School-supervised use of a once-daily inhaled corticosteroid regimen: A cluster randomized trial

School-supervised use of a once-daily inhaled corticosteroid regimen: A cluster randomized trial Joe K. Gerald, MD, PhD,a,b Julia M. Fisher, PhD, MS,c Mark A. Brown, MD,d,e Conrad J. Clemens, MD, MPH,f Melissa A. Moore, MD,f Scott C. Carvajal, PhD, MPH,g Donna Bryson, RN, AE-C,h Nikki Stefan, RN, BSN,i Dean Billheimer, PhD,c,j and Lynn B. Gerald, PhD, MSPHb,g Tucson, Ariz, and Denver, Colo GRAPHICAL ABSTRACT

Supervising a once-daily inhaled corcosteroid regimen at school did not improve asthma control among Lano elementary students with self-reported asthma 10 Delayed Intervenon Schools

10 Immediate Intervenon Schools

Usual Care in Year 1 / Supervised Therapy in Year 2

Supervised Therapy in Year 1 & Year 2

189 Students Year 1

143 Students Year 1 Primary Analysis -0.08 (95% CI -0.31 – 0.14)

169 Students Years 1 & 2

Secondary Analysis -0.07 (95% CI -0.22 – 0.08)

217 Students Years 1 & 2

Background: School-supervised use of a once-daily inhaled corticosteroid regimen (supervised therapy) can improve medication adherence and asthma control. Objective: We sought to evaluate the effectiveness of supervised therapy in a unique setting and population. Methods: We conducted a cluster randomized trial of supervised therapy in 20 elementary schools with a disproportionate enrollment of low-income Latino students. Schools were purposively selected, matched, and randomized to

receive 9 months of supervised therapy with mometasone furoate or usual care. All English- or Spanish-speaking students with self-reported asthma were eligible. The Asthma Control Questionnaire (ACQ) was interviewer administered quarterly at school. Students in supervised therapy schools were hypothesized to have lower ACQ scores than students in usual-care schools. Results: Of 393 enrolled students, 189 students receiving immediate intervention and 143 students receiving delayed

From athe Department of Community, Environment and Policy, Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson; bAsthma and Airways Disease Research Center, University of Arizona, Tucson; cBIO5 Institute, University of Arizona, Tucson; dthe Department of Pediatrics, University of Colorado, Denver; e The Breathing Institute, Children’s Hospital Colorado, Denver; fthe Department of Pediatrics, University of Arizona College of Medicine, University of Arizona, Tucson; g the Department of Health Promotion Sciences, the Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson; hAmerican Lung Association of Southern Arizona, Tucson; ithe Department of Health Services, Tucson Unified School District, Tucson; and jthe Department of Epidemiology and Biostatistics, Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson. Supported by National Institutes of Health/National Heart, Lung, and Blood Institute grant 1R18HL110858-01A1; a research grant from the Investigator-Initiated Studies Program of Merck Sharp & Dohme Corporation who provided Asmanex (mometasone furoate) and Proventil (albuterol sulfate); and Thayer Medical Corporation, who

provided disposable valved holding chambers (LiteAire). The opinions expressed in this paper are those of the authors and do not necessarily represent those of Merck Sharp & Dohme Corp. Disclosure of potential conflict of interest: The authors declare that they have no relevant conflicts of interest. Trial registration: NCT01997463. Received for publication August 9, 2017; revised June 22, 2018; accepted for publication June 28, 2018. Corresponding author: Joe K. Gerald, MD, PhD, Department of Community Environment and Policy, Mel and Enid Zuckerman College of Public Health, Asthma and Airways Disease Research Center, University of Arizona, 1295 N Martin Ave, PO Box 245210, Tucson, AZ 85724-5210. E-mail: [email protected]. 0091-6749/$36.00 Ó 2018 American Academy of Allergy, Asthma & Immunology https://doi.org/10.1016/j.jaci.2018.06.048

1

2 GERALD ET AL

intervention provided 1 or more ACQ data points, were between 6 and 10 years of age, and were included in the primary analysis. At baseline, 39% of students reported taking a controller medication, and 24% had well-controlled asthma. Eighty percent of students receiving immediate intervention were prescribed mometasone. Schools administered 98% of prescribed doses when students attended school. Absences, weekends, and holidays reduced calendar adherence to 53%. During the first year, the mean ACQ score for students receiving immediate and delayed intervention was 1.55 (95% CI, 1.41-1.70) and 1.64 (95% CI, 1.47-1.80), respectively. The estimated treatment effect was 20.08 (95% CI, 20.31 to 0.14). Discussion: Compared with usual care, supervised therapy did not improve asthma control among this population of Latino students. Additional research is warranted to confirm these results. (J Allergy Clin Immunol 2018;nnn:nnn-nnn.) Key words: Schools, anti-inflammatory agent, medication adherence, asthma, child, directly observed therapy, randomized controlled trial, inhaled corticosteroid, cluster randomized trial

