National guidelines recommend daily use of controller medications for children with persistent asthma. Although studies suggest low rates of controller use, little is known about predictors of underuse among low-income children in whom asthma morbidity is greatest.
To determine the frequency of underuse of controller medications among Medicaid-insured children in a variety of managed care arrangements, and to examine demographic factors and processes of asthma care associated with underuse.
Cross-sectional telephone survey of parents of children and adolescents aged 2 to 16 years with asthma, insured by Medicaid, and enrolled in 1 of 5 managed care plans. The main outcome was parent-reported underuse of controllers among children with persistent asthma. Survey items included demographic factors and reports of specific processes of care. Current symptom level was determined by recall of the number of days with symptoms in the previous 14 and by the Physical Function Score of the American Academy of Pediatrics (AAP) Child Health Status Assessment for Asthma. Logistic regression was used to identify factors independently associated with underuse.
The response rate was 66%, with 1648 children included in the analysis; 1083 were classified as having persistent asthma. Of these, 73% were underusers of controller therapy, with 49% reporting no controller use and 24% reporting less than daily use. A multivariate model that adjusted for age, managed health care organization, and AAP Physical Function Score found that black (odds ratio [OR], 1.7; 95% confidence interval [CI], 1.2-2.4) or Latino (OR, 2.2; 95% CI, 1.3-3.8) race were associated with underuse and that parental education beyond high school was protective (OR, 0.6; 95% CI, 0.4-0.8). Having a primary care physician (OR, 0.4; 95% CI, 0.2-0.8), written action plan (OR, 0.5; 95% CI, 0.4-0.7), or a follow-up visit (OR, 0.5; 95% CI, 0.4-0.8) and having seen an asthma specialist (OR, 0.5; 95% CI, 0.4-0.7) were associated with lower rates of underuse.
Underuse of controller medications among Medicaid-insured children is widespread. Racial minorities and children whose parents are less educated are at higher risk for underuse. Patients who have received action plans or had follow-up visits or specialty consultations are less likely to be symptomatic underusers of controller medications.
ASTHMA PREVALENCE and morbidity remain high among children of low-income families compared with their more advantaged peers.1- 3 Although reasons for such differences are complex, a broad literature suggests that poor children, especially of racial and ethnic minorities, lack access to appropriate primary health care services1,4,5 and may not receive the range of available, effective treatment modalities recommended in national guidelines.6 Such guidelines focus on asthma as a chronic, inflammatory condition. Recommended pharmacological management includes use of anti-inflammatory agents as controller medications for all children with persistent asthma, defined as those with symptoms occurring more frequently than 2 days per week or with nighttime symptoms occurring more than twice per month. Anti-inflammatory agents, especially inhaled corticosteroids, have been shown to decrease symptoms7,8 and hospitalizations.9- 11 Routine checkups for asthma and the use of written, structured action plans are believed to help implement these guidelines and improve patient adherence to prescribed treatments.6
Previous studies12- 18 have documented underuse of controller medications (inhaled corticosteroids and mast-cell stabilizers) in general and particularly in low-income populations.19,20 The Asthma Care Quality Assessment (ACQA) Study is an investigation of patterns of asthma-related health care for Medicaid-enrolled patients in 5 managed health care plans. Given the high levels of asthma morbidity in low-income children, we must understand better the reasons for underuse of controller medications and what might alter these treatment patterns. Therefore, as part of this study, we sought to identify demographic factors and processes of care that are associated with underuse of controller medications among children with evidence of persistent asthma.
The data used for this analysis were collected during the baseline telephone survey for the ACQA Study. The following 5 large, nonprofit health care plans participated: Harvard Vanguard Medical Associates (an affiliate of Harvard Pilgrim Health Care) in Massachusetts, Kaiser Permanente in Northern California, Group Health Cooperative in Washington State, Partnership Healthplan of California, and Neighborhood Health Plan in Massachusetts. The former three are group-model health maintenance organizations (HMOs) in which a large provider group contracts exclusively or almost exclusively with the health insurance plan. Medicaid members are in the minority in these health plans, in which most members are commercially insured. In contrast, the latter two are mixed-model Medicaid managed care organizations (MCOs) that contract with a variety of provider groups, including community health centers.
