TELEMEDICINE JOURNAL AND e-HEALTHVolume 10, Supplement 2, 2004© Mary Ann Liebert, Inc.

Asthma Control and Hospitalizations Among Inner-CityChildren: Results of a Randomized Trial

SYLVIA GUENDELMAN, Ph.D.,1 KELLEY MEADE, M.D.,2 YING QING CHEN, Ph.D.,3and MINDY BENSON, P.N.P.2

ABSTRACT

Asthma prevalence is increasing among poor and minority children. We examined the effec-tiveness of a novel interactive device programmed for self-management of pediatric asthmain reducing asthma control problems and hospitalizations. A randomized controlled trial (66children in the intervention group and 68 in the control group) was conducted at home andin an outpatient hospital clinic with 8–16-year-old inner-city children with physician-diag-nosed asthma. During a 12-week period, children in the experimental group received anasthma self-management and education program, the Health Buddy® (Health Hero Network),designed to enable them to monitor their symptoms and transmit this information to a casemanager through a secure website. Control group participants used an asthma diary. Afteradjusting for baseline asthma control problems, asthma severity, and seasonality, childrenrandomized to automated self-management had a significantly lower mean number of asthmacontrol problems at 6 weeks (2.0, SD � 1.6) as compared to children assigned to the asthmadiary (2.7, SD � 1.6) (p � 0.03). By 12 weeks, after adjusting for time and the other covariates,asthma control problems dropped markedly in both groups, and did not differ by interven-tion modality (p � .07). The intervention modality was not a significant predictor of hospi-talization. Educational interventions that encourage children’s active involvement in theirown care and symptom monitoring would help children increase their control of asthma prob-lems. Compared to the asthma diary, the automated self-management had a significant short-term impact on asthma control problems. Its initial effectiveness and more consistent use sug-gest that remote monitoring may be successfully used in short term interventions and insettings where staffing for case management is weak.

1Maternal and Child Health Program, Division of Health Policy and Management; and 3Division of Biostatistics,School of Public Health, University of California, Berkeley, Berkeley, California.

2Ambulatory Services, Children’s Hospital Oakland, Oakland, California.

INTRODUCTION

SEVERAL STUDIES HAVE DOCUMENTED an in-crease in childhood asthma prevalence in

the United States in recent years.1–5 The largestincrease in asthma morbidity seems to have oc-

curred among poor and minority children liv-ing in the inner city.6,7 Prevalence rates amongpoor and minority children have been reportedfrom 10% to 20%,8,9 whereas the prevalence forall U.S. children is 6%.2 Among these children,the greater use of emergency departments and

S-6

5292_e01_pS6-S14 11/9/04 1:00 PM Page 6

the greater number of hospitalizations suggestthat asthma is not under control.10

Asthma morbidity in children is largely pre-ventable with improved patient/family educa-tion regarding risk factors and compliance withprescribed therapy,2 and with medical man-agement11 that preferably considers parents’perceptions and health beliefs.12 NationalAsthma Education and Prevention Program(NAEPP) guidelines for clinicians recommendthe use of written asthma plans, dust mite al-lergen and cockroach control, smoking cessa-tion, use of inhaled corticosteroids, and peakflow monitoring to keep asthma symptoms un-der control and to avoid preventable hospital-izations.13,14 In addition, NAEPP guidelinesrecommend that patients monitor their asthmawith a daily diary or a periodic self-assessmentsheet before a visit to the physician in order tocapture the patient’s impression of whether hisor her asthma is controlled and to assess self-management skills.14

Monitoring approaches, although useful, de-mand intensive physician involvement, and theyhave been hard to implement in clinical set-tings.15 Many providers have turned to pre-scribing daily medications to reduce symptoms.Yet, this approach poses other challenges. Sev-eral studies have documented poor adherence tomedication regimes, particularly among inner-city children.16–18 Some studies suggest that thenon-compliance in medication use is due to pooraccess to quality medical care.7

Interactive, web-based health communica-tion devices can engage patients in their owncare, while reducing and reinforcing the role ofthe physician as educator and counselor in atimely manner. Through providing informa-tion, immediate feedback, and decision sup-port, web-based devices may reinforce writtenasthma plans while they also have the poten-tial to change behaviors and reduce the burdenof illness. Furthermore, they may change be-liefs about the patient’s effectiveness in con-trolling asthma.19