Medication adherence among children with asthma in the general population is insufficient to maximize symptom control and minimize exacerbation risk. For example, 14% of patients prescribed an inhaled corticosteroid (ICS) will never fill it, 29% will fill it once; 54% will refill it sporadically, and a mere 3% will refill it frequently enough to exceed 75% adherence.1 Long-term adherence in clinical trials with motivated and monitored participants is only marginally better. For example, just one quarter of participants in the Childhood Asthma Management Program exceeded 80% adherence during the study’s fourth year.2 Because nonadherence is an important cause of preventable morbidity,3 the National Asthma Education and Prevention Program and the American Academy of Asthma, Allergy & Immunology have urged researchers to develop more effective adherence programs.4,5 Schools are a logical setting to deploy such programs because schools are where children congregate, spend much of their day, and are frequently monitored.6 Engaging schools can also cost effectively reach the urban disadvantaged minority populations with the lowest adherence and highest morbidity.7,8 Some programs have supervised the use of a once-daily ICS regimen at school (supervised therapy) to increase medication adherence. Supervised therapy programs typically involve case identification, clinical assessment, medical care, asthma education, and care coordination.9 The specifics of program implementation are influenced by the research team’s experience, the school’s health resources, the population’s unique sociocultural determinants, and the local health system’s configuration. Several randomized controlled trials have shown supervised therapy to improve asthma control, reduce school absences, and decrease health care use.10-14 Successful program implementation requires intellectual, organizational, and operational capacity that many schools lack. Researchers have used their expertise and institutional resources to create the necessary conditions to implement and evaluate supervised therapy’s efficacy to address these gaps. For example, researchers have used pulmonary function tests to confirm asthma diagnoses, provided medications at no cost, coordinated care with medical providers and health insurance companies, and built Web-based monitoring systems to track daily symptoms.

J ALLERGY CLIN IMMUNOL nnn 2018

Abbreviations used ACQ: Asthma Control Questionnaire ICS: Inhaled corticosteroid MAR: Medication Administration Record MID: Minimally important difference NSLP: National School Lunch Program OCS: Oral corticosteroid SAMS: Supervised Asthma Medicine in Schools TUSD: Tucson Unified School District

Although necessary, these activities have limited our understanding of supervised therapy’s effectiveness under real-world conditions. To address this knowledge gap, we conducted a comparative effectiveness study in elementary schools serving primarily Latino students from disadvantaged communities in Tucson, Arizona. Although these schools lacked the resources to independently implement supervised therapy, we minimized our research team’s engagement by modifying a study design we previously used in Birmingham, Alabama (see Table E1 in this article’s Online Repository at www.jacionline.org).12 We hypothesized that supervised therapy would improve asthma control, reduce school absences, and prevent asthma-related health care use.

METHODS The Supervised Asthma Medicine in Schools (SAMS) study was a 2year, cluster randomized, delayed intervention trial that compared supervised therapy with usual asthma care practices (Fig 1). During 2014-2015 (year 1), schools were randomized to immediately initiate supervised therapy or delay implementation until 2015-2016 (year 2). The delayed intervention design, which is similar to a stepped-wedge design,15 simplified program implementation, enhanced recruitment, and ensured all schools would benefit from participation.16 Cluster randomization standardized asthma care practices within schools and minimized cross-contamination risk. Twenty Tucson Unified School District (TUSD) elementary schools were purposively selected to ensure that large urban schools serving students from low-income communities were disproportionately represented. Each elementary school had 1 full-time health assistant whose minimum qualifications included having a high school diploma, 3 years’ experience, cardiopulmonary resuscitation certification, and first-aid certification. TUSD also used full-time school nurses who supervised 2 to 4 health assistants. Selected schools were matched on student enrollment, turnover, ethnicity, and National School Lunch Program (NSLP) participation and then randomized in a 1:1 ratio to immediate or delayed intervention groups. Schools were randomized before students were recruited to facilitate implementation. Because placebo inhalers were unaffordable, the SAMS study was an unblinded open-label study. The SAMS study was approved and monitored by the Institutional Review Board at the University of Arizona and the Office of Curriculum and Instruction at TUSD. All English- or Spanish-speaking students reporting asthma on school health forms were eligible. School health personnel began recruiting students in March 2014 by sending flyers home through the ‘‘backpack express’’ and making follow-up telephone calls to parents. Students could enroll at any point during the school year, with the largest number enrolling from August to October 2014 (see Fig E1 in this article’s Online Repository at www.jacionline.org). Interested parents were contacted by research personnel who obtained written informed consent and acquired baseline demographic information. Verbal assent was obtained from students during their first encounter with the study team.