Automated registration and claims files of each health care plan were used to identify Medicaid-insured children and adolescents aged 2 to 16 years who met at least 1 of the following criteria: an asthma-related ambulatory, emergency department visit or hospitalization in the past 12 months or medication use suggestive of a diagnosis of asthma. The medication use criteria included at least 2 dispensings of a β-agonist or at least 1 dispensing of a controller medication (inhaled corticosteroid, cromolyn sodium, or nedocromil sodium).
The study was approved by the institutional review board at each of the study centers. An initial contact letter included elements of informed consent and offered a $10 incentive for participation. Research assistants conducted telephone interviews in English or Spanish from a single site from February through October, 1999. During the initial telephone call, parents confirmed the diagnosis of asthma in their child and provided verbal consent to participate. In August and September 1999, an identical version of the survey was mailed to families not yet reached by telephone. The interview included demographic items such as patient age and race and parental age and education. Access to health care was assessed by whether parents reported the child to have a primary care physician. Items reflecting processes of care received in the previous 6 months included receiving a written care plan, having an asthma-related follow-up visit (defined as going to a doctor's office or clinic for asthma follow-up care or checkups), and having seen an asthma specialist. Current asthma burden was assessed using the Physical Function Score of the American Academy of Pediatrics (AAP) Child Health Status Assessment instrument for asthma.21 Medication use was determined by asking parents to list prescription asthma medications they had at home, and for each, how often they had been used by their child in the past 2 weeks (daily, 3-6 d/wk, 1-2 d/wk, or not at all).
Current symptoms were determined by asking about the number of days during the past 14 with any asthma symptom (including cough, wheeze, shortness of breath, or limited activity). We categorized patients as having persistent asthma if they were symptomatic on 5 or more days during the past 14, corresponding to the definition of persistent asthma (>2 symptom-days per week) in the National Asthma Education and Prevention Program (NAEPP) guideline.6 We also included patients using daily controller medications in the persistent asthma category, even if they did not meet the symptom threshold, under the assumption that, without medication, they would have experienced symptoms. Children with symptoms less frequent than 5 days during the past 14 (and no daily controller use) were considered to have intermittent asthma.
According to the NAEPP guideline, children with persistent symptoms should receive daily controller medication (including inhaled corticosteroids or mast-cell stabilizers). We considered patients to be underusers if no use (nonusers) or less than daily use of such medications was reported. We did not include leukotriene antagonists, long-acting β-agonists, or theophylline in our definition of controllers, as rates of use of these medications were low during the study period. Our bivariate analyses compared underusers of controller medications with those who reported daily use among patients with evidence of persistent asthma (Figure 1). Patients for whom only intermittent symptoms were reported and who were not daily users of controller medications were excluded from these analyses.
Division of respondents by symptom burden and self-reported medication frequency. Data are given as number (percentage) of patients. AI indicates anti-inflammatory agent, including inhaled corticosteroids, cromolyn sodium, and nedocromil sodium. The patients in the shaded areas were excluded from analyses.
We used the general approach of Aday and Andersen22 to build logistic regression models to examine associations of demographic and process-of-care variables with underuse (vs daily use) among symptomatic children. We began by including patient age, MCO, and AAP Physical Function Score (a proxy for asthma symptom severity) based on the a priori judgment that these were potential confounders for which adjustment is usually made in asthma research. We tested patient race, parental education, and family income in a stepwise fashion and retained predictors that were significant at P<.10. We then tested having a primary care physician as a variable representing general access to health care and again retained it if the results were significant. Finally, we tested whether reports of the following specific processes of care in the past 6 months were associated with underuse: receiving a written action plan, having had an asthma-related follow-up visit, and having seen an asthma specialist. Odds ratios (ORs) and 95% confidence intervals (CIs) are reported for the final model only. We performed all analyses using SAS Version 8 (SAS Institute, Cary, NC). To evaluate the predictors of controller medication underuse in a subgroup of extremely symptomatic children (at least 10 symptom-days during the past 14), we applied the final model to predict underuse in these patients.
