our study, although the proportions of those who received cervical cancer screenings and contraceptive counseling were higher in the IG1 group, those differences were not statistically significant. We also found no statistically significant differences in the proportions of avoidable ED visits and hospitalizations. However, 6 months might have been too short a period to see differences in this outcome for this younger, generally healthy population.
Many studies have shown that working with a CHW can increase individuals' health-related knowledge.74,78,82,90,110-114 In our study, members working with a CHW had statistically significant pre-post improvements in several measures of knowledge, especially about finding and changing PCPs, where to go for care, what to do to prepare for a visit, and navigating the health plan. Compared with the 6-month responses of the IG2 and CG groups, a number of these improvements may be attributable to working with a CHW rather than simply to 6 months' experience with enrollment. Over the 6 months, the IG1 group members increased their likelihood of reporting that they needed someone's help to read instructions, pamphlets, or other written material from their doctor or pharmacy.
When looking at who accepted the offer of CHW services, we first examined the characteristics of those who were reachable for the offer because a member could not accept without first being contacted, and then of those who accepted. It seems that those who were reachable were older than those unreachable, and of the reachables, those who accepted were also older. However, in each case the average difference was so small—a year or less—that it might not be of practical significance. Also, when comparing reachables with unreachables and then accepts with declines, the reachables and accepters were more likely to be Hispanic or Latina and less likely to be White or “other,” and they were more likely to live closer to where the CHWs were based—that is, near Tucson and the UA. It is unknown why these characteristics would be associated with reachability other than possibly the more urban/interconnected environment of Tucson and the UA. Regarding the higher acceptance of the offer, CHW services are often designed for and target minority groups, including Hispanics, and 1 characteristic of CHWs is that they are known within and draw on social networks in their communities—ie, their impact is higher close to home.70,115 Finally, those who accepted were more likely to give Spanish as their language preference; a non-English preference is a known barrier to health care.63,65-69
In order to explore potential determinants of primary care use (our primary outcome) we estimated 2 logistic regression models, each with 3 possible endings. In both models we included available variables for need, enabling, and predisposing factors (the base models) and then, in the first version, included working with a CHW to see if it added significant explanatory power. In the base models the blocks of need factors and enabling factors were statistically significant. It is unsurprising that those with comorbidities (more severe/complex health conditions) were much more likely to have a PCP visit in their first 6 months, and that age for these generally younger women would not make much difference. It is also interesting that living in a small metro county (containing cities or towns with populations of 50 000-250 000) increases the odds of seeing a PCP and that living farther away from Tucson decreases the odds. Adding working with a CHW to these base models explained a statistically significant amount of additional variance, with these women having 1.8 times the odds of seeing a PCP. However, adding measures of reachability to the base models instead of working with a CHW also explained a statistically significant amount of additional variance, with the unreachable women generally having lower odds of seeing a PCP. This is consistent with what has been seen in other studies of vulnerable or unstable populations.106,116,117 It is also not surprising that those who reported soon after study start that they were no longer with the health plan would have much lower odds of seeing a PCP than those who made this report only at 6 months. The third version of the models adds both working with a CHW and the reachability variables at the same time. As would be expected, this block explained a statistically significant amount of additional variance. However, only the odds ratios for a report of leaving the health plan remained statistically significant in both models. This indicates that part of the beneficial effect of working with a CHW might come from the same characteristics that make those individuals reachable in the first place.
As discussed above, this study showed that the health plan offer of CHW services to new Medicaid members was ineffective, and as such the implementation of a similar type of offer (ie, trying to contact these new members using the health plan's call center) is not recommended.
Our main study result, that the offer of CHW services by a health plan via its call center is not effective, is likely generalizable to other Medicaid health plan populations of women aged 18 to 39 years who are new members with roughly the same characteristics: 32% Hispanic and 39% White, using the mutually exclusive categories for each available in the health plan's data, and highly unreachable by phone.