GERALD ET AL 3

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

Year 1 Aug 2014 – May 2015 10

R

Year 2 Aug 2015 – May 2016

Supervised Therapy (Immediate Intervenon)

Supervised Therapy Summer Break

Schools Usual Care (Delayed Intervenon)

10 Child Visit:

Supervised Therapy

V1

V2

V3

V4

V5

V6

V7

V8

N (immediate)= 146

141

146

151

130

133

123

116

N (delayed)= 111

109

104

115

104

97

91

83

Primary Outcome

Primary Analysis Unplanned Secondary Analysis Parent Call:

P1

P2

P3

P4

N (immediate) = 154

157

138

134

N (delayed) = 119

120

108

106

Secondary Outcomes

P5

P6

P7

P8

102

110

105

99

84

84

78

67

FIG 1. SAMS study design and data collection timeline.

Intervention Students’ need for daily controller medication was assessed at back-toschool clinics held at immediate intervention schools during the first year (in person) and at all schools during the second year (telemedicine). At these visits, study clinicians prescribed treatment based on parental responses from the baseline telephone interview, student-reported asthma control from that day’s visit, and a brief physical examination. Treatment options included as-needed albuterol sulfate with or without a 110, 220, or 440 mg/d dose of mometasone furoate delivered through a breath-actuated dry powder inhaler. The original design planned to duplicate the students’ at-home medication regimen, but unexpected federal budget cuts led us to seek a manufacturer donation of a standardized regimen instead. After the clinic visit, individualized asthma action plans were sent home with students, mailed to parents and their child’s primary asthma care provider, and maintained in the school’s health office. Although parents were encouraged to continue their child’s previous at-home medication regimen on weekends, holidays, and absences, no effort was made to facilitate adherence with this recommendation. By supervising medication administration on 5 of 7 weekdays, the SAMS study could have achieved a maximum calendar adherence level of 61% during the 9-month school year after accounting for weekends and school holidays. Study medications for school use were provided to participants at no cost, secured in the school’s health office, and refilled by research personnel as necessary. Supervised therapy was initiated when parents returned signed TUSD Medication Administration Record (MAR) forms to the school’s health office. All medication administration was supervised once daily by school health personnel and documented in the student’s MAR. Once initiated, treatment was continued until the end of the school year, up to 9 months in total. Treatment was discontinued over the summer break and resumed after clinical re-evaluation during the second year. The research team provided certain resources at no cost to help schools initiate supervised therapy. All school nurses and health assistants were provided a half-day in-service training on asthma management practices and appropriate inhaler technique before the start of each school year. A certified asthma educator employed by the American Lung Association made frequent school visits to answer health assistants’ questions, reinforce asthma training, and manage medication refills. During the first year only, a bilingual Latina nurse practitioner employed by the SAMS study provided clinical support and coordinated treatment changes with parents and asthma care providers. Letters were mailed to all local pediatricians,

pulmonologists, and health clinics before starting supervised therapy to inform providers. The study was also presented at scheduled meetings of the various local medical associations.

Standardized asthma education The Open Airways for Schools asthma self-management program was offered at immediate intervention schools during year 1 and all schools during year 2.17 The program’s 6 standardized, 45-minute, interactive, group-based lessons were conducted in the students’ own schools and were led by trained public health undergraduate and graduate students. Although Open Airways for Schools was designed for students in grades 3 to 5, we modified it with American Lung Association assistance for use in grades kindergarten to 2 as well.