We identified a sample of 4094 children with asthma based on health plan data. After excluding 603 (15%) who were no longer plan members and another 141 (3%) who returned postcards declining participation, we attempted contact of 3350 parents of eligible patients with asthma and completed contact with 2568. Of these, 628 were deemed ineligible because they did not have asthma (n = 430), had a sibling in the study (n = 66), had a language barrier (n = 65), were no longer enrolled in the health plan (n = 54), or for other reasons (n = 13). A total of 1663 consented to participate in the interview for a response rate of 66%, based on the method of the Council of American Survey Research Organizations, which subtracts the estimated number of ineligible nonrespondents from the denominator.23 An additional 15 respondents were excluded from this analysis because they did not answer the primary symptom frequency question, for a final sample of 1648. Table 1 gives the demographic characteristics of the sample. Among the identified patients with asthma, we found a predominance of boys (58% vs 42%). The sample included representation of white (31%), black (38%), and Latino (19%) patients; the remaining 12% were categorized as "other." The median educational level of parents was a completed high school education.
Figure 1 shows the division of respondents by symptom burden and self-reported frequency of controller use. We categorized a total of 1083 children as having persistent asthma for this analysis. Of these, 980 patients (90%) reported at least 5 symptom-days during the past 14, with 651 (66%) of these reporting at least 10 symptom-days during the past 14. An additional 103 (10%) reported daily use of controller medication with intermittent or no symptoms. Among children with persistent asthma, 293 (27%) reported daily use of controller medication as recommended by the NAEPP guideline.6 The remaining 790 (73%) were designated controller medication underusers, with 529 (49%) reporting no controller medication use and 261 (24%) reporting use less frequently than every day. The other 565 patients (shaded categories in Figure 1) reported fewer than 5 symptom-days during the past 14 and were not daily users of controller medication. These patients were excluded from the main comparisons of this study, which sought to identify predictors of daily controller medication use among the 1083 patients who met criteria for persistent asthma.
Table 2 presents the association of demographic factors and process-of-care variables with controller medication underuse. Underuse varied with age, and was more common in children younger than 4 years (79%) and in adolescents (81%) compared with school-aged children (P<.01). Parents with more than a high school education were less likely than those with a high school degree or less to have a child who underused controller medications (67% vs 78%; P<.001). Race was also a significant predictor, with parents of black (75%) and Latino (79%) patients reporting higher rates of underuse than parents of white patients (69%) (P<.001). Self-reported family income was not associated with differential underuse.
Variables that reflect access to medical care and processes of care were also associated with differential rates of underuse. Patients who reported having no primary care physician had higher rates of underuse (88%) compared with those who did (71%) (P<.001). Children who had not seen an asthma specialist were also more likely to report underuse (77% vs 55%; P<.001). Having been given a written care plan was associated with lower likelihood of underuse of controller medications (60% vs 79%; P<.001). A similar result was seen for having had a follow-up appointment in the past 6 months (68% vs 85%; P<.001).
The final multivariate model simultaneously adjusted for age, MCO, AAP Physical Function Score, and demographic and process-of-care variables (Table 3). Race remained a significant predictor of underuse, with an adjusted OR of 1.7 (95% CI, 1.2-2.4) for black race and 2.2 (95% CI, 1.3-3.8) for Latino race. Parental education beyond high school also remained protective against underuse (OR, 0.6; 95% CI, 0.4-0.8). Patients with a primary care physician were less likely to report underuse (OR, 0.4; 95% CI, 0.2-0.8). The magnitude of the independent effects of having a written care plan (OR, 0.5; 95% CI, 0.4-0.7) or a follow-up visit (OR, 0.5; 95% CI, 0.4-0.8) and having seen a specialist (OR, 0.5; 95% CI, 0.4-0.7) were similar.
We used the threshold from the NAEPP guidelines for defining persistent symptoms (>2 symptom-days per week). We also tested whether the same variables would be associated with underuse among children who were extremely symptomatic (≥10 symptom-days during the past 14). In this analysis, we found virtually identical point estimates and CIs for each of the variables of interest.