The 2 main limitations of this study were that the health plan was unable to contact most of the population to make the offer of CHW services and that the young, generally healthy population made it difficult to detect impacts on avoidable ED visits and hospitalizations within a 6-month period. However, neither was a source of systematic error or bias, and the first is the main outcome of the study—that the direct offer of these services by a health plan likely will not work because of factors that affect the receipt of the intervention offer (ie, the determinants of reachability, measured and unmeasured). Although a health plan responds to its members' needs and queries, it is not usually the initiator of member contact, and this study clearly illustrates this limitation. On the other hand, the offer by the health plan provided 2 advantages to the study of CHW services. First, Medicaid health plan membership constitutes a clearly defined, complete, and generally vulnerable population in which to study the effects of interventions to improve health care access and utilization. Second, health plan data are available on all members. In contrast to studies that examined the effectiveness of an intervention only in the population sample that could be reached, this full population and full set of data also allowed examination of who was being missed. Finally, it should be noted that those administering 6-month surveys were not masked to treatment allocation—study personnel surveyed those who worked with a CHW (IG1), and health plan call center personnel surveyed all others.
Medicaid health plans have an incentive to support interventions that help their members use health care in a manner that allows for the more efficient use of these resources and improves members' overall health. Numerous studies have shown that CHWs can be an important and effective element of these interventions. Further research is needed to find a better way to combine the health plan's incentive to offer these services via a delivery mechanism that will reach the members who will most benefit from them. As discussed above, 1 possible mechanism would be to have the CHWs embedded in the local clinic, ED, and/or hospital to successfully guide new members through their first, and subsequent, encounters with the health care system. These services could then be covered by the health plan; members who are approached for the intervention and those missed could be tracked; and members' health care utilization could be captured to determine effectiveness.
The direct telephone-based offer of CHW services by a Medicaid health plan to its new members to increase their use of a PCP was unsuccessful. However, the lack of success cannot be attributed to the CHW services, which were themselves well received and might have been effective. Instead, it is likely due to the inability of the health plan to connect with, and make the offer to, enough of its members to generate a noticeable effect.
The initial protocol called for 3 call attempts to invite members for participation, and although the protocol indicated that calls be made at different times and on different days, successful connection with members was often unachievable, even when members were interested in participating. We had anticipated that increasing the number of call attempts would improve the probability of a successful connection with potential participants. In addition to increasing the number of call attempts, mailings that included compensation amounts were sent out as postcards as an additional form of contact to individuals who had been randomized to the intervention. Based on the current study results, recommendations for future studies include incorporating a multimodel approach to communicating with highly mobile Medicaid populations. In this age of social media, it could be beneficial for researchers to incorporate texting, emailing, and instant messaging on various platforms into their recruitment and engagement protocol.
In retrospect, the direct telephone-based offer of CHW services should not be surprising given that a health plan's job is to cover the health care utilization of its members, which is for the most part handled electronically between the health care provider and the plan. A health plan is expected to respond to members' inquiries; its business plan does not require it to initiate contact with its members. On the other hand, targeting a Medicaid health plan's members for this intervention makes sense on a number of other levels—both the health plan and its members benefit from the appropriate use of health care; the Medicaid population is often in need of services that can improve access and reduce disparities; and the health plan has data on all its members, which allows examination not only of the effects of the intervention but also of the characteristics of members who are being missed. The benefits of health plan involvement combined with the fact that direct contact for this population does not work begs the question of whether there is a more effective way for a Medicaid health plan to offer CHW services to its members.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#IHS-1306-04356) Further information available at: https://www.pcori.org/research-results/2013/does-offer-phone-community-health-worker-support-increase-access-primary-care
Detail on the Results of the Call Attempts (PDF, 137K)
Garcia FA, Wilkinson-Lee AM, Herman PM, et al. (2019). Does an Offer by Phone of Community Health Worker Support Increase Access to Primary Care for Women Who Are Newly Enrolled in a Health Plan? Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/10.2019.IHS.130604356
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