Primary and secondary outcomes The primary outcome was the 6-item child version of the Asthma Control Questionnaire (ACQ).18 This standardized instrument asked students to report their symptom frequency, symptom burden, and albuterol use during the past week. Lower scores on the 0- to 6-point Likert scale indicated better asthma control. Because the ACQ was only validated for children 6 to 10 years of age, results obtained from students outside this range at the time of the interview were excluded. The instrument was interviewer administered at school in English or Spanish up to 8 times over the 2-year study beginning at baseline and repeated quarterly thereafter. Secondary outcomes included all-cause absences (TUSD school records) and asthma-related absences (parental interviews). Asthma-related oral corticosteroid (OCS) use, unscheduled clinic visits, emergency department visits, and hospitalizations were obtained from parental telephone interviews conducted up to 8 times during the 2-year study beginning at baseline and repeated quarterly thereafter. Controller medication adherence at school was obtained by means of annual review of student MAR forms. Controller medication adherence at home was not measured.

Power calculation The primary hypothesis was that students in immediate intervention schools would have lower mean ACQ scores than students in delayed intervention schools during year 1. The SAMS study was originally designed to have 80% power to detect a mean group difference of 0.37 ACQ units or

4 GERALD ET AL

J ALLERGY CLIN IMMUNOL nnn 2018

Eligible TUSD Elementary Schools (N= 66)

Excluded Schools (n= 46) Schools were purposively sampled to include those with highest enrollment of low-income, minority students.

Allocaon

Supervised Therapy (Immediate Intervenon) Schools recruing students = 10 517 of 4504 students had asthma (11.8%) 224 of 517 enrolled (43.3%) 22.4 per school

Year 1

Analyzed = 10 schools 16 students transferred prior to matriculaon 19 students not 6-10 years of age 189 students included in primary analysis 18.9 per school (range 7 – 36)

Analyzed = 10 schools 16 students transferred prior to matriculaon 10 students not 6-10 years of age 143 students included in primary analysis 14.3 per school (range 9 – 19)

Year 2

20 Randomized Schools 995 of 8916 students reported asthma (11.1%)

Analyzed = 10 schools 6 students aged-in from Year 1 22 students newly enrolled during Year 2 217 students included in unplanned analysis 21.7 per school (range 8 – 47)

Analyzed = 10 schools 4 students aged-in from Year 1 22 students newly enrolled during Year 2 169 students included in unplanned analysis 16.9 per school (range 12 – 21)

4 3

Usual Care (Delayed Intervenon) Schools recruing students = 10 478 of 4412 students had asthma (10.8%) 169 of 478 enrolled (35.4%) 16.9 per school

FIG 2. CONSORT diagram for the SAMS study cluster randomized trial.

greater, assuming 490 students were enrolled in 16 schools, and the intraclass correlation coefficient was 0.15 or less. In actuality the SAMS study only enrolled 332 students with allowable ACQ scores, but this loss of power was offset by recruiting 4 additional schools and observing a lower than expected intraclass correlation coefficient value of 0. Ultimately, the SAMS study had 80% power to detect a mean group difference of 0.26 ACQ units or greater. The ACQ’s minimally important difference (MID) is 0.4 to 0.5 units.18,19

Analysis plan A Gaussian hierarchical linear mixed-effects model was used to conduct a prespecified intent-to-treat analysis comparing group mean ACQ scores obtained during the 4 student visits during year 1. The model included fixed factors for intervention group, study visit, and an intervention group 3 study visit interaction. Random intercepts were included for each school and each student nested within the school. No clinical or demographic variables were included. The treatment effect was estimated by subtracting the model-estimated mean ACQ scores of students receiving delayed intervention from those receiving immediate intervention. Results are presented by using a point estimate and 95% CI. A more comprehensive unplanned secondary analysis was undertaken that used ACQ scores from all 8 visits over the 2-year study and estimated the treatment effect by using 2 linear contrasts. The first contrast measured differences between students receiving immediate and those receiving delayed intervention during year 1, and the second measured differences between year 1 and year 2 among students receiving delayed intervention after adjusting for time trends. These 2 contrasts were combined to estimate the total treatment effect. A similar second unplanned analysis investigated the effect of supervised therapy after excluding students who were presumed to have intermittent asthma: students receiving immediate

