 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ELECTRONIC ARTICLE |
* Center for Child Health Care Studies, Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massachusetts
Division of General Pediatrics, Children’s Hospital Boston, Boston, Massachusetts
Center for Health Studies, Group Health Cooperative, and Child Health Institute, Department of Pediatrics, University of Washington, Seattle, Washington
|| Division of Research, Kaiser Permanente, Oakland, California
¶ Department of Pediatrics, Kaiser Permanente, Vallejo, California
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Methods. A prospective cohort study with 1-year follow-up was conducted in 5 health plans in California, Washington, and Massachusetts. Data were collected via telephone interviews with parents at baseline and 1 year, surveys of practice sites and clinicians, and computerized databases. The practice site survey asked about policies to promote cultural competence, the use of several types of reports to clinicians, support for self-management of asthma, case management and care coordination, and access to and continuity of care. Quality of care was evaluated on the basis of 5 measures: 1) preventive medication underuse based on parent report; 2) the parent’s rating of asthma care; 3) the 1-year change in the child’s asthma physical status based on a standardized measure; 4) preventive medication underprescribing based on computerized data; and 5) the occurrence of a hospital-based episode.
Results. Of the 1663 children in the study population, 67% had persistent asthma at baseline based on parent report of symptoms and medications. At 1-year follow-up, 65% of the children with persistent asthma were underusing preventive medication based on parent report. In multivariate analyses, patients of practice sites with the highest cultural competence scores were less likely to be underusing preventive asthma medications based on parent report at follow-up (odds ratio [OR]: 0.15; 95% confidence interval [CI]: 0.06–0.41 for the highest vs lowest categories) and had better parent ratings of care. The use of asthma reports to clinicians was predictive of less preventive medication underprescribing based on computerized data (OR: 0.33; 95% CI: 0.16–0.69), better parent ratings of care, and better asthma physical status at follow-up. Patients of practice sites with policies to promote access and continuity had less underuse of preventive medications (OR: 0.56; 95% CI: 0.34–0.93). Among the 83 practice sites, the practice site’s size, organizational type, percentage of patients insured by Medicaid, mechanism of payment for specialty care, and other primary care features were not consistently associated with quality measures.
Conclusions. Practice-site policies to promote cultural competence, the use of reports to clinicians, and access and continuity predicted higher quality of care for children with asthma in managed Medicaid.
Key Words: asthma • Medicaid • cultural competence • quality of care
Abbreviations: ACQA, Asthma Care Quality Assessment • ED, emergency department • HMO, health maintenance organization • OR, odds ratio • CI, confidence interval
Medicaid-insured children with asthma pose important challenges in clinical practice. Poor and minority children are at elevated risk for asthma hospitalization, and many do not receive appropriate outpatient treatment.1–4 In general, Medicaid-insured children may receive worse care than privately insured children, even when they are in the same health plans.5 Scant information exists to help clinicians and policy makers understand how to improve health care quality effectively for children in managed Medicaid arrangements.6
Clinicians who treat Medicaid-insured children need better information about which practice-site policies most enhance quality of care for this vulnerable group. Three general categories of practice-site policies are those that are tailored to low-income populations, chronic care policies, and primary care elements. Policies that are tailored to low-income populations attempt to promote cultural competence,7–9 communication, and empowerment. Policies described in the Chronic Care Model include self-management support, decision support, clinical information systems, delivery system design, health care organization, and community resources.10 Primary care elements have been defined as first-contact care (commonly called access), longitudinality (continuity), comprehensiveness, and coordination.11 Other features of practices that could potentially influence health care quality include how care is financed (capitated or fee for service), size, organizational type (eg, private office or community health center), and whether the population is predominantly Medicaid insured or commercially insured.