We evaluated the efficacy of a new home-based interactive device, the Health Buddy®

(Health Hero Network [HHN], MountainView, CA) programmed for the care of inner-city children with asthma. It monitors asthmasymptoms, and sends information through a

secure website to the health care provider. Weexamined whether the device can help patientscontrol their asthma and subsequently reducehospital admissions. The level of asthma con-trol is an indicator of the adequacy of healthcare provided, taking into account both sever-ity of asthma and the patient’s adherence tomedical management.20,21 We hypothesizedthat by providing children an opportunity tolearn about asthma, including symptom recog-nition and feedback on their appraisals and be-haviors, asthma problems would be broughtunder control.22 When this occurs, the need forhospitalization would also diminish.

MATERIALS AND METHODS

Sample and study protocol

A randomized, controlled trial of a comput-erized interactive asthma self-management andeducation program was conducted in the pri-mary care clinic at Children’s Hospital Oak-land. This clinic is a comprehensive pediatrichealth center and resident teaching facility thatserves a predominantly low-income popula-tion. Patient appointment lists were reviewedthe day before clinic, and all children withasthma as the reason for the visit were screenedfor study eligibility. Potential recruits were alsoreferred by the attending physicians in the pul-monary and adolescent clinics and identifiedthrough hospital administrative lists if theyhad substantial health care utilization (two ormore emergency room visits and/or at leastone inpatient admission due to asthma duringthe year preceding the study).

Children would be eligible to participate inthe study if they were between 8 and 16 yearsof age, had an English-speaking caregiver, hada telephone at home, and were diagnosed withpersistent asthma following NAEPP clinicalpractice guidelines.14 If patients were involvedin other asthma or drug efficacy studies, in re-search that required behavior modification,had co-morbid conditions that affect their qual-ity of life, or who had mental or physical chal-lenges, they were excluded from the study. Institutional Review Board approval was ob-tained from the hospital prior the study.

ASTHMA CONTROL AND HOSPITALIZATIONS S-7

5292_e01_pS6-S14 11/9/04 1:00 PM Page 7

Children diagnosed with persistent or out ofcontrol asthma identified as eligible were ap-proached for participation in the study be-tween April 1999 and July 2000 (n � 136). Fam-ilies were told that the purpose of the studywas to find out if keeping track of asthmasymptoms at home would help the families andproviders to manage their children’s asthmaand that two methods of keeping track werebeing evaluated. The nurse coordinator ob-tained written informed consent on 134 chil-dren from their parents or legal guardians.Only two families declined to participate be-cause of time constraints.

After consent was given, the nurse conducteda standardized teaching session at the clinic, ad-ministered the baseline questionnaire, and ran-domized the children into the control or the in-tervention groups. At the teaching session, eachchild was given a peak flow-measuring deviceand instructed on how to establish their per-sonal best based on the best of three readings,how to determine whether the reading was inthe “green-yellow or red zone.” Readings in thegreen zone (80–100% of personal best) indicatedthat their asthma was under control. A peakflow reading (PFR) in the yellow zone (50–80%of personal best) indicated that asthma was notunder sufficient control and required additionalmedication. Readings in the red zone (�50% ofpersonal best) signaled severe asthma exacer-bation. The teaching session also covered theappropriate use of medications and of healthcare services and reviewed the child’s (unmod-ified) written asthma plan. Subsequently, thenurse coordinator administered the baselinequestionnaire to the child and the accompany-ing family member, and gave each family a $20incentive for completing the interview.

Following the interview, the nurse openeda sealed envelope containing the treatmentassignment, and children were randomized toeither the intervention (n � 66) or the controlgroup (n � 68). They were given instructionson how to use the assigned tracking methodto record their PFR and symptomatology.Sample size calculations indicated that wehad 97% power to detect differences in oneasthma control problem between the inter-vention and control groups at the 5% level ofsignificance.

Intervention

Health Buddy, the personal and interactivecommunication device connected to a homephone, was developed by HHN. The device canbe programmed to present questions and in-formation on a screen and to record responses.The nurse coordinator sends a set of querieseach day using a standard Internet browser.The patient answers the queries, called dia-logues, by pressing one of four buttons on thedevice. Subsequently, the device would auto-matically call a data-processing center at night,which processes the responses and publishesthem to a secure website on the next day.