intervention not prescribed ICSs and students receiving delayed intervention who were not taking controller medication at home and whose baseline ACQ score was less than 1.25. Results are presented as point estimates and 95% CIs. Secondary outcomes of absences, OCS use, and unscheduled care visits were modeled by using generalized linear models patterned on the prespecified intent-to-treat ACQ model described for the primary analysis. These models included a fixed effect for intervention group and estimated the treatment effect with the same linear contrast used in the primary analysis. Absences, OCS use, and unscheduled health care visits were aggregated over year 1. Because of this, fixed effects for study visit and the visit 3 group interaction were not included in the model. Similarly, because each student only provided 1 data point per analysis, these models did not include a random intercept for student. A random intercept for school was still included. These 3 secondary outcomes were fit to hurdle models to account for zero-inflated data. In the truncated Poisson portion of the models, offsets for log time were included; in the logistic portion, time was included as a covariate. Hurdle models were fit by using Bayesian regression methods with either an uninformative prior (unscheduled care visits) or an informative prior (absences and OCS use) distribution.20 Emergency department visits were fit to a logistic regression model because very few children reported more than 1 visit. As in the logistic portion of the models above, time was included as a covariate; however, no random intercepts were included because of convergence issues. Because almost no children reported overnight hospitalizations, differences between groups were assessed with a Fisher exact test. The relationship between asthma control (ACQ scores) at each child visit and medication adherence in the immediately preceding 28-day period was investigated by fitting a model to the ACQ data, with fixed effects for visit and a 5-knot restricted cubic spline for adherence and random intercepts for school and student within school.

GERALD ET AL 5

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

TABLE I. Baseline parent, household, and student characteristics by randomization assignment Total (n 5 332), % (no.)

Parent and household characteristics Maternal education Less than high school High school Associate’s degree Bachelor’s degree or greater Monthly housing budget, mean (SD) Student demographics Age (y), mean (SD) Male sex Race White Other American Indian African American Ethnicity Latino ETS exposure Exposed in home Exposed outside of home BMI Underweight (<5th) _5th-<85th) Normal weight (> _85th-<95th) Overweight (> _95th) Obese (> Health status Excellent Very good Good Fair/poor

18 47 24 11 $702

Immediate intervention (n 5 189), % (no.)

(58) (147) (76) (35) ($262)

21 47 24 8 $680

(37) (84) (43) (14) ($239)

Delayed intervention (n 5 143), % (no.)

15 46 24 15 $730

(21) (63) (33) (21) ($287)

8.1 (1.5) 63 (210)

8.2 (1.5) 62 (118)

8.1 (1.5) 64 (92)

42 34 13 11

41 32 15 12

44 37 9 10

(129) (105) (38) (33)

(70) (55) (26) (20)

(59) (50) (12) (13)

75 (238)

75 (136)

75 (102)

15 (46) 7 (22)

14 (25) 7 (13)

15 (21) 7 (9)

3 51 18 28

(8) (170) (60) (94)

2 51 20 28

(3) (97) (37) (52)

4 51 16 30

(5) (73) (23) (42)

15 38 36 11

(49) (123) (116) (35)

15 39 36 10

(28) (71) (66) (18)

15 37 36 12

(21) (52) (50) (17)

Missing data: maternal education, 16; monthly housing budget, 84; second-hand ETS, 14; race, 27; ethnicity, 14; and health status, 9. No differences were observed between students receiving immediate or delayed intervention at a significance level of a P value of .05. ETS, Environmental tobacco smoke.

RESULTS Although 393 students enrolled during year 1, only 332 (84%) provided sufficient data to be included in the primary analysis (Fig 2). A greater proportion of eligible students enrolled from immediate intervention schools than delayed intervention schools (43% vs 35%, respectively; P 5 .01). Over the 2-year study, 93% of planned student interviews and 73% of parent interviews were completed. Detailed enrollment completion data are provided in Fig E1 and Table E2 in this article’s Online Repository at www. jacionline.org. Baseline demographic and clinical characteristics were balanced across groups (Table I). Briefly, the students’ mean age was 8.1 years, 63% were male, 75% were Latino, and 93% were eligible for the NSLP. Based on a mean monthly housing budget of $702 (SD, $262), students were estimated to reside in households with an average annual income of approximately $21,000, which is near the federal poverty level.21 Nevertheless, few caregivers reported difficulty accessing or paying for asthma-related care, likely because most were publicly (74%) or privately insured (17%, Table II). At baseline, 39% of students reported controller medication use, and only 24% had well-controlled asthma. After clinical evaluation, 80% of students assigned to immediate intervention schools were prescribed mometasone during year 1, with 45% prescribed the 110-mg dose, 25% prescribed the 220-mg dose, and 10% prescribed the 440-mg dose. The remaining 20% of students were prescribed albuterol only.

