Electronically monitored adherence to medications by newly diagnosed patients with juvenile rheumatoid arthritis.код для вставкиСкачать
Arthritis & Rheumatism (Arthritis Care & Research) Vol. 53, No. 6, December 15, 2005, pp 905–910 DOI 10.1002/art.21603 © 2005, American College of Rheumatology ORIGINAL ARTICLE Electronically Monitored Adherence to Medications By Newly Diagnosed Patients With Juvenile Rheumatoid Arthritis MICHAEL A. RAPOFF, JOHN M. BELMONT, CAROL B. LINDSLEY, AND NANCY Y. OLSON Objective. To describe patterns of adherence to nonsteroidal antiinﬂammatory drugs (NSAIDs) in newly diagnosed patients with juvenile rheumatoid arthritis (JRA), and to examine demographic and disease-related variables as potential predictors of adherence. Methods. Adherence to NSAIDs was monitored in 48 children with JRA (mean age 8.6 years) over 28 consecutive days using an electronic monitoring device. Measures of disease activity (active joint counts, morning stiffness) and demographics (age, sex, ethnicity, socioeconomic status) were also obtained. Results. Using an 80% adherence cut point, 25 (52%) patients were classiﬁed as adherent and 23 (48%) as nonadherent. There was considerable variability across patients, with full adherence (taking all doses on time) ranging from 0 to 100% of the monitored days. Logistic regression showed that active joint count and socioeconomic status were the only signiﬁcant predictors. Both were positively related to adherence. The model correctly classiﬁed 70.5% of patients as either adherent or nonadherent (Cox and Snell R2 ⴝ 0.295, P ⴝ 0.0005). Conclusion. Children newly diagnosed with JRA are more likely to adhere to an NSAID regimen if they have a greater number of active joints or their families have higher socioeconomic status. The former ﬁnding suggests that children’s adherence is symptom-driven, while the latter suggests that families of lower socioeconomic status deserve special attention to address adherence issues. KEY WORDS. Adherence; Juvenile rheumatoid arthritis; Nonsteroidal antiinﬂammatory drugs; Electronic monitoring. INTRODUCTION Juvenile rheumatoid arthritis (JRA) is the most common pediatric rheumatic disease and a major cause of shortand long-term disability. Medical treatment objectives for JRA include reduction of pain and inﬂammation, preservation of function, and prevention of joint deformities and destruction. Comprehensive treatment of JRA includes pharmacotherapy (frequently nonsteroidal antiinﬂammatory drugs [NSAIDs]), physical and occupational therapy, and psychosocial support (1). Children with JRA and their parents are required to adhere consistently and over a Dr. Rapoff is supported by a grant from Maternal and Child Health (MCJ-200617). Michael A. Rapoff, PhD, John M. Belmont, PhD, Carol B. Lindsley, MD, Nancy Y. Olson, MD: University of Kansas Medical Center, Kansas City. Address correspondence to Michael A. Rapoff, PhD, University of Kansas Medical Center, Department of Pediatrics, 3901 Rainbow Boulevard, Kansas City, KS 66160-7330. Email: email@example.com. Submitted for publication May 25, 2005; accepted in revised form July 21, 2005. long period of time to medication regimens, such as daily NSAIDs, whose beneﬁts may be delayed even though adverse side effects such as gastrointestinal irritation are immediate. This constellation of factors associated with JRA and its treatment is predictive of greater adherence problems to medical regimens in pediatric chronic disease (2). There are few studies that have speciﬁcally addressed adherence to treatment regimens for JRA. Two retrospective studies by Litt and colleagues found that 55% of children and adolescents with JRA were adherent to salicylate medications as determined by serum assays (3,4). In 3 separate within-subject design studies involving 5 patients with JRA (ages 3 to 14 years), baseline adherence with medications was assessed by parental observations or pill counts and ranged from 38 –59% (5–7). Therefore, adherence to medications for JRA can vary widely across different patient samples and across methods of assessing adherence, but appears to be similar to the average adherence levels (50 –55%) found with medications for other chronic