Percentage of General Relief (GR) beneficiaries who visited their designated General Relief Health Care Program provider within 4 months of enrollment.
Diamant AL, Brook RH, Fink A, Gelberg L. Assessing Use of Primary Health Care Services by Very Low-Income Adults in a Managed Care Program. Arch Intern Med. 2001;161(9):1222–1227. doi:10.1001/archinte.161.9.1222
To assess the effect of providing free health care services to low-income adults.
We measured access to primary care services by enrollees with 4 chronic medical conditions in the General Relief Health Care Program (GRHCP), a program designed for adults receiving General Relief (GR). Implemented by the Los Angeles County Health Department in October 1995, the GRHCP is composed of private and public health care facilities. As adults registered for GR, they were asked to complete a baseline health survey, were enrolled in the GRHCP, and assigned a health care provider. A total of 8520 surveys were completed between September and November 1996 (98% response rate). The analyses of this article are limited to individuals (N = 2164) who reported a history of hypertension, diabetes mellitus, a nonresolving cough, or substance dependence. We reviewed medical records to determine whether new GR recipients had visited their designated GRHCP provider within 4 months of enrollment and used multivariate logistic regression to assess the effect of individual patient factors on the use of free health care.
A total of 17% of individuals visited their assigned GRHCP provider within 4 months of enrollment. In multivariate analysis, patients were more likely to have made a visit if they were younger than 50 years, were female, were Asian/Pacific Islander, reported needing to see a physician, or had seen a physician within 12 months.
It is not sufficient to merely supply the name and address of a health care provider to this population. More aggressive efforts should be attempted to increase utilization of services for patients with medical conditions responsive to ambulatory care.
AT THE SAME TIME that many privately insured individuals have made the transition to managed care, the number of uninsured has continued to rise. In response, some state and local governments have implemented programs that provide health care to low-income children and adults.1- 9 Increasingly, these programs have been based on managed care models. The goals of these programs are to (1) expand access to needed health care, (2) optimize the quality of health care provided, and (3) limit the costs of medical care.3,4,10 This article reports an assessment of a program specifically designed to provide free and accessible primary care services to poor, uninsured adults who were not receiving Medicaid but were eligible for public assistance (ie, General Relief [GR]) in a large urban county).
Beginning in October 1995, adults in Los Angeles County, California, who were receiving GR benefits were enrolled in a county-organized health care network, the General Relief Health Care Program (GRHCP). Before October 1995, most of these individuals relied on Los Angeles County Department of Health Services facilities, including emergency departments, for their medical care and were uninsured for health care.11,12 The GRHCP was designed to establish a usual source or site of care for enrollees and thereby increase access to primary care services. The GRHCP assigned a health care provider to newly registered GR beneficiaries, and individuals were instructed to go to that provider for both their urgent and routine primary health care needs.
The first primary care physicians who joined the GRHCP were within private community organizations or were individual practitioners; later, 4 traditional Department of Health Services facilities were added to the program. However, Los Angeles County Department of Health Services retained responsibility for providing specialty, emergency, and hospital care to patients referred to them by the GRHCP provider. At the time of this study, the GRHCP was composed of a network of 11 community health care organizations and 4 Department of Health Services facilities. One of the original community health care organizations in the GRHCP was composed of 4 clinics and an independent practice association that included more than 25 individual physicians.
Our study had 2 main objectives: (1) to measure use of free primary care services in the form of a visit by GRHCP enrollees to their designated care provider and (2) to assess which characteristics of enrollees were associated with their use of health care services. We focused on a sample of very low-income adults on GR with at least 1 of 4 marker conditions: hypertension, diabetes mellitus, a nonresolving cough, and substance dependence. We selected these conditions because of (1) their prevalence in the population, (2) the potential for serious complications if untreated, (3) the known benefit of regular health care in treating them, and (4) the existence of accepted standards of care for treatment.13- 19
To apply for GR (ie, public assistance, welfare benefits) in Los Angeles County, adults submit their application to an eligibility worker at 1 of the 14 Department of Public Social Services offices. Beginning in October 1995, applicants judged to be eligible for GR were referred to the health benefits representative at that site. The GR beneficiaries may have been found later to be eligible for Medicaid but at the time of application were only eligible for the GRHCP. The health benefits representatives were responsible for explaining the GRHCP to the new GR beneficiaries, including their eligibility for free health care through this system. The beneficiaries were asked to complete a baseline health history questionnaire, developed by the Los Angeles County Department of Health Services. Adults chose or were assigned to a specific GRHCP provider organization based on the location of the Department of Public and Social Services office at which they had registered for GR benefits. The health benefits representative recorded this information on the health questionnaire.