Primary and secondary outcomes During year 1, the group mean ACQ score for students receiving immediate intervention was 0.08 units (CI 95%, 20.31 to 0.14 units) less than that for students receiving delayed intervention (Table III). This estimate was not statistically significant, and the 95% CI did not contain the ACQ’s reported MID of 0.4 to 0.5 units. No differences in absences or health care use were observed (Table III). Medication adherence and school absences During years 1 and 2, school health personnel administered 98% of all prescribed doses when students were present at school; however, adherence was lower than the hoped for 61% because of frequent absences. Supervising ICSs on days when students were present at school achieved a calendar adherence level of 53% (interquartile range, 51% to 57%) among students receiving immediate intervention during year 1 and 53% (interquartile range, 52% to 56%) among all students during year 2. According to school records, students in immediate intervention and delayed intervention schools missed an average of 16 and 15 days, respectively, of the 180-day school year during year 1 (Table III). Only 4 to 5 of these absences were reported by parents to be asthma related. Using a threshold of 15 absences per year,22 43% of students receiving immediate intervention and 37% of those receiving delayed intervention were chronically absent during year 1.

6 GERALD ET AL

J ALLERGY CLIN IMMUNOL nnn 2018

TABLE II. Access to health care, prior health care use, and clinical features at baseline by randomization assignment

Access to health care Health insurance Public Private Uninsured Medical home Specialist Primary care None At-home regimen Controller, ICS Controller, LTRA Reliever only None Needed medication but did not have it Difficulty obtaining asthma care Difficulty paying for care Health care use in past 12 mo Used OCS ED visit Hospitalization Clinical features History of atopic dermatitis ACQ, mean (SD) ACQ, categorical Well controlled (<0.75) _0.75-<1.5) Indeterminate (> _1.5) Poorly controlled (> Clinician-rated control Well controlled Not well controlled Poorly controlled Study medication regimen* 110 mg/d 220 mg/d 440 mg/d None

Total (n 5 332), % (no.)

Immediate intervention (n 5 189), % (no.)

74 (238) 17 (53) 9 (29)

75 (135) 17 (30) 8 (15)

74 (103) 16 (23) 10 (14)

6 (19) 78 (253) 16 (50)

7 (12) 78 (142) 16 (29)

5 (7) 80 (111) 15 (21)

31 8 41 19 23 19 17

33 7 45 15 24 16 17

29 10 37 25 25 23 16

(97) (26) (129) (60) (75) (60) (54)

(58) (12) (79) (26) (44) (30) (31)

Delayed intervention (n 5 143), % (no.)

(39) (14) (50) (34) (31) (30) (23)

42 (132) 30 (94) 6 (20)

41 (73) 28 (51) 7 (12)

43 (59) 31 (43) 6 (8)

37 (116) 1.7 (1.2)

41 (73) 1.7 (1.2)

31.2 (43) 1.7 (1.2)

27 (89) 24 (81) 49 (162)

28 (52) 23 (44) 49 (93)

26 (37) 26 (37) 48 (69)

— — —

24 (45) 64 (117) 12 (22)

— — —

— — — —

45 25 10 20

— — — —

(82) (46) (18) (37)

Missing data: health insurance, 12; medical home, 10; at-home medication, 20; needed medication but did not have it, 12; difficulty obtaining care, 9; difficulty paying for care, 9; used OCS, 19; had emergency department visit, 14; hospitalized, 13; clinician-rated control, 5, study medication regimen, 4; and atopic dermatitis, 14. No differences were observed between immediate and students receiving delayed intervention at a significance level of a P value of .05. ED, Emergency department; LTRA, leukotriene receptor antagonist. *Two students were maintained on their at-home regimen, with 1 dose supervised at school.

Unplanned secondary analyses of asthma control using year 1 and year 2 data We examined the robustness of our findings to alternative specifications using data from both years, thereby increasing the sample size to 386 students and increasing the total number of possible observations per student to 8. These analyses estimated the between-group difference during year 1 and the differences between years 1 and 2 for students who transitioned from delayed intervention to supervised therapy. This latter estimate was adjusted for a statistically significant temporal trend of 0.28 (CI 95%, 0.12-0.43) units of improvement per year (Fig 3). By using both years’ data, the estimated treatment effect of supervised therapy was 20.07 units (95% CI, 20.22 to 0.08). We repeated this approach among the 288 students most likely to have persistent asthma: students receiving supervised therapy prescribed mometasone by study clinicians in either year and students receiving delayed intervention during year 1 who used controller medication before

enrollment or whose ACQ scores were 1.25 or greater at baseline. The estimated treatment effect was 20.12 units (95% CI, 20.30 to 0.05).