pediatric diseases (2). Although optimal levels of adherence for most regimens have not been established, these average adherence levels could compromise the ther905 906 apeutic beneﬁts of medical regimens for JRA and increase unnecessary health care costs (8). Because not all patients have problems with adherence, discovering factors that predict poor adherence would be useful for identifying those patients at risk who might beneﬁt from special attention to adherence issues. Moreover, factors that predict adherence could be used to improve the internal validity of clinical studies by stratifying patients on relevant dimensions (e.g., age, sex, socioeconomic status) before randomly assigning them to intervention or control groups. Few studies have examined predictors of adherence to medication regimens for JRA. Litt and Cuskey (3) found that lower adherence to salicylate medications among 38 adolescents with JRA was associated with younger age at disease onset, longer disease duration, shorter duration of subspecialty care, greater delay between disease onset and ﬁrst subspecialty clinic visit, and fewer clinic visits. Litt et al (4) found that higher selfesteem and autonomy (assuming more responsibility for health care in the context of a positive relationship with parents) were predictive of higher adherence to salicylate medications among a separate group of 38 adolescents with JRA. Chaney and Peterson (9) discovered that family relationship and stress variables predicted adherence to medications among 18 children and adolescents with JRA. A recent Finnish study found that self-reported adherence of adolescents with JRA was predicted by motivation, fear of acute problems, support from nurses, energy and willpower, and threat to social well being (10). These studies provide useful data on potential predictors of adherence to medication regimens for JRA, but they are limited by having assessed adherence using periodic drug assays (which measure adherence over short time periods) or self-report (which tends to overestimate adherence) (2). The most important advance in adherence assessment is the electronic monitor that detects actions necessary to administer medications, such as removing the cap from a pill bottle. These monitors store downloadable time- and date-stamped information in real time for up to several months. They can provide precise information on the frequency and timing of presumptive dosing and reveal daily patterns of adherence, such as full adherence (taking all daily doses), partial adherence (taking some but not all daily doses), and drug holidays (2 or more consecutive days without dosing). However, electronic monitors can only indicate that a maneuver consistent with dosing has been executed. It is assumed that the patient actually ingests the medication close to the time the monitored action is performed and that the monitored container is the only one in use by the patient (11). In spite of these potential limitations, electronic monitors have been proposed as the new “gold standard” in adherence measurement (12). To our knowledge, only 2 studies on adherence to NSAIDs have used electronic monitors. One involved patients with ankylosing spondylitis (13), and the other targeted a subset of patients in the current study who participated in an adherence intervention trial (14). The latter involved a 28-day preintervention baseline period, which is the focus of the current study. In addition to the use of electronic monitoring, the current study is unique in its focus on newly diagnosed patients with JRA. Rapoff et al The aims of this study were to describe patterns of adherence to NSAIDs (full, partial, or no adherence and drug holidays) among newly diagnosed patients with JRA during a 28-day period using an electronic monitoring device, and to examine demographic and disease-related variables as potential predictors of adherence. PATIENTS AND METHODS Patients. Patients included children and adolescents who were diagnosed with JRA within the past year according to established standards (15) and had been on an unvarying NSAID regimen for at least 1 month. Recruitment occurred over a 42-month period. The parents of 90 patients were sent