A total of 8520 baseline health history surveys were completed by adults who met with a health benefits representative between September 1, 1996, and November 31, 1996 (response rate, 98%). Adults were eligible for inclusion in our study if they reported a history of 1 or more of 4 medical conditions—hypertension, diabetes mellitus, a nonresolving cough, or substance dependency—on the baseline health history questionnaire20,21 and a GRHCP provider was recorded on their baseline health history (n = 2771). Because of logistic constraints (25 individual practices), we did not include in our study GR beneficiaries assigned to the independent practice association (n = 533). In addition, we excluded patients if information identifying their GRHCP provider (n = 66) or their date of registration (n = 8) was incomplete or missing. Our study sample comprised 2164 low-income adults.
Between July 1997 and April 1998, we attempted to locate the medical records for each person in our sample. From the baseline health history questionnaire, we obtained the name of each individual's designated GRHCP primary care provider. We provided clinic staff with a list of individual patient identifiers (ie, name and Social Security number). If a provider organization had more than 1 site, we sent the pertinent patient lists to all of the sites to determine if medical records existed for those patients. We reviewed each medical record obtained to determine whether the patient had made a visit to his/her designated GRHCP provider within 4 months of enrollment. Only a few of the providers had a computerized appointment system, which allowed us to review the appointment records for the 4-month period beginning on the date the baseline health history questionnaire was completed. We chose a 4-month window as the interval from the time of enrollment in GR to their appointment with a GRHCP provider after extensive discussions with both primary care providers and specialists who care for patients with the 4 medical conditions included in this study. In addition, this time window is consistent with published guidelines for the care of patients with hypertension and diabetes.
At each site we worked with clinic staff to locate medical records for all study patients designated to receive care at that site. We checked file rooms, physicians' offices, and other patient care locations. If we determined that a patient had been seen at that site, but we were unable to review the medical record at our initial visit, we asked the clinic staff to find the chart, and we scheduled a return visit to that site. We assessed the test-retest reliability of our method for measuring patients' use of free primary care services by submitting the same patient lists to a limited number of clinic sites at a later date. No new medical records were identified, and all previously reviewed medical records were eventually reobtained.
Our outcome variable was a visit to a designated GRHCP provider within 4 months of enrollment in the GRHCP.
We used information obtained from individuals' baseline health history questionnaire: age, sex, race/ethnicity, preferred language, education, and current residential status (housed in one's own home or apartment or with family or friends vs homeless, such as living in a hotel or motel, in a shelter, in a residential treatment facility, on the streets, near a freeway, or in an abandoned building); type of transportation to a medical appointment (public, private, or none other than walking); any history as a survivor of violence; psychiatric history (≥1lifetime overnight hospitalizations at a psychiatric facility); current use of tobacco; health status; a patient's reported need to see a physician; and prior use of health services (ie, last visit to a provider and use of prescription medications).
To measure use of primary care services through the GRHCP, we calculated the proportion of our sample who had made 1 or more visits to their designated provider within 4 months of enrollment in the GRHCP. We then used the χ2 and Fisher exact tests to assess the associations between our outcome variable and the explanatory variables. We used multivariate logistic regression (relative risks and 95% confidence intervals) to measure the effect of the explanatory variables on the outcome of having made a visit to a designated GRHCP provider within 4 months of enrollment. We included independent variables in the multivariate logistic regression if they were part of conceptual models that have been used to predict the use of care22- 25 or if they were significantly associated with the outcome variable in the bivariate analyses (P<.05). The independent variables in the model were as follows: age, sex, race/ethnicity, homeless status, current tobacco use, health status, perceived need to see a physician, usual site for health care (clinic, emergency department, private physician), current use of prescription medications, and visit to a physician within 1 year before entering the program. SAS 6.12 statistical software (SAS Institute Inc, Cary, NC) was used for all analyses.