The study team developed a protocol withsoftware programmers and asthma specialistsat HHN. Following the NAEPP clinical prac-tice guidelines,14 the protocol consisted of 10questions about asthma symptoms, PFRs, func-tional status, such as school attendance and ac-tivity limitations, and use of medications andof health services. In addition to the core im-mutable questions, asthma facts and triviaquestions, which changed daily, were includedto provoke children’s curiosity and enhancelearning. The device provided immediate feed-back for the questions, praising a correct an-swer or encouraging the user to try again. Thefollowing excerpt illustrates the dialogue be-tween the child and the device:

Hi, [Child’s name]! Thanks for hanging outwith your Health Buddy today. Your questionsare now ready for you.

Do you know how much a McDonald’s ham-burger cost in 1963? (Trivia)

Have you had any coughing or wheezing inthe last day?

[If child answers Yes:] This could be a signthat your asthma is acting up. You may needto take your Albuterol as directed by your doc-tor when you are coughing and wheezing.

[If No:] That’s great! You must be taking yourpreventor medicine.

Have you been coughing during the night sothat it wakes you up?

[If child answers Yes:] You should take yourAlbuterol as directed by your doctor just be-fore going to bed to prevent nighttime cough-ing and wheezing.

GUENDELMAN ET AL.S-8

5292_e01_pS6-S14 11/9/04 1:00 PM Page 8

[If No:] That is super good news! A goodnight’s rest is important for staying healthy.

Did you take your preventor medicine to-day?

[If child answers Yes:] That’s great you tookyour preventor medicine! You are takingcharge of your body and health.

[If No:] Go ahead and take it now as in-structed by your doctor. If you are not surewhich of your medicines is the preventor med-icine, call your doctor.

[If Don’t Know:] Your preventor medicineworks to prevent your asthma from acting up.If you are not sure which of your medicines isthe preventor medicine, call your doctor.23,*

Children were instructed to use the deviceonce a day, preferably at a regular time. Anasthma diary was used as the comparison ap-proach for tracking symptoms. Children wereencouraged to log their symptoms and to mon-itor peak flow, medication use, and restrictedactivity on a daily basis.

Families whose children were assigned to theintervention group were given a demonstrationon how to use the device and how to install itat home. The nurse coordinator only calledthose families who initially failed to transmittheir responses to her website, to review in-stallation instructions (n � 2). All childrenwere asked to return for two follow-up visits,at 6 weeks and 12 weeks. At each follow-upvisit, families were interviewed by the nursecoordinator and given a standardized teachingsession that reinforced peak flow measure-ment, compliance with medicines, tracking ofsymptoms, and review of written action plan.Families received an incentive for participa-tion, and children were asked to hand in theirhealth diaries. They were also examined by aphysician. Medical management remained atthe discretion of the physicians, and it followedNAEPP guidelines. Physicians were notblinded to the child’s group assignment. Treat-ment regimes were kept constant between vis-

its and changed at the follow-up visits only ifthere was a significant deterioration or im-provement in asthma symptoms.

Measures

Asthma control was assessed as the numberof self-reported asthma control problems.20 Re-spondents were queried on whether in the last14 days they had: (i) any symptoms of chesttightness, coughing, shortness of breath, orwheezing; (ii) trouble sleeping at night as a re-sult of cough or wheeze; (iii) restrictions inphysical activity such as exercise or play dueto asthma; and (iv) PFRs in the yellow-redzone. Finally, children were asked about (v)any missed school days because of asthma inthe last 6 weeks. Following work by Vollmer etal.,20 each of the five asthma control indicatorswas given a score of 0 if the respondent did nothave the problem and 1 if she or he did. Thenumber of current control problems wassummed as an index ranging from 0 to 5.20

Hospitalization was measured by whether ornot the child had one or more hospital admis-sions for asthma during the 12-week trial.10,24

This information was abstracted from hospitalrecords. Missing data and outside hospitaliza-tions were checked against self-reported data.5

Asthma control problems, seasonality ofsymptoms (i.e., whether they occurred pre-dominantly in the winter/fall or in thespring/summer), information on hospital ad-missions in the last 6 weeks, and the demo-graphic characteristics was obtained from in-terviews. The nurse coordinator administeredthe questionnaires at baseline and at the twofollow-up visits at 6 and 12 weeks, respectively.