recruitment letters. Thirty-six (40%) declined to participate because of time constraints. The remaining 54 agreed to participate, but 6 had incomplete adherence data (3 because of device malfunctions, 1 from failure of the patient to bring in the device for downloading, and 2 from dropping out during baseline). The ﬁnal sample consisted of 48 participants. The age range of the participants was 2.3–16.7 years (mean ⫾ SD 8.6 ⫾ 4.0), 71% were female, and 90% were white. Measures. Adherence. Adherence was assessed using the Medication Event Monitoring System (MEMS; Aprex Corporation, Fremont, CA). This electronic medication bottle cap records the date and time of each bottle opening. It can record 1,800 openings and has an 18-month battery life. MEMS data are downloaded to a portable computer and analyzed using the manufacturer’s software, which provides daily and continuous data on adherence. Patients were asked to use the MEMS for their primary NSAID, which we targeted because the regimen for most patients (75%) involved only a single NSAID, and over 80% of those were instructed to take it twice daily. Patients and their parents were informed about the purpose of the MEMS cap. The daily MEMS adherence score was the percent of prescribed doses taken within the recommended dosing interval, with a ⫾2-hour forgiveness interval (e.g., NSAID twice daily, could be taken 10 to 14 hours between doses). Extra bottle-cap openings were not considered in calculating adherence, because they were infrequent and reportedly resulted from ﬁlling the pill vial. Each participant yielded 28 contiguous days of data. Demographics. The parents’ demographic questionnaire yielded the child’s age (dates converted to decimal years), sex (male/female), ethnicity (white/nonwhite), mother’s marital status (married/not married), mother’s and father’s education levels (high school/some college/ college graduate/graduate school), and the parents’ occupations. The Hollingshead Four Factor Index of Social Status (16) was computed as the measure of socioeconomic status (SES), including the whole score (continuously variable from 8 – 66) and the 5 social strata (ranging from unskilled to major professional). Disease-related variables. The participant’s JRA subtype (pauciarticular, polyarticular, or systemic onset) and regimen complexity (1 NSAID alone versus at least 2 med- Adherence to Medications By Patients With JRA 907 ications) were recorded from the medical record. During routine clinical examination the rheumatologist recorded standard clinical indices of disease activity, including number of active joints (those with pain, swelling, and limitation of motion) and the number of minutes of morning stiffness. Procedure. Patients were given the MEMS cap to ﬁt over the vial used to dispense their primary NSAID. They were asked to dispense medications only from this vial and not to use any other vials or pill reminder containers to dispense their primary NSAID. Demographic data were obtained at the beginning of the 28-day MEMS monitoring period; disease activity measures were obtained from the medical record of the participant’s last clinic visit prior to the ﬁrst day of MEMS use (median delay ⫽ 5.5 weeks, mean ⫾ SD 7.5 ⫾ 6.7 weeks). Informed permission/assent forms were approved by the Institutional Review Board. Statistical analysis. For each participant, each day was deﬁned as full adherence (all prescribed doses taken on time ⫾ 2-hour forgiveness interval), partial adherence (some but not all taken on time), and no adherence (no doses taken at all). A drug holiday was deﬁned as 2 or more consecutive days with no doses taken preceded and followed by at least 1 day with at least partial adherence. The means, medians, and dispersions of the number of days ﬁtting these classiﬁcations were calculated and plotted to provide descriptive data on the patterns of adherence. To examine demographic and disease-related variables as potential predictors of adherence, patients were labeled adherent if they demonstrated full adherence on at least 80% of the 28 monitored days. This 80% adherence cut point is now a convention, which originated in early studies on adherence to antihypertensive medications (2). The