We submitted the study protocol to the Human Subjects Protection Committee of the University of California at Los Angeles and received institutional review board approval. In addition, the study protocol was presented to representatives of all the GRHCP provider organizations and was submitted as required to their individual institutional review boards for approval. We obtained permission to collect data from the Los Angeles County Department of Health Services sites by applying to the institutional review board at the Los Angeles County–University of Southern California Medical Center.
Table 1 describes our study sample. The median age of the sample was between 35 and 40 years, and 36% were female, 48% African American, 29% Latino/Hispanic, 15% white, 1% Asian/Pacific Islander, and 7% multiracial and other. Less than half of the sample were high school graduates, a similar proportion were homeless, and 12% preferred a language other than English. Seventy percent rated their health status as fair or poor, and an equal proportion reported they thought they needed to see a physician. Eighteen percent had ever been hospitalized overnight at a psychiatric facility, and 77% had seen a physician within the preceding year.
A total of 358 (17%) of the study sample visited their GRHCP provider within 4 months of enrolling in the program. However, as Figure 1 shows, visit rates varied by medical condition.
About one quarter of patients with diabetes or persistent cough visited their provider within the 4-month window. However, only 18% of patients with hypertension and 12% of patients with substance dependence did so. Bivariate analysis revealed that individuals who had made at least 1 visit to their designated GRHCP provider were more likely to be older than 50 years, female, and Asian/Pacific Islander (Table 2). In addition, individuals were more likely to have made a visit to their GRHCP provider if they were not homeless, reported fair or poor health status, stated a perceived need to see a physician, had a private physician as their prior usual source of care, reported using prescription medications, or had visited a physician within the preceding 12 months. Individuals were less likely to have made a visit if they reported regular tobacco use. Education, mode of transportation, a history of having been a survivor of violence, or a history of psychiatric hospitalization were not significantly associated with having made a visit within 4 months.
Results of the multivariate logistic regression analyses are found in Table 3. Women and adults older than 50 years were about a third more likely to have visited their designated GRHCP provider within the 4-month window of interest. Those who had seen a physician in the preceding 12 months were about half again more likely. The most powerful effect was perceived need to see a physician; patients who expressed such a need were more than 75% more likely than others to have made a physician visit. Factors that were not independently associated with having made a visit included the following: current residential status, smoking, self-rated health status, having a prior usual source for health care, and current use of prescription medications. Because prior use may account for most of future use, we reran the multivariate model, excluding prior use, and found no differences.
The GRHCP represents an important attempt by local government to provide health care for adults at high risk for poor health because of poverty. To increase access to health care for very low-income adults receiving GR benefits and to optimize the geographic distribution of services available through this new program, the county contracted with community health care providers throughout Los Angeles County. Many GRHCP providers had extensive experience caring for very low-income adults and were located in areas that had previously been identified as underserved.
However, as this study demonstrates, the provision of free primary care services to very low-income adults does not guarantee that the services will be used. We found that only 17% of very low-income adult patients with a medical condition shown to benefit from regular ongoing medical care had visited their designated health care provider within 4 months of enrollment, despite the fact that these patients were enrolled in a publicly funded managed care program that was geographically dispersed and free of charge. We also found that the proportion of individuals who made a visit to their assigned provider varied considerably by medical condition. Those with diabetes were the most likely to have made a visit within the 4-month time frame; those with a history of substance dependence were the least likely; however, in the multivariate model, the individual medical conditions did not have significant main effects.
We attempted to identify characteristics that could be used to explain use or lack of use of primary care services by very low-income adults who were enrolled in a program that provided them medical care free of charge. Although prior research has shown that sociodemographic factors such as sex, age, and race/ethnicity are associated with differential access to health care—women and the elderly are more likely than men and younger adults to seek care, but they may report greater difficulty receiving care—other barriers to care also exist.1,25- 31 These barriers may take the form of competing or conflicting needs, including a lack of housing, food, transportation, and child care, as well as fears for personal safety.32- 34 In these analyses, we studied the effects of specific competing needs (housing status and mode of transportation) but were unable to include others such as food and child care due to limited information. Although we found significant bivariate associations for health behaviors and competing needs with use of health care, these factors were not significant when we controlled for demographic and other explanatory variables, including patients' prior usual site for health care.