Data analysis

Data analysis was based on intention-to-treat. Chi-square tests for categorical variablesand two-sample t tests for continuous vari-ables were used to compare the two groups inthe study with respect to demographic char-acteristics (at baseline) and to asthma out-comes at baseline and at the two follow-up vis-its. Generalized linear models25 were used tomodel the covariate effects on the number ofasthma control problems and the probabilityof having a hospitalization. The results with a

ASTHMA CONTROL AND HOSPITALIZATIONS S-9

*From Guendelman S, Meade K, Benson M, Chen YQ,Samuels S. Improving asthma outcomes and self-man-agement behaviors of inner-city children. Arch Ped Ado-lesc Med 2002;114–120, and reprinted with permissionfrom the American Medical Association, Copyright © 2002. All rights reserved.

5292_e01_pS6-S14 11/9/04 1:00 PM Page 9

p value �0.05 were justified as significant.When repeated observations of the two follow-up visits were considered, the technique ofGeneralized Estimation Equations26 was em-ployed to estimate the covariate effects and toaccount for within-subject correlation. The co-variates included the baseline response (i.e.,hospitalizations, asthma control problems),the asthma severity score, seasonality of thevisit, and treatment allocation. The treatmenteffect is presented as the intervention Odds Ra-tio (OR), which is the ratio of odds of an out-come in the intervention group to the odds ofthe same outcome in the asthma diary group,adjusted for the covariates. Interaction termsbetween treatment and the covariates wereevaluated for both outcomes. Missing data(which were rare) were handled by assumingmissing at random.27

RESULTS

The intervention and asthma control groupshad comparable demographic characteristics,illness history, daily use of quick-relief medica-tions, and in-patient hospital use at baseline as

expected from random assignment (Table 1).Follow-up of participants occurred similarlyacross all four seasons in the two groups.Ninety-six percent of participants (63 interven-tion and 65 control) returned for the first follow-up visit at 6 weeks, and 91% (62 interventionand 60 control) returned for the follow-up visitat 12 weeks. Reasons for dropping out from thestudy included moving out of the area (n � 3)or life crises experiences (n � 4). Additionally,five families were unavailable for contact. Base-line characteristics of children who did and didnot complete the trial did not differ.

At baseline, coughing or wheezing was themost frequent asthma control problem (88%),followed by activity limitations due to asthma(75%) and nocturnal awakenings (63%) (Fig. 1).Ninety-two percent of the children had at leastone asthma control problem, and 14% had allfive control problems (data not shown). Al-though almost half of the children did not usea peak flow device at baseline (30/66 for inter-vention and 34/68 for control, p � 0.49), at the6-week follow-up visit, 89% of the participantsreported that they used it (60/66 for interven-tion and 54/68 for control, p � 0.03). By 12weeks 91% reported that they used it (62/66 for

GUENDELMAN ET AL.S-10

TABLE 1. BASELINE CHARACTERISTICS OF CHILDREN ENROLLED IN THE STUDY

Health Buddy Asthma diary Chi-square(n � 66) (n � 68) p value

Age (years)a 12 (2.3) 12.2 (2.9) 0.65b

Male 40 (60.6) 37 (54.4) 0.47Race/ethnicity

African American 52 (78.8) 50 (73.5) 0.69White 5 (7.6) 8 (11.8)Other 9 (13.6) 10 (14.7)

Type of health insurance coveragePublic 61 (92.4) 63 (92.6) 0.98Private 5 (7.6) 4 (5.9)

Parents is primary asthma caregiver 47 (71.2) 55 (80.9) 0.35Education of asthma caregiver

High school diploma or below 26 (39.3) 35 (51.4) 0.50At least some college/technical 40 (60.6) 33 (48.6)

Asthma severityMild persistent 15 (23.1) 20 (29.4) 0.66Moderate persistent 43 (66.1) 40 (58.8)Severe persistent 7 (10.8) 8 (11.8)

Daily puffs of quick-relief medicationa 1.6 (0.7) 1.5 (0.7) 0.55b

Patients hospitalized during 6 weeks 7 (10.6) 7 (10.3) 0.95prior to trial

Data are number (percent) except as indicated.aData are mean (SD).bp value based on t test.

5292_e01_pS6-S14 11/9/04 1:00 PM Page 10

intervention and 60/68 for control, p � 0.25)(data not shown). By 6 and 12 weeks, eachasthma control problem showed a decreasingtrend except for the prevalence of activity lim-itations, which remained similar at 6 (37%) and12 (39%) weeks (Fig. 1).