analysis aimed to develop a model for prediction of adherence. Potential demographic and disease-related predictors were screened using univariate tests of the group difference (adherent versus nonadherent) at a liberal ␣ ⫽ 0.2 for entry. Continuous variables were tested using t-test, or Mann-Whitney U test as appropriate; categorical variables were tested using Fisher’s exact test, chi-square test, or linear-by-linear chi-square test as appropriate. All variables that met the screening criterion were entered into a preliminary forward stepwise logistic regression model with the adherence group as the dependent variable. Using ␣ ⫽ 0.05 for entry, variables identiﬁed as signiﬁcant predictors were then entered into a ﬁnal logistic regression model. All statistics were computed using SPSS software, version 11.5 (SPSS, Chicago, IL). RESULTS Descriptive statistics and univariate analyses. Using median levels, patients showed full adherence on 70% of the monitored days, partial adherence on 14%, and no adherence on 7% of the monitored days (Figure 1). There was considerable variability across patients, with full adherence ranging from 0 to 100% of the days. Seventy-nine Figure 1. For each of 48 patients with newly diagnosed juvenile rheumatoid arthritis, each of the 28 monitored days was classiﬁed as having full, partial, or no adherence to their nonsteroidal antiinﬂammatory drug regimens. Ranges, quartiles, means, and medians (mdn) for the percentage of monitored days falling into each classiﬁcation are shown. max ⫽ maximum; min ⫽ minimum. percent of patients took no drug holidays, 13% took 1, and the remaining 8% of the patients took between 2 and 4. The 80% adherence cut point yielded 25 (52%) adherent and 23 (48%) nonadherent patients. Table 1 shows the results for the demographic and disease activity variables for the entire sample and for the 2 adherence groups. The initial screen for candidate predictors showed that the adherence groups did not differ (P ⬎ 0.20) on age, sex, mother’s marital status, father’s education, participant’s JRA subtype, duration of morning stiffness, or complexity of NSAID regimen. The groups did differ, however, on 4 variables: the adherent group had a higher concentration of white patients (P ⫽ 0.180), their mothers had more education (P ⫽ 0.052) and had a higher Hollingshead Index (SES) (P ⫽ 0.001). The adherent group also had a higher active joint count (P ⫽ 0.016). Multivariate model. Of the 48 patients, 4 did not have joint count recorded in their medical record on the visit prior to the electronic monitoring period. The remaining 44 had complete data sets, and these were included in the multivariate model. The preliminary stepwise logistic regression analysis admitted 2 of the 4 candidate predictors: the active joint count and SES. Entering joint count ﬁrst, 65.9% of the patients were correctly classiﬁed as adherent/ nonadherent (Cox and Snell R2 ⫽ 0.158, P ⫽ 0.006). Adding SES to the prediction equation, the ﬁnal logistic model correctly classiﬁed 70.5% of the patients (Cox and Snell R2 ⫽ 0.295, P ⫽ 0.0005). This model had 74% sensitivity, 67% speciﬁcity, 71% positive predictive value, and 70% negative predictive value. 908 Rapoff et al Table 1. Demographic and disease-related variables for 48 newly diagnosed children with juvenile rheumatoid arthritis (JRA) and for subsamples* Variable Age, mean ⫾ SD years Sex, % female Ethnicity, % white Mother’s marital status, % married Mother’s education, % High school Some college College graduate Graduate school Father’s education, % High school Some college College graduate Graduate school Hollingshead Index, no. ⫾ SD JRA subtype, % Pauciarticular Polyarticular Systemic onset Active joint count, no ⫾ SD AM stiffness, minutes ⫾ SD Regimen of 1 NSAID only, % Total sample (n ⴝ 48) Nonadherent (n ⴝ 23) Adherent (n ⴝ 25) 8.6 ⫾ 4.0 71 90 77 8.4 ⫾ 4.4 78 83 70 8.8 ⫾ 3.7 64 96 84 35 21 35 8 48 22 26 4 24 20 44 12 41 20 22 17 45 ⫾ 13 52 14 19 14 39 ⫾ 13 32 24 24 20 51 ⫾ 9 33 52 15 1.3 ⫾ 2.6 43 ⫾ 47 75 35 57 9 0.4 ⫾ 0.9 40 ⫾ 50 83 32 48 20 2.2 ⫾ 3.2 46 ⫾ 44 68 Statistical test P† t-test 2 Exact 2 L-b-L 2 0.717 0.278 0.180‡ 0.235 0.052‡ L-b-L 2 0.282 t-test 2 0.001‡ 0.537 Mann-Whitney U t-test 2 0.016‡ 