As a secondary analysis, we examined whether correlates of controller underuse differed for patients reporting no use compared with those reporting partial use (1-6 d/wk). Among nonusers, ORs for demographic variables were almost identical to those in the primary analysis. Associations with having a primary care physician (OR, 0.4; 95% CI, 0.2-0.8), receiving an action plan (OR, 0.4; 95% CI, 0.3-0.6), having had a follow-up visit (OR, 0.5; 95% CI, 0.3-0.7), and having seen a specialist (OR, 0.4; 95% CI, 0.3-0.7) were significant and similar to those found in the primary analysis. In models predicting partial controller use, the results were also similar to those of the primary analysis, although the ORs for black (OR, 1.5; 95% CI, 1.0-2.4) and Latino (OR, 1.8; 95% CI, 1.0-3.5) race no longer reached statistical significance at a level of P<.05. Having a primary care physician (OR, 0.9; 95% CI, 0.4-2.1) was not associated with partial controller use, but the remaining 3 process-of-care measures, ie, receiving an action plan (OR, 0.7; 95% CI, 0.5-1.0), having had a follow-up visit (OR, 0.7; 95% CI, 0.4-0.9), and having seen an asthma specialist (OR, 0.7; 95% CI, 0.5-1.1), showed effects in the same direction, although of decreased magnitude.
These data show substantial underuse of anti-inflammatory agents by Medicaid-insured children with asthma in 5 geographically dispersed health care plans. We found 72.9% percent of children with persistent asthma were underusers of these medications; approximately two thirds of these patients reported no use of anti-inflammatory agents. We examined demographic factors and process-of-care variables. Associations with the latter are particularly important, since they may point toward effective strategies to improve health care. Minority race was associated with underuse, and education beyond high school was protective. All 3 of the process-of-care measures examined were associated with lower risk for underuse, even when entered into a single model that controlled for current symptom burden and all significant demographic predictors.
Underuse of controller medications in the presence of persistent symptoms is concerning and is not limited to Medicaid-insured populations. Goodman et al24 found that only 29% of children in an HMO who were prescribed an anti-inflammatory agent had more than 2 canisters dispensed during a 1-year period. A study from the Pediatric Asthma Care PORT (Patient Outcomes Research Team) reported that 69% of those with 3 dispensings of a β-agonists had a controller medication dispensed, and only 48% had repeated dispensings (≥3).17 In surveys of MCO members, Diette et al16 reported that 64% of adult users of inhaled corticosteroids reported underdosing (use <5 d/wk). This group more recently reported similar data for children in MCOs, and found only 55% of patients with moderate and severe asthma reported daily use of long-term controller medications.18 Early work by Bosco et al19 showed low rates of use of asthma medication and inhaled steroids among Medicaid-insured individuals in particular. More recent studies in Medicaid-insured and inner-city populations support and extend this finding.5,20,25 For example, Eggleston et al20 reported that of patients with 8 or more symptom-days in the past month, only 12% were taking anti-inflammatory agents.
The underlying reasons for underuse of controller agents remain unclear. A recent comparison of Medicaid-insured and commercially insured children in an HMO showed differences in dispensing rates of controller agents, but similar rates of prescriptions written.26 This finding raises the possibility that reported underuse may be the result of patient adherence to prescribed regimens as well as underprescribing by clinicians. Leickly et al27 identified concern about adverse effects and doubts about the usefulness of medications as common in the inner-city sample they interviewed. Focus groups of parents of low-income children with asthma confirm that health beliefs and misunderstanding about preventive asthma medications are important factors in adherence.28 The present study, based on parental report without confirmation by means of medical record review, provides an estimate of what is actually being received by patients but cannot separate the effects of inadequate prescribing of controller medications from patient adherence to prescribed preventive regimens. Even among children with asthma in a study monitoring use of inhaled medication, parent report has been shown to overestimate, by up to 50%, what is actually received.29
The finding that race and parental education are associated with medication underuse, even among insured low-income populations, is disturbing, but is consistent with the findings of other studies on access to prescription medications.30 In asthma, patient race or ethnicity has been associated with increased prevalence31- 34 and morbidity5,35 and lower quality of health care.36,37 Nonwhite race has been associated specifically with suboptimal medication regimens in adults.16 A separate analysis of racial and ethnic variation in the processes and outcomes of asthma care is currently underway as part of the ACQA Study.38
A structured, proactive approach to caring for children with asthma has been outlined by the NAEPP guidelines.6 This stepwise approach recommends regular asthma checkups to assess current asthma symptoms and to optimize preventive medication regimens. These steps are intended to ensure symptom control and that children are prescribed the lowest effective dose of controller medications. An observational study in a single HMO has suggested a benefit of written plans.39 The present study, in a Medicaid-insured population, also suggests that written plans and follow-up visits are associated with more appropriate medication use. Of course, in this cross-sectional study, we are not able to discern whether these processes of care are responsible for improved medication use or are markers for other health care processes or unmeasured patient attributes. However, given that these techniques are tested, inexpensive methods to standardize medical care for chronic disease,40 this study supports attempts to increase the use of such asthma care plans by primary care providers for low-income children.