Children in both groups reported a decreasein the mean number of asthma control prob-lems at 6 and 12 weeks compared to baseline(Table 2). By 6 weeks, after adjusting for base-line asthma control problems, asthma severity,and seasonality, children randomized to the in-tervention group had a significantly lowermean number of asthma control problems (2.0,SD � 1.6) than children assigned to the asthmadiary group (2.7, SD � 1.6) (p � 0.03). How-ever, by 12 weeks, after adjusting for time andthe other covariates, the mean number ofasthma control problems was not significantlydifferent in the two groups (p � 0.07). Further-more, the drop in asthma control problems

from the 6 to 12 weeks in the asthma diarygroup was not significant. No interactions be-tween the treatment modality and the covari-ates were found.

The proportion of children admitted to thehospital for asthma increased significantly asthe number of asthma control problems in-creased. Whereas children who had no asthmacontrol problems did not have a hospital ad-mission for asthma during the trial period, 12%of those with five asthma control problemswere admitted (Fig. 2). The difference in theproportion of children who were hospitalizedduring the trial between the two groups wasnot statistically significant after controlling forother covariates (OR � 0.62; 95% confidence in-terval � 0.17, 2.32) (Table 3).

On average over the 12-week trial period,77% of children used their tracking system 3days a week or more (i.e., every other weekday) (89.4% for intervention vs. 65.1% for con-

ASTHMA CONTROL AND HOSPITALIZATIONS S-11

Coughing/Wheezing0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

NocturnalAwakening

PFR in Yellow/RedZone

Type of Asthma Control Problem

Per

cent

age

Limitation inActivities

Missed School

Visit 1Visit 2Visit 3

FIG. 1. Frequency of individual asthma control problems during a 90-day trial for all children. PFR � peak flowreading.

TABLE 2. MEAN NUMBER OF ASTHMA CONTROL PROBLEMS BY TREATMENT MODALITY DURING THE 12-WEEK TRIAL

Treatmenteffect

Mean (SD) p value Mean (SD) p valuea Mean (SD) p value p valueb

Health Buddy 3.5 (1.4) 0.87 2.0 (1.6) 0.03 1.8 (1.5) 0.34 0.07Asthma diary 3.5 (1.5) 2.7 (1.6) 2.1 (1.4)

aAdjusted for baseline asthma control problems, asthma severity, and seasonality.bAdjusted for baseline asthma control problems, asthma severity, and seasonality, and visits.

Baseline 6 Weeks 12 Weeks

5292_e01_pS6-S14 11/9/04 1:00 PM Page 11

trol, p � 0.001). Decline in compliance wasfaster in the early stages of the interventionfor the asthma diary group. At 6 weeks par-ents reported that 70% of the children in theintervention group used the device with no orfew reminders compared to 49% of childrenusing the asthma diary without reminders(p � 0.08). By 12 weeks 69% of children in theintervention group compared to 41% in theasthma diary group used it with few or no re-minders (p � 0.001). Children who used theirtracking systems with no or few reminderswere less likely to have asthma control prob-lems (p � 0.001).

DISCUSSION

The results of this study show that by 6weeks into the trial children randomized to theintervention group had significantly fewerasthma control problems than children as-signed to the asthma diary, after controlling for

other covariates. These differences may havebeen accounted for by children’s interest in thenovelty of the device, as reflected in the highcompliance rate in its use and their higher com-pliance with monitoring peak flows at 6 weeks.These children received feedback from the de-vice when entries were completed or whenthey were missing, and this could have pro-moted compliance. After 12 weeks the effectsin both groups were weaker and not statisti-cally significant. This could be due to satura-tion with the educational messages. Alterna-tively, children successfully recruited to thestudy could have been motivated to controltheir asthma independent of any intervention.Hence, they would improve over time

Since both study groups received significantinterventions, this could have limited our abil-ity to demonstrate an incremental impact ofelectronic self-monitoring. Although it is diffi-cult to isolate the independent effects of the ed-ucational intervention, the decline in asthmacontrol problems could have resulted from theasthma action plans alone.24

Control of asthma requires access to qualitymedical care. In this single-center clinical trial,access to quality care, although not assessed ex-tensively, was available to the study partici-pants. All had health insurance coverage and awritten asthma action plan that specifiedasthma triggers and medications, and all re-ceived follow-up care in a highly equippedasthma clinic. Similar to the National Cooper-ative Inner-City Asthma Study (NCICAS), thisstudy recruited a large proportion of moder-ate-to-severe asthmatics, and the number ofbaseline hospitalizations was comparable. Al-though the study participants were older(mean age 12 years vs. 6 years for NCICAS) the

GUENDELMAN ET AL.S-12

0 1–2 3–4 50.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

Number of Asthma Control Problems

Hos

pita

lized

FIG. 2. Percentage of children with at least one hospitaladmission during the trial period by number of asthmacontrol problems.