This study addresses the critical gap in information about which practice-site policies and features influence asthma care quality for Medicaid-insured children. This research is unique in that we measured potential predictors simultaneously at the practice site, clinician, and patient levels, then evaluated quality based on parent report as well as computerized data over a 1-year follow-up period. Our aims were to 1) evaluate asthma care quality for children with managed Medicaid and 2) identify practice-site and clinician-level predictors of asthma care quality for these children.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Data Collection
Four respondent groups contributed data for this study: 1) parents of children with asthma, via telephone interviews at baseline and 1-year follow-up; 2) clinicians who were the primary care clinicians for these children, via written surveys at baseline and follow-up; 3) practice site administrators, via written surveys; and 4) health plan administrators and other key informants, via face-to-face and telephone interviews in which we gathered contextual information. Health care use data, including hospitalizations, emergency department (ED) visits, outpatient visits, and medications, were collected by analysis of computerized databases for the years before and after the baseline interviews. Asthma-related use was identified using International Classification of Diseases, Ninth Revision 493.0 to 493.9 and a list of appropriate medications.
The target population was Medicaid-insured children aged 2 to 16 years with asthma confirmed by both computerized utilization data and parent report. Potentially eligible children were identified via 1) a physician’s diagnosis of asthma based on an International Classification of Diseases, Ninth Revision code of 493.0 to 493.9 at an outpatient clinic visit, ED visit, or hospitalization or 2) 1 or more prescriptions for cromolyn, nedocromil, or inhaled corticosteroids or 2 or more prescriptions for ß-agonist medications during the 12 months before October 1998, the index date for cohort identification. All children identified as potentially eligible were included in interview and survey attempts, except at Neighborhood Health Plan, where we drew a random sample of 1000 because of the large population size.
Parent Interviews
Using a closed-ended telephone interview, we collected information from parents about the child’s asthma status, medication use, parent reports and ratings of care, demographics, and family structure. Measures were based on previously validated instruments, including the American Academy of Pediatrics Children’s Health Survey for Asthma and the Consumer Assessment of Health Plans Survey instrument for childhood asthma.12 Baseline telephone interviews were conducted during February to October 1999, with follow-up interviews 1 year later. Surveys with identical content were mailed to families who were not reached by telephone. A $10 incentive for interview participation was offered, and the study was approved by the Institutional Review Boards at participating sites.
Clinician Surveys
The primary provider for each child in the initially identified sample was identified using utilization and personal physician databases at the time of cohort identification. Each clinician was mailed a closed-ended, self-administered survey at baseline and at follow-up. The survey asked clinicians to report their experiences with selected features of asthma care, for example, whether they used asthma guidelines, reports, registries, written management plans, or disease management programs; had experienced any barriers to specialty referrals; or had been subject to any financial incentives that might influence asthma care. To gauge clinicians’ propensity to use inhaled anti-inflammatory medications, we presented 4 vignettes that described standardized patients.13 The survey also included questions on the demographic characteristics of the clinicians and their practice groups.
Practice Site Surveys
Two surveys were mailed to each practice site, defined as a provider or a group who saw patients in the same location(s). These closed-ended, self-administered instruments were usually completed by a clinician or a nurse manager at each site, sometimes with the input of other clinicians. The surveys asked questions about the basic features of each practice site, including its size, organizational type (eg, community health center, health maintenance organization [HMO], private office) and the demographics and insurance coverage of its patient population. It included questions about selected practice-site policies for low-income populations, chronic care, and primary care.10,11 Details about the specific items asked for each policy are included in Table 1. For example, there were 6 questions on cultural competence policies, including whether the practice attempted to recruit ethnically diverse providers or offered cross-cultural or diversity training. Questions on care for low-income populations were derived from literature review and focus groups with clinicians and staff at 3 urban primary care clinics.
|
|
The following practice-site policies were identified as potentially important predictors in a priori hypotheses: cultural competence (composite score), support for continuity of care, use of reports to clinicians about asthma patients, and use of guidelines by clinicians. In addition, a priori hypotheses specified the following practice-site features as predictors of interest: percentage of patients with Medicaid, type of practice site (multispecialty group practice, academic health center, community health center, or solo or 2-person practice), size of practice group, and method of payment for specialty care (fee for service vs capitated).
Statistical Analysis
The multi-item measures for each policy in the practice site survey (eg, cultural competence) were refined using tests of inter-item correlation, principal component analysis, and factor analysis. Bivariate analyses to evaluate the association of each quality measure with each predictor included the 
2 test for categorical variables and the Wilcoxon rank-sum test and Spearman correlation coefficient for ordinal or continuous variables.