Relationship between adherence and asthma control In an analysis comprising 230 students and 892 observations over 2 years, there was no relationship between ACQ scores and the preceding 28-day calendar adherence (P 5 .62, Fig 4). This was true by level of assigned study medication dose (P 5 .15) and by adherence 3 dose interaction (P 5 .16). DISCUSSION Supervised therapy, as implemented in this unique population of elementary school students residing in urban disadvantaged Latino communities, did not improve asthma

GERALD ET AL 7

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

TABLE III. ACQ scores, all-cause absences, and asthma-related absences and health care use during year 1 by randomization assignment

ACQ,* mean (95% CI) All-cause absences  _1 absence Percentage with > _1 absence Mean absences if > Asthma-related absencesà _1 absence Percentage with > _1 absence Mean absences if > OCS use _1 OCS use Percentage with > _1 OCS use Mean OCS uses if > Unscheduled visits _1 visit Percentage with > _1 visit Mean visits if > ED visits _1 ED visit Percentage with > Hospitalizations _1 hospitalization Percentage with >

Immediate intervention

Delayed intervention

Estimated treatment effect

1.6 (1.4 to 1.7)

1.6 (1.5 to 1.8)

98 (95 to 100) 15.5 (12.5 to 18.9)

99 (97 to 100) 13.4 (10.8 to 16.4)

OR: 0.6 (0.2 to 1.6) RR: 1.2 (0.9 to 1.5)

68 (53 to 81) 6.5 (5.4 to 7.8)

73 (59 to 85) 6.6 (5.4 to 7.9)

OR: 0.8 (0.3 to 1.7) RR: 1.0 (0.8 to 1.3)

30 (17 to 45) 2.8 (1.5 to 5.0)

30 (17 to 46) 2.6 (1.5 to 4.4)

OR: 1.0 (0.5 to 1.9) RR: 1.3 (0.5 to 2.7)

50 (35 to 66) 1.9 (1.3 to 2.6)

56 (40 to 71) 2.1 (1.5 to 2.9)

OR: 0.8 (0.4 to 1.5) RR: 0.9 (0.3 to 1.8)

32 (18 to 50)

33 (17 to 53)

OR: 1.0 (0.5 to 2.0)

3 (0.6 to 9)

3 (0.3 to 9)

20.1 (20.3 to 0.1)

P 5 1.0§

ED, Emergency department; OR, odds ratio; RR, rate ratio. *The model estimated mean ACQ score by intervention group from year 1 based on 332 students with 1 or more ACQs and 6 to 10 years of age.  All-cause absences reported by TUSD. àAsthma-related absences reported by parents. §Fisher exact test: No statistically significant differences were observed between students receiving immediate and those receiving delayed intervention at a significance level of a P value of .05.

FIG 3. Model estimated mean ACQ scores during year 1 (intervention year, visits 1-4) and year 2 (delayed intervention year, visits 5-8) for students receiving supervised therapy and those receiving usual care.

control, as measured by using the ACQ. The lower bound of the 95% CI estimating improvement in ACQ score, 0.31 units, was smaller than the instrument’s 0.4 to 0.5-unit MID.18,19 Therefore, it is unlikely that a clinically meaningful improvement was missed because of inadequate power. This conclusion was also robust to alternative model specifications and was corroborated by a lack of improvement in absenteeism or health care use. Comparisons with other large supervised therapy trials suggest several reasons why our approach might not have produced the expected improvements. Before addressing these points, it is worth briefly reviewing these past trials. Three trials conducted in

Rochester, New York, evaluated the effect of supervised therapy on symptom-free days among children 3 to 10 years of age, 60% of whom were African American. In chronological order, these 3 supervised therapy trials increased the number of symptom-free days by a nonsignificant amount, by 0.92 (95% CI, 0.5–1.33) days per 2 weeks, and by 0.69 (95% CI, 0.15–1.22) days per 2 weeks.11,13,14 A single trial conducted in Birmingham, Alabama, evaluated the effect of supervised therapy on a composite outcome, episodes of poor asthma control (EPAC), among children 7 to 16 years of age, 90% of whom were African American. For students in supervised therapy schools, the treatment group 3 year interaction was marginally significant at a P value

8 GERALD ET AL

J ALLERGY CLIN IMMUNOL nnn 2018

FIG 4. Association between 28-day adherence and ACQ scores, with least2squares means curve of ACQ scores averaged over time with 95% CIs. Data from 230 students who provided 1 or more ACQ values and had exactly 28 days of calculable controller use before ACQ measurement (230 participants, 892 data points). Only data fitting the above criteria and with adherence between 25% and 75% are shown. The modeled relationship between 28-day adherence and ACQ core is nonsignificant (P 5 .62).