TABLE 3. PROPORTION OF CHILDREN WITH ANY HOSPITALIZATION BY TREATMENT MODALITY DURING THE 12-WEEK TRIAL

Treatment effectb

n % n % n % OR (95% CI)

Health Buddy 7/66 (10.6) 3/63 (4.8) 1/62 (1.6) 0.62 (0.17, 2.32)Asthma diary 7/68 (10.3) 2/65 (3.1) 4/60 (6.6)

aHospitalizations during 6 weeks prior to the trial.bOR is the ratio of odds of having had a hospital admission for asthma among the Health Buddy group to the odds

of having had a hospital admission for asthma in the asthma diary group, adjusted for asthma severity, baseline hos-pitalizations, seasonality, and visits. CI � confidence interval.

Baselinea 6 weeks 12 weeks

5292_e01_pS6-S14 11/9/04 1:00 PM Page 12

samples were also comparable with respect tosex, race, a family history of asthma, and ex-posure to second-hand smoking.7 These char-acteristics suggest that our findings are quitegeneralizable to inner-city children who havehealth insurance, are attached to a health caresystem, and who have an English-speakingcaregiver.

The results must be interpreted with cautiongiven the study limitations. Despite the at-tempts by the nurse coordinator to check the re-liability of the self-reported data on asthma con-trol at each interview and verify each child’sresponse with the participating parent or care-giver, there could have been recall bias due tothe retrospective nature of the interviews. Nev-ertheless, we found a strong association be-tween the self-reported number of asthma con-trol problems and hospitalizations abstractedfrom hospital records. Children in the asthmadiary group who filled out diaries retrospec-tively may have overstated compliance. Despiteestimates of high reported compliance withhealth diaries,28 a recent study using a paper di-ary electronically instrumented with photosen-sors to capture the opening and closing of thediary reported that actual compliance with pa-per diaries is very low and not filled accordingto protocol.29 Therefore, we relied on family re-call to measure asthma control.

The initial effectiveness of the remote moni-toring device in reducing symptoms and itsmore consistent use suggest that it may lend it-self to targeted brief interventions for address-ing asthma control problems.

ACKNOWLEDGMENTS

We would like to thank Lora Santiago for herclerical support and Dr. Doug Oman for hishelpful comments. This study was partiallyfunded by Health Management Services,Merck and Co., Inc. through an unrestricted ed-ucational grant.

REFERENCES

1. Centers for Disease Control and Prevention. Asthmamortality and hospitalizations among children and

young adults. MMWR Morbid Mortal Wkly Rep 1996;45:350–353.

2. Centers for Disease Control and Prevention. CDC sur-veillance summaries. Surveillance for asthma—United States, 1960–1995. MMWR Morbid Mortal WklyRep 1998;47(SS-1):1–27.

3. Weiss KB, Gergen PJ, Hodgson TA. An economicevaluation of asthma in the United States. N Engl JMed 1992;326:862–866.

4. Russo M, McConnochie K, McBride J, Szilagyi PG,Brooks AM, Roghmann RJ. Increase in admissionthreshold explains stable asthma hospitalizationrates. Pediatrics 1999;104:454–462.

5. Vollmer W, Osborne M, Buist AS. Uses and limita-tions of mortality and health care utilization statisticsin asthma research. Am J Respir Crit Care Med 1994;149:S79–S87.

6. Gergen PJ, Weiss KB. Changing patterns of asthmahospitalization among children: 1979 to 1987. JAMA1990;264:1688–1692.

7. Kattan M, Mitchell H, Eggleston P, Gergen P, CrainE, Redline S, Weiss R, Evans R 3rd, Kaslow R, Kercs-mar C, Leickly F, Malveaux F, Wedner HJ. Charac-teristics of inner-city children with asthma: the Na-tional Cooperative Inner-City Asthma Study. PediatrPulmonol 1997;24:253–262.