0.701 0.243 * Nonadherent ⫽ ⬍80% of nonsteroidal antiinﬂammatory drug (NSAID) doses taken on time; adherent ⫽ 80% of NSAID doses taken on time; L-b-L 2 ⫽ linear x linear 2 for ordered categories; exact ⫽ Fisher’s test (used because expected frequencies ⬍5). † Differences between the subsamples. ‡ Variables that were P ⱕ 0.20 and were then entered into the regression analysis. DISCUSSION The overall adherence rate in this study (25 [52%] of 48 patients) is consistent with previous studies in the adult and pediatric literatures, including a wide range of disease regimens and methods of measuring adherence (2). This adherence rate is based on classifying as adherent all patients who take ⱖ80% of their prescribed doses. This cutoff score is a convention in the adherence literature (2), which apparently had its origin in early studies on adherence to antihypertensive medications. It was reported that patients who took ⱖ80% of their medications had better blood pressure control than those who took ⬍80% (17). A recent study of children with human immunodeﬁciency virus (HIV) (18) also supports the 80% cutoff score. In this study children with detectable viral loads (a measure of active disease) were ⬍80% adherent to highly active antiretroviral therapy (18). Our descriptive information can also be compared with previous studies using electronic monitoring. In our study, full adherence occurred on a median of 70% of the monitored days, which is close to the 78% full adherence found in patients with ankylosing spondylitis (the only other study using MEMS to assess NSAID adherence) (13). In contrast, full adherence as assessed by electronic monitoring has been much lower in other pediatric studies. For example, full adherence to inhaled steroids or bronchodilator medications for children with asthma has ranged from 5 to 50% of monitored days (19,20). Days with no detectable adherence were quite low in our study, occur- ring on only 7% of the days, compared with 19% of the monitored days in the other NSAID study using the MEMS (13). In contrast, studies on children with asthma have reported a higher percentage of days with no adherence to inhaled medications (range 20 – 48%) (19,21,22). One might expect better adherence to NSAIDs than to inhaled medications for asthma because it may be easier to swallow a pill than to use a metered dose inhaler. Moreover, NSAID therapy may provide more immediate symptom relief than inhaled medications for asthma, particularly inhaled steroids, which are preventive and therefore do not provide immediate symptom relief. Another possible reason for more days with full adherence and fewer with no adherence in our study is that patients were newly diagnosed with JRA. Shorter disease duration has been associated with better adherence to chronic disease regimens (2). The multivariate model in our study showed that a higher active joint count prior to monitoring of adherence and higher SES were signiﬁcantly associated with being classiﬁed as adherent to NSAID. The active joint count ﬁnding contradicts pediatric studies that reported greater disease activity to be associated with lower adherence to medications for HIV (18), renal disease (23), and seizures (24). However, these studies were cross-sectional, and it is possible that lower adherence produced increased disease activity. In our study, active joint counts were obtained prior to the adherence-monitoring period. It appears that children with more active joints, including pain as a major Adherence to Medications By Patients With JRA symptom, were more likely to take their NSAID to relieve their symptoms. An applied behavior analytic perspective would suggest that taking NSAIDs would be negatively reinforced; that is, taking medications more consistently would provide relief or escape from aversive symptoms such as pain (2). The most robust predictor of adherence to NSAIDs in this study was SES, with patients from lower SES strata being more likely to be classiﬁed as nonadherent. This is in agreement with a number of pediatric studies showing that lower SES predicts lower adherence to regimens for asthma, cystic ﬁbrosis, diabetes, and renal disease (23,25– 27). Lower SES has also been found