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

of .065 indicating that children in the supervised therapy group showed greater improvement.12 Unlike previous trials, we allowed all students with self-reported asthma to participate without requiring evidence of persistent asthma. For this reason, 20% of our participants were not prescribed study medication and were unlikely to benefit substantially from our intervention. However, when these participants were excluded in a secondary analysis, there was no evidence that the remaining participants experienced a benefit either. Unlike previous trials, we did not facilitate ICS treatment at home during weekends, holidays, and absences. Because community-based controller adherence is notoriously low,1 many of our participants were using only ICSs administered at school. Because weekends, holidays, and absences reduced actual calendar adherence to 53%, supervised therapy might not have been sufficient by itself to improve asthma control. This hypothesis is supported by the absence of a relationship between 28-day adherence and ACQ scores. At the population level, higher adherence is associated with better outcomes; however, at the individual level, the amount of adherence needed to achieve adequate symptom control remains controversial.23 Some have argued that 80% or greater adherence might be necessary to achieve adequate symptom control3,24; others have argued that levels between 40% and 60% might be sufficient.25 Across adherence levels between 25% and 75% in this study, there was no relationship between adherence and asthma control. If substantiated, this has important implications because these adherence levels are comparable with those routinely achieved in clinical practice. Future study is needed to more accurately define the minimum adherence needed to induce asthma control, minimize exacerbation risk, or both.23 Unlike previous trials, we used a substantially different approach to measure asthma control. Instead of relying on parent-reported asthma symptoms based on daily diary recall, we relied on child-reported recall of similar domains captured on a standardized asthma control instrument. Based on recommendations established by the National Institute of Health/Agency for Healthcare Research and Quality, neither approach is clearly superior to the other in prospective trials of children with asthma.26 Although it is tempting to attribute our null finding to this decision, it seems unlikely given that ACQ scores were sensitive to change in asthma control over time. Lastly, the SAMS study was conducted in schools whose predominant minority population was Latino students of Mexican origin rather than African American students. Might our students have been less responsive to ICSs? Historically, the studies used to establish the efficacy of ICSs in children have relied on its effect in white children with additional oversampling of African American children.27 In general, Latino children, particularly those of Mexican origin, have been underrepresented in clinical trials.27 Although data on differential responsiveness are limited, a secondary analysis examining predictors of ICS responsiveness in children observed a clear benefit among non-Latino white children (hazard ratio, 0.37; 95% CI, 0.18-0.75) but not Latino children (hazard ratio, 1.16; 95% CI, 0.47-2.86).28 Given that Latino children comprise 20% or more of the general population,29 this question deserves future investigation. Serving disadvantaged youth is one element all supervised therapy trials have had in common. Poverty is detrimental to a

GERALD ET AL 9

wide range of health outcomes, as evidenced by the 10- to 15-year life expectancy gap between the highest and lowest income earners in the United States.30 Compared with children nationally, SAMS study students were half as likely to report excellent/very good health and were 3 times more likely to report fair/poor health.31 Even when compared with children living in households at less than the 200% federal poverty level, SAMS study students were 25% less likely to have excellent/very good health and 50% more likely to have fair/poor health.31 Disparities in asthma burden by race are well documented.32 Disentangling the biologic, social, and environmental drivers of these disparities suggest that 50% can be attributable to socioeconomic hardship (eg, income and access to care) and another 25% to environmental exposures (eg, mold, dander, and pollutants).33 For this reason, the effect of supervised therapy might have been attenuated by the social and environmental conditions of the student’s lived experience. One expression of this potential disadvantage is the dramatic rate of chronic absenteeism in our students. On average, SAMS study students missed 14 school days per year, with only 4 to 5 being attributable to asthma. By missing 15 or more days of school, 41% met the federal definition of chronic absenteeism.22 Nationally, only about 10% of elementary students are chronically absent22; however, schools serving low-income minority students are at greatest risk.34 A comparable study of predominately African American elementary students with asthma reported half as many absences, 8 per year, but a similar proportion being asthma-related (31%).35 Risk factors for absenteeism include poor health, obesity, poorly controlled asthma, low household income, and attending schools where many students are NSLP eligible.36,37 Our null finding should be viewed cautiously in light of the robust literature on school-based asthma programs in general and supervised therapy in particular.38 By minimizing the