Our finding that having seen an asthma specialist is associated with higher rates of use of anti-inflammatory agents is consistent with the results of studies in non–Medicaid-insured populations, such as those found in a large HMO.13 Those patients who seek or are referred to specialists may be more motivated or have more severe disease. However, studies suggest that specialists prescribe more guideline-adherent regimens, independent of disease severity,13,41 and may have systems that allow more effective patient education.42 The finding that specialty care is associated with more appropriate medication use among low-income children reinforces the need for access to specialists, or equivalent systems of care, for Medicaid-insured and other low-income children.
The strengths of this study include its relatively large sample of low-income children, with an overall response rate of 66%. The number of collaborating health plans ensures geographic and health care delivery system diversity. Previous studies using the administrative data of health plans only have been criticized for inferring asthma symptom levels by counting pharmacy dispensings of β-agonist medications. It is not clear whether automated pharmacy data or patient reports should be considered the gold standard for estimation of medications received. With automated data, it is impossible to know the frequency of use of these medications at any point in time. In this study, we have concurrent reports of symptoms, β-agonist use, and controller use. However, by relying on parental report, we may still overestimate medication use by these children.29 By repeating the analysis on a subgroup of extremely symptomatic children, we have shown that the associations of underuse with demographic and process-of-care variables are similar, even in patients with moderate to severe asthma.
Although Medicaid managed care programs are expanding throughout the country, these data may not be representative of all Medicaid enrollees. However, we have no reason to believe that the effects of demographic and process-of-care indicators would differ for Medicaid-insured patients in fee-for-service systems. We excluded from analysis patients with intermittent symptoms who were not taking their controller medication daily. Patients who had a controller medication prescribed may have had a diagnosis of persistent asthma; however, since these patients experienced good asthma control (during the 14 days before the interview), we did not feel justified in categorizing them as underusers. Finally, a cross-sectional study cannot establish a causal relationship between, for example, the use of written care plans and decreased underuse of controller medications. We report associations with these process-of-care measures but rely on supporting literature to conclude that institution of written action plans and asthma checkup visits among those who do not now receive them will actually decrease underuse of controller agents among symptomatic children.
Even with these caveats, the fact that we are able to show independent effects of demographic factors and specific processes of care in this population of low-income children is important. We have identified straightforward actions that providers and parents can take that may enhance the quality of health care received by low-income children. Given the well-documented effectiveness of controller medications in reducing morbidity due to asthma, adoption of these measures may help to reduce some of the disproportionate burden of asthma on low-income children. Managed care plans serving Medicaid members should continue to experiment with structured ways of improving the quality of care provided to their enrollees.
Accepted for publication February 7, 2002.
The ACQA Project was supported by grant HS09935 from the Agency for Healthcare Research and Quality (Rockville, Md), by the American Associations of Health Plans Foundation (Washington, DC), and by the Maternal and Child Health Bureau (Rockville).
We thank the core study team, including Nancy Jensvold, MPH, Angela Capra, MA, and Felicia Chi, MPH, for their intellectual contributions and their hard work. This study would not have been possible without the support of senior leaders at each of the health care plans who provided access and guidance. We also thank Bernard Friedman, PhD, at the Agency for Healthcare Research and Quality for his thoughtful input and support of this work. We are grateful to the hundreds of parents and children who shared their experiences and opinions in the interviews for this study.
Corresponding author and reprints: Jonathan A. Finkelstein, MD, MPH, Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, 133 Brookline Ave, 6th Floor, Boston, MA 02215 (e-mail: Jonathan_Finkelstein@harvardpilgrim.org).
Finkelstein JA, Lozano P, Farber HJ, Miroshnik I, Lieu TA. Underuse of Controller Medications Among Medicaid-Insured Children With Asthma. Arch Pediatr Adolesc Med. 2002;156(6):562–567. doi:10.1001/archpedi.156.6.562