to be a signiﬁcant risk factor for poorer health, mental health, and cognitive/ academic outcomes for children (28,29). Most studies utilize income, education, and occupation to represent SES (28). We used the Hollingshead Index, which combines caregivers’ educational level and type of occupation. The connection between SES and adherence to medical regimens may be explained by a number of potential mechanisms. Lower education levels among lower SES caregivers may make it more difﬁcult for them to understand and accommodate prescribed regimens (30). Part of the difﬁculty in accommodating or integrating regimens into their daily lives may originate in lower SES families’ experiencing increased stress and reduced social support (28). Also, they may have fewer ﬁnancial resources, which may make it difﬁcult for them to ﬁll or renew prescriptions in a timely fashion. Given all of these potential factors, one might expect that lower SES families in particular would have a more difﬁcult time adhering to NSAIDs at times when their children are asymptomatic, as there are other stressors and demands that would take precedence. In our study we found that no low-SES patient was adherent when joint count was zero, whereas 61% of high-SES patients were adherent when joint count was zero (2 ⫽ 7.67, with Yates’ correction, P ⫽ 0.006); when joint count was ⬎0, the respective ﬁgures were 50% and 83% (2 ⫽ 0.444, P ⫽ 0.505). There are several limitations to this study. It was cross sectional and correlational, and it utilized a convenience sample of newly diagnosed children with JRA who were participating in a randomized trial to prevent an anticipated drop in adherence over time (14). The sample was rather small, and there were other potentially important predictors, such as self-efﬁcacy or barriers to adherence, that were not considered. In addition, we could only speculate about potential mechanisms that link SES with adherence, for we did not assess such factors as stressors in the family and ﬁnancial resources. In spite of these limitations, this is one of only 2 studies that monitored adherence to NSAIDs among children with JRA using electronic monitoring, which is considered the gold standard in adherence assessment (12). Unlike all other assessment methods, electronic monitors can capture highly speciﬁc daily medication events such as full, partial, and no dosing, as well as drug holidays. Future studies would beneﬁt from periodic assays to conﬁrm that medications removed from the MEMS container are ingested. Further research is also needed using biologic criteria for deﬁning optimal cutoff points for adherence/non- 909 adherence, the target being the level of adherence necessary to achieve an acceptable therapeutic response (31). Clearly, this would need to be done on a disease-bydisease, regimen-by-regimen basis. Longitudinal studies are also needed to predict adherence over time, with repeated measures of adherence, disease activity, and psychosocial predictors that might vary over time (such as family stressors, self-efﬁcacy, and barriers to adherence). The results of this study suggest that lower SES families deserve special attention to address nonadherence to NSAIDs in the treatment of JRA. Educational and behavioral interventions to improve adherence must address the unique challenges and lack of practical and psychological resources among these families. REFERENCES 1. Cassidy JT, Petty RE. Textbook of pediatric rheumatology. 5th ed. New York: Elsevier; 2005. 2. Rapoff MA. Adherence to pediatric medical regimens. New York: Kluwer/Plenum; 1999. 3. Litt IF, Cuskey WR. Compliance with salicylate therapy in adolescents with juvenile rheumatoid arthritis. Am J Dis Child 1981;135:434 – 6. 4. Litt IF, Cuskey WR, Rosenberg A. Role of self-esteem and autonomy in determining medication compliance among adolescents with juvenile rheumatoid arthritis. Pediatrics 1982; 69:15–7. 5. Rapoff MA, Lindsley CB, Christophersen ER. Improving compliance with medical regimens: case study with juvenile rheumatoid arthritis. Arch Phys Med Rehabil 1984;65:267–9. 6. Rapoff MA, Purviance MR, Lindsley CB. Improving medication compliance for juvenile rheumatoid arthritis and its effect on clinical outcome: a single-subject analysis. Arthritis Care Res 1988;1:12– 6. 