8. Crain EF, Weiss KB, Bijur PE, Hersh M, Westbrook L,Stein RE. An estimate of the prevalence of asthma andwheezing among inner-city children. Pediatrics 1994;94:356–362.

9. Halfon N, Newacheck PW. Childhood asthma andpoverty: differential impacts and utilization of healthservices. Pediatrics 1993;91:56–61.

10. Lozano P, Sullivan SD, Smith DH, Weiss KB. The eco-nomic burden of asthma in US children: estimatesfrom the National Medical Expenditure Survey. J Al-lergy Clin Immunol 1999;104:957–963.

11. Murphy SJ, Kelly HW. Advances in the managementof acute asthma in children. Pediatr Rev 1996;17:227–235.

12. Mansour ME, Lanphear BP, De Witt TG. Barriers toasthma care in urban children: parent perspectives.Pediatrics 2000;106:512–519.

13. National Asthma Education Program. Guidelines forthe diagnosis and management of asthma. Bethesda, MD:US Department of Health and Human Services, 1991.

14. National Asthma Education and Prevention Program.Guidelines for the Diagnosis and management of asthma.Report 2. Bethesda, MD: U.S. Department of Healthand Human Services, 1997.

15. Cabana MD, Rand CS, Becher OJ, Rubin HR. Reasonsfor pediatrician nonadherence to asthma guidelines.Arch Pediatr Adolesc Med 2001;155:1507–1062.

16. Halterman JS, Aligne CA, Auinger P, McBride JT, Szi-lagyi PG. Inadequate therapy for asthma among chil-dren in the United States. Pediatrics 2000;105(1 Pt 3):272–276.

17. Ordonez GA, Phelan PD, Olinsky A, Robertson CF.Preventable factors in hospital admissions for asthma.Arch Dis Child 1998;78:143–147.

ASTHMA CONTROL AND HOSPITALIZATIONS S-13

5292_e01_pS6-S14 11/9/04 1:00 PM Page 13

18. Homer CJ, Szilagyi P, Rodewald L. Does quality ofcare affect rates of hospitalization for childhoodasthma? Pediatrics 1996;98:18–23.

19. Zimmerman B, Bonner S, Evans D, Mellins RB. Self-regulating childhood asthma: A developmental modelof family change. Health Educ Behav 1999;26:55–71.

20. Vollmer WM, Markson LE, O’Connor E, Sanocki LL,Fitterman L, Berger M, Buist AS. Association of asthmacontrol with health care utilization and quality of life.Am J Respir Crit Care Med 1999;160:1647–1652.

21. Cockcroft DW, Swystun A. Asthma control vs. asthmaseverity. J Allergy Clin Immunol 1996;98:1016–1018.

22. Bandura A. Social learning theory. Englewood Cliffs,NJ: Prentice Hall, 1977.

23. Guendelman S, Meade K, Benson M, Chen YQ,Samuels S. Improving asthma outcomes and self-management behaviors of inner-city children. ArchPed Adolesc Med 2002;156:114–120.

24. Lieu TA, Quesenberry CP Jr, Capra AM, Sorel ME,Martin KE, Mendoza GR. Outpatient managementpractices associated with reduced risk of pediatricasthma hospitalization and emergency departmentvisits. Pediatrics 1997;100:334–341.

25. McCullagh P, Nelders NA. Generalized linear models,2nd ed. New York: Chapman, 1989.

26. Zeger SL, Liang KY. Models for longitudinal data: ageneralized estimating equation approach. Biometrics1988;44:1049–1106.

27. Rubin DB. Inference and missing data. Biometrika1976;63:581–592.

28. Norman GR, McFarlane AH, Streiner DL, Neale K.Health diaries: strategies for compliance and relationto other measures. Med Care 1982;20:623–629.

29. Stone AA, Shiffman S, Schwartz JE, Broderick JE, Huf-ford MR. Patient non-compliance with paper diaries.Br M J 2002;324:1193–1194.

Address reprint requests to:Sylvia Guendelman, Ph.D.

Maternal and Child Health ProgramDivision of Health Policy and Management

School of Public Health404 Earl Warren Hall, #7360

University of CaliforniaBerkeley, CA 94720-7360

E-mail: sylviag@uclink4.berkeley.edu

GUENDELMAN ET AL.S-14

5292_e01_pS6-S14 11/9/04 1:00 PM Page 14