Interviews were completed with the families of 413 children. After subtracting the unreachable families from the original sampling frame, the overall response rate was 49%. Children without a regular source of care were excluded from the analyses, leaving 403 respondents in the analytic sample.
Nonrespondents were similar to respondents in terms of child’s sex, race and ethnicity, and school in which the child was enrolled. Respondents were slightly more likely than nonrespondents (P < .05) to have a younger child (mean age = 8.1 vs 8.4 years). Data for these comparisons were available through an administrative data set provided by the school district and assembled each school year through an enrollment form completed by parents.
Measurement
Race and ethnicity. Race data were available through the parent-completed school enrollment files provided by the school district. The categories of race or ethnicity were white (non-Hispanic), Hispanic, black (non-Hispanic), Asian, Filipino, Pacific Islander, and American Indian. To ensure a sufficiently large sample size, we combined Asian, Filipino, and Pacific Islander into a single category called Asian. We also excluded American Indian from the study sample because of extremely small numbers.
Primary care quality. For this study we used the Pediatric Primary Care Assessment Tool (PCAT) developed by the Johns Hopkins Primary Care Policy Center for the Underserved to evaluate 4 cardinal attributes of primary care quality: first-contact care, longitudinality, comprehensiveness, and coordination Table 1. Scale scores were generated for each attribute based on summed responses to questions, with 4 Likert-type response choices: definitely (score = 4), probably (score = 3), probably not (score = 2), and definitely not (score = 1). “Don’t know” responses were coded as the middle score (2.5) because we assumed that not knowing about an important feature of primary care signified some partial failure to convey the availability of that particular feature. For example, parents’ not knowing whether their child could receive immunizations from the provider signified some partial communication failure on the part of the provider. Both child and adult versions of the instrument have been developed, the reliability and validity of which are reported elsewhere.29,30
Within each cardinal attribute, the PCAT assesses structural characteristics of the facility or provider that reflect the capacity to achieve quality primary care and processes of care that indicate the achievement of the function in actual practice. Only patients who reported a regular source of care (n = 403) were asked to assess the quality of primary care.
First-contact care. Two subdomains of first-contact care are measured by the PCAT: accessibility of the provider and the degree to which the provider is used as a single point of entry into the medical care system. Accessibility is evaluated with 8 questions about characteristics of the health system that facilitate access (eg, If the facility is open on weekends, would the provider see the child the same day?). The utilization subdomain is scored with an algorithm that assigns a higher score for each type of service (acute illness, regular check-up, and immunizations) that is sought from the parent-identified regular source of care.
Longitudinality. Two subdomains of longitudinality are measured by the PCAT: interpersonal relationship with the provider and extent of affiliation. The relationship subdomain is evaluated with 14 questions concerning the parents’ perception of the “person orientation” of the interactions between provider, parents and child (eg, the degree of interest the provider has in the child as a person rather than as someone with a medical problem). The extent of affiliation subdomain addresses the extent of the child’s relationship with a specific provider. This is scored with an algorithm that assigns a higher score if the provider identified as the regular source of care also knows the child best and is the provider from whom care would be sought for a new problem.
Comprehensiveness. Two subdomains of comprehensiveness are measured by the PCAT: services available and services provided. Six questions address the availability of specific primary care services (eg, immunizations and tests for lead poisoning). Another 5 questions address the services received from the primary care source (eg, discussions of ways to stay healthy such as eating nutritious foods and getting enough sleep).
Coordination. For children who have visited a specialist (n = 135), 7 questions address the degree of interaction and integration between the primary care physician and specialist services (eg, Did the primary care provider know that you made the visit to the specialist?).
Covariates. We selected covariates based on previous studies demonstrating a relation between the variables and aspects of primary care quality such as accessibility and continuity of care. We controlled for socioeconomic status (income, employment, and education), characteristics of the health care system (provider specialty, practice setting, and cost-sharing requirements), and demographics (child’s age, sex, general health status, and insurance coverage). Because of extensive managed care penetration in California, we clarified the range of practice settings by using names of local managed care clinics, public health centers, and group practices as examples.