7. Rapoff MA, Purviance MR, Lindsley CB. Educational and behavioral strategies for improving medication compliance in juvenile rheumatoid arthritis. Arch Phys Med Rehabil 1988; 69:439 – 41. 8. Rapoff MA. Compliance with treatment regimens for pediatric rheumatic diseases. Arthritis Care Res 1989;2:S40 –7. 9. Chaney JM, Peterson L. Family variables and disease management in juvenile rheumatoid arthritis. J Pediatr Psychol 1989; 14:389 – 403. 10. Kyngas H. Motivation as a crucial predictor of good compliance in adolescents with rheumatoid arthritis. Int J Nurs Pract 2002;8:336 – 41. 11. Urquhart J, de Klerk E. Contending paradigms for the interpretation of data on patient compliance with therapeutic drug regimens. Stat Med 1998;17:251– 67. 12. Cramer JA. Microelectronic systems for monitoring and enhancing patient compliance with medication regimens. Drugs 1995;49:321–7. 13. De Klerk E, van der Linden SJ. Compliance monitoring of NSAID drug therapy in ankylosing spondylitis, experiences with an electronic monitoring device. Br J Rheumatol 1996; 35:60 –5. 14. Rapoff MA, Belmont J, Lindsley C, Olson N, Morris J, Padur J. Prevention of nonadherence to nonsteroidal anti-inﬂammatory medications for newly diagnosed patients with juvenile rheumatoid arthritis. Health Psychol 2002;21:620 –3. 15. Brewer EJ Jr, Bass J, Baum J, Cassidy JT, Fink C, Jacobs J, et al. Current proposed revision of JRA criteria. Arthritis Rheum 1997;20 Suppl 2:195–9. 16. Hollingshead AB. Four factor index of social status. New Haven (CT): Yale University; 1975. 17. Sackett DL, Snow JC. The magnitude of compliance and noncompliance. In: Haynes RB, Taylor DW, Sackett DL, editors. Compliance in health care. Baltimore: The Johns Hopkins University Press; 1979. p. 11–22. 910 18. Farley J, Hines S, Musk A, Ferrus S, Tepper V. Assessment of adherence to antiviral therapy in HIV-infected children using the Medication Event Monitoring System, pharmacy reﬁll, provider assessment, caregiver self-report, and appointment keeping. J Acquir Immune Deﬁc Syndr 2003;33:211– 8. 19. Bender B, Milgrom H, Rand C, Ackerson L. Psychological factors associated with medication nonadherence in asthmatic children. J Asthma 1998;35:347–53. 20. Gibson NA, Ferguson AE, Aitchison TC, Paton JY. Compliance with inhaled asthma medication in preschool children. Thorax 1995;50:1274 –9. 21. Bender B, Wamboldt FS, O’Connor SL, Rand C, Szeﬂer S, Milgrom H, et al. Measurement of children’s asthma medication adherence by self report, mother report, canister weight, and Doser CT. Ann Allergy Asthma Immunol 2000;85:416 – 21. 22. Milgrom H, Bender B, Ackerson L, Bowry P, Smith B, Rand C. Noncompliance and treatment failure in children with asthma. J Allergy Clin Immunol 1996;98:1051–7. 23. Brownbridge G, Fielding DM. Psychosocial adjustment and adherence to dialysis treatment regimes. Pediatr Nephrol 1994;8:744 –9. 24. Hazzard A, Hutchinson SJ, Krawiecki N. Factors related to Rapoff et al 25. 26. 27. 28. 29. 30. 31. adherence to medication regimens in pediatric seizure patients. J Pediatr Psychol 1990;15:543–55. Bobrow ES, AvRuskin TW, Siller J. Mother-daughter interaction and adherence to diabetes regimens. Diabetes Care 1985; 8:146 –51. Patterson JM. Critical factors affecting family compliance with home treatment for children with cystic ﬁbrosis. Fam Relations 1985;34:79 – 89. Radius SM, Becker MH, Rosenstock IM, Drachman RH, Schuberth KC, Teets KC. Factors inﬂuencing mothers’ compliance with a medication regimen for asthmatic children. J Asthma Res 1978;15:133– 49. Bradley RH, Corwyn RF. Socioeconomic status and child development. Annu Rev Psychol 2002;53:371–99. Chen E, Matthews KA, Boyce WT. Socioeconomic differences in children’s health: how and why do these relationships change with age? Psychol Bull 2002;128:295–329. Goldman DP, Smith JP. Can patient self-management help explain the SES health gradient? Proc Natl Acad Sci U S A 2002;99:10929 –34. Gordis L. Conceptual and methodologic problems in measuring patient compliance. In: Haynes RB, Taylor DW, Sackett DL, editors. Compliance in health care. Baltimore: The Johns Hopkins University Press; 1979. p. 23– 45.