In the first stage of multivariate analysis, we created a separate multivariate model that evaluated the association between each eligible predictor of interest and each quality measure while adjusting for demographics (age, gender, and race) and baseline asthma status (asthma physical status and medication dispensing). To be eligible for the first-stage multivariate analysis, a predictor variable had to be associated with the dependent variable at P 
.10 in bivariate analysis or identified in the conceptual model’s a priori hypotheses. Income, education, and employment were not significantly associated with quality measures in bivariate analyses and thus were not routinely included in models. In the final stage of modeling, we created iterative, forward-selection models for each quality measure. The final models sought to identify independent effects among predictor variables that were significant at P .10 in the first stage. Demographic and asthma status variables were forced into the final models.
Models used Poisson regression for rates (eg, hospital-based episodes), logistic regression for dichotomous variables (eg, preventive medication underuse), and linear regression for continuous variables (eg, asthma physical status). Parameter estimation, including regression coefficients and associated standard errors, was conducted using generalized estimating equation techniques to account for the fact that patients of a given clinician may be correlated and thus are not independent observations.16 Clustering of clinicians by practice site and practice sites by health plan was not included in the analysis. We limited our adjustment to clustering of patients by physician because a previous study suggested that within-physician correlation was likely to be stronger than within-facility correlation.17
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The study group was 38% black, 19% Latino, and 31% white (Table 2) . Among the families, 53% had household incomes below the poverty level. Among the children, 67% had persistent asthma at baseline based on symptoms and/or medications reported by the parent. The hospitalization rate during the year before the baseline interview was 3.8, and the ED visit rate was 13.3 per 100 person-years of follow-up.
|
Most practice sites in this study were multispecialty group practice or HMO centers (48%) or community health centers (31%). Practice sites varied widely in their proportions of patients insured by Medicaid and proportions of minority patients (Table 3).
|
ranging from .70 to .86, denoting high reliability based on inter-item correlation. The final cultural competence summary score included 6 items. On the basis of the results of principal component analysis, 3 additional items on communication-related practices (Table 1) did not fit with the others in the cultural competence score; each of these was treated as a separate predictor variable in analyses. Practice sites varied widely in their policies to support cultural competence (Table 1). Although 71% reported recruiting ethnically diverse nurses and clinicians, only 23% offered training to clinicians to develop communication skills. Practice sites also varied widely in their use of reports to clinicians. Only 15% routinely provided clinicians with lists of asthma patients, although 30% gave feedback reports to clinicians to improve performance in care.
Policies Predictive of Asthma Care Quality
Quality Measures Based on Parent Report
Of the 729 children with persistent asthma by parent report, 65% were underusing preventive medications based on parent report of less-than-daily use. The mean rating of overall asthma care given by parents was 8.6 on a scale of 1 to 10, with a standard deviation of 1.9. Among all children, 17% experienced a decline of 0.5 or more standard deviations in their asthma physical status score between baseline and follow-up.
Table 4 shows the results of final multivariate models that simultaneously controlled for all significant predictors. Underuse of preventive asthma medications was lower among patients of practice sites with higher scores for cultural competence (odds ratio [OR]: 0.15 for the highest vs lowest scoring sites; 95% confidence interval [CI]: 0.06–0.41; Table 4). Preventive medication underuse was also less common among practice sites that had policies to promote access and continuity (OR: 0.56 for the highest-scoring sites vs all others; 95% CI: 0.34–0.93). These associations were identified in first-stage multivariate models that adjusted for demographics and baseline asthma status, as well as in the final multivariate model that evaluated both predictors simultaneously.
|
Quality Measures Based on Computerized Data
Of the 316 children with persistent asthma based on computerized data, 21% had medication underprescribing. The number of patients with persistent asthma based on computerized data was smaller than the number based on parent report because the computerized criterion was more stringent, likely resulting in a higher severity subgroup of children. The rate of hospital-based episodes was 11.0 per 100 person-years, including 2.7 hospitalizations and 9.5 ED visits.
Underprescribing of preventive medication based on computerized data was lower among practice sites that issued lists of asthma patients to clinicians (OR: 0.33; 95% CI: 0.16–0.69) and among clinicians who routinely gave written asthma management plans to families (OR: 0.27; 95% CI: 0.13–0.54; Table 4). Several other predictors were associated with less underprescribing of preventive medication in the first-stage models only. These included the use of self-management tools, policies to promote access and continuity, and the receipt of routine reports by the clinician.
Hospital-based episodes were associated with several predictors in first-stage models, but none of these predictors was significant in the final multivariate model (Table 4). In first-stage models, patients were less likely to have hospital-based episodes when their practice site–issued reports to clinicians or used self-management tools or when their clinician said that they were aware of disease management programs for asthma or received routine reports about their asthma patients.
In secondary analyses, adjusting for HMO did not change the magnitude of most of the observed associations, although selected findings became statistically less significant. In the secondary models, HMO was not independently associated with any quality measure except hospital-based episodes (P = .0051), for which the commercial HMOs had lower rates than Medicaid HMOs.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This work is unique in that we evaluated predictors of quality from the practice site, clinician, and patient levels simultaneously in multiple settings in 3 states. Previous studies have identified variables associated with asthma care processes or outcomes by measuring the characteristics of patients and clinicians.3,18,20–24 However, research that evaluates variables from practice sites or that combines variables from practice sites, clinicians, and patients is uncommon, perhaps because of the complexity of identifying the interrelationships among these entities, then collecting and linking data from multiple classes of respondents.
Practice-Site Policies
This study provides some of the strongest available evidence that policies to promote cultural competence are associated with quality of care. Cultural competence has received increasing attention in recent years, especially in managed Medicaid.25,26 Many studies suggest that cultural issues influence the effectiveness of health care.27–32 However, scant research exists on whether health care system efforts to promote cultural competence are effective or about which techniques are most useful.8 Our findings suggest that cultural competence policies are associated with specific aspects of health care quality in low-income populations and that this association deserves additional testing in interventional studies.
Among policies to support chronic care, reports to clinicians emerged as being associated with more appropriate medication prescribing, better parent ratings of care, and improved asthma physical status. Methods of reporting ranged from providing clinicians with lists of asthma patients to giving clinicians feedback reports to stimulate improvements in care or using prompts to enhance guideline adherence during individual patient encounters. The most simple of these methods—giving clinicians lists of their patients with asthma—was associated with 2 of the quality measures. This suggests that policies to improve chronic care do not need to be complex or highly technology dependent to be effective.
In the seminal description of primary care, access (first-contact care) and continuity (longitudinality) are 2 separate concepts.11 These were combined as a single measure in our final analyses because in preliminary analyses, the items that measured these concepts were highly correlated and seemed to represent a single factor. Our findings that access and continuity were associated with better parent-reported use of preventive medications is in accordance with other studies suggesting that continuity of care is linked with desirable outcomes.33–37
Other practice-site policies that were associated with quality measures in preliminary multivariate models included case management, support for self-management, and the use of written asthma management plans. These variables were not statistically significant in final models, but this in part might have been because these policies were highly correlated with other policies that accounted for more of the variation in quality. We interpret these results as suggesting that these policies are potentially useful, although the evidence was not as strong as that for other policies with independent effects in the final models.
Quality Measures
This study measured quality on the basis of both parent interviews and computerized data. Unsurprising, different practice-site policies were associated with different measures of quality. Among our quality measures, the process-based measures (preventive medication underuse and underprescribing, and parent ratings of care) seemed to be more sensitive indicators than the outcome-based measures (hospital-based episode rates and asthma physical status scores). Hospital-based episode rates were not significantly associated with any predictor variables in the final multivariate model and may be less modifiable by quality improvement efforts than other measures.
Limitations
Some of the policies that we attributed to practice sites may actually have represented or have been highly influenced by the policies of the health plans with which the practice sites were associated. For example, some of the health plans in this study routinely distributed reports to clinicians about their asthma patients. However, a preliminary study that we conducted suggests that there is often high variability in policies even among practice sites associated with the same health plan.38 Because this study included only 5 health plans, we did not attempt to evaluate specific health plan features as predictors of quality. The children in this study were served by relatively mature and stable health plans, and these findings may not generalize to all managed Medicaid arrangements.
The observational design of this study precludes inferences that practice-site policies actually caused improvements in quality of care. The policies that we identified could be simply a marker for other, unstudied characteristics that actually caused the variation in quality. For example, practice sites with policies to promote cultural competence or reports to clinicians might also be better organized or more responsive to patients’ needs in other ways that we did not evaluate. Another explanation of the association could be that practice sites that were already delivering high-quality care might have had the luxury of initiating activities such as cultural competence policies. Interventional studies are required to evaluate both the effectiveness of specific practice-site policies and the effectiveness of quality improvement programs that promote such policies.
Most variables in this study were collected via surveys of practice-site administrators, clinicians, and parents. Although nonresponse could affect the results, the response rate of 63% is considered excellent for a study of Medicaid patients.39 The multi-item measure of cultural competence that we developed was based on questions derived from focus groups and existing literature. The measure was reliable on the basis of inter-item correlation and had construct validity in that it was associated with quality of care. Various other instruments have subsequently been developed to evaluate cultural competence, but limited data exist about their reliability and validity.26 Additional development of standardized measures of cultural competence is warranted.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We thank Bernard Friedman, PhD, Irene Fraser, MD, Charles Cutler, MD, and Marie Mann, MD, at our funding agencies for support throughout the project. We thank our senior advisors, Joe Selby, MD (Kaiser Permanente), Rich Platt, MD (Harvard Pilgrim Health Care), and Ed Wagner, MD (Group Health Cooperative) for thoughtful guidance. We are grateful for the contributions of the medical directors and analysts of the other participating health plans, including Chris Cammisa, MD, and Cindi Turner of Partnership Healthplan of California and Robert Master, MD, James Glauber, MD, and Ming Yuan of Neighborhood Health Plan.
This study benefited from the input of Lynn Olson, PhD, and Linda Asmussen, MS, of the American Academy of Pediatrics, who shared the Children’s Health Status Assessment for Asthma, and of Charles Homer, MD, and William Vollmer, PhD. We appreciate the efforts of Kachen Streiff, MS, project coordinator at the Harvard site, and of Charlene Gay, who assisted with manuscript preparation.
This project’s success is attributed to the efforts of our research assistants Georgina Berrios, Sylvia Chan, Yoko Chavez, Pete Dorin, MPA, Danielle Hauptman, Diane Hinkley, Regina Mason, Valerie Ramirez, MS, Monica Soto, and Lynda Tish. We gratefully acknowledge the many parents, clinicians, practice-site administrators, and health plan administrators who contributed perspectives for this research.
|
FOOTNOTES |
|---|
Reprint requests to (T.A.L.) Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, 133 Brookline Ave, 6th Fl, Boston, MA 02215. E-mail: tracy_lieu{at}hphc.org
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. B. Diette and C. Rand The Contributing Role of Health-Care Communication to Health Disparities for Minority Patients With Asthma Chest, November 1, 2007; 132(5_suppl): 802S - 809S. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. D. Cabana, M. Lara, and J. Shannon Racial and Ethnic Disparities in the Quality of Asthma Care Chest, November 1, 2007; 132(5_suppl): 810S - 817S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Lara, M. D. Cabana, C. R. Houle, J. W. Krieger, L. L. Lachance, J. R. Meurer, M. P. Rosenthal, and I. Vega Improving Quality of Care and Promoting Health Care System Change: The Role of Community-Based Coalitions. Health Promot Pract, April 1, 2006; 7(2_suppl): 87S - 95S. [Abstract] [PDF]
|
|
|
|
 |
|
 G. M. Hunninghake, S. T. Weiss, and J. C. Celedon Asthma in Hispanics Am. J. Respir. Crit. Care Med., January 15, 2006; 173(2): 143 - 163. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Lurie, M. Jung, and R. Lavizzo-Mourey Disparities And Quality Improvement: Federal Policy Levers Health Aff., March 1, 2005; 24(2): 354 - 364. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