Surprisingly, although poor health status has been shown to predict adults' use of health care and although it has been documented that low-income populations tend to have worse health status,35- 37 self-rated health status was not a significant predictor of use after adjusting for other variables. However, adults who indicated a need to see a physician were almost twice as likely to have made a visit to their GRHCP provider within 4 months than those who did not report a need to be seen. Of note, slightly more than two thirds of adults who did not make a visit to their assigned GRHCP provider reported a need to see a physician; the highest rate of perceived need was among patients with diabetes and the lowest among patients with substance dependence. We did not measure the beneficiaries' understanding of the services available, the presence of barriers to seeking services through the GRHCP, or the ability to obtain health care elsewhere, and all of these unmeasured attributes could have contributed to the low use rate.
Prior research has demonstrated that having a regular source for health care improves access to and use of health care28,38,39 and reduces the number of hospitalizations for ambulatory sensitive conditions (eg, diabetes, hypertension, asthma, congestive heart failure).28,40,41 In addition, prior use of health care has been shown to predict current and future use.39 We found that an individual's prior use of health care, including where he/she previously received care, current use of prescription medications, and the duration since the last visit to a physician, was associated on bivariate analysis with current use of care. However, in our multivariate analysis, only a visit to a health care provider within the preceding year remained significant.
Our study has 4 limitations. First, we relied on patient self-report to identify individuals with 1 of the 4 medical conditions of interest for this study; thus, we did not include people who had the condition but did not know it. If we had included this group, visit rates would probably have been even lower. There may also be the possibility of overreporting, although patients' perceptions of need related to having 1 or more of the 4 conditions would have been expected to increase use in the form of a visit to a designated GRHCP provider. Prior research has demonstrated that homeless adults accurately report the presence or absence of ambulatory medical conditions.20,21
Second, we were not able to obtain information about whether individuals sought primary care services from someone other than their designated GRHCP provider. According to Los Angeles County Department of Health Services regulations, providers other than a patient's designated GRHCP provider were not reimbursed for primary care services provided to GR recipients. In addition, because data were obtained by patient survey only at the time of their enrollment into the GRHCP, we were unable to determine if they made visits outside the GRHCP subsequent to their enrollment into it.
Third, we relied on obtaining and reviewing the medical record to determine if a visit had been made within the designated 4-month window. Our inability to obtain medical records could have caused us to underreport the proportion of people who had made a visit to their provider. However, we assessed our ability to obtain specific medical records and found very high test-retest reliability.
Fourth, the study sample was not population based but relied on identifying GR registrants who met with a health benefits representative and completed the baseline health survey. Thus, the study sample may not be representative of all very low-income adults or even all GR recipients.
We believe that providing a free health care benefit at the time a GR applicant is enrolled in the welfare program represents an innovative approach to increasing appropriate use of primary health care services by very low-income adults. However, although necessary, this attempt is not sufficient to increase use of care to medically appropriate rates. Our study results suggest that the current program needs to be enhanced. Although evidence does not currently exist for the effectiveness of any of the following enhancements, they are listed herein for consideration. Of course, their impact would need to be carefully evaluated. At the time of enrollment in the welfare program, the beneficiary's medical history could be checked and a specific medical appointment made (1) for all beneficiaries or, if that is not possible, (2) for those beneficiaries who either expressed a need for a physician visit or had a chronic condition whose care required such an appointment. In addition, we suggest that education about the need for routine and continuous medical care will also be provided to patients who have treatable medical conditions but who also express no need to see a physician.
Accepted for publication September 14, 2000.
Dr Diamant's work on this article was funded as a Robert Wood Johnson Clinical Scholar and a National Research Service Award Primary Care Research fellow. Dr Gelberg is a Robert Wood Johnson Generalist Physician Faculty Scholar.
We thank staff at the Los Angeles County Department of Health Services for their assistance with this project and David Klein for programming.
This work does not necessarily represent the opinions of the funding organizations or of the institutions with which the authors are affiliated.
Corresponding author and reprints: Allison L. Diamant, MD, MSHS, UCLA, Department of Medicine, Division of General Internal Medicine and Health Services Research, 911 Broxton Ave, Los Angeles, CA 90095 (e-mail: firstname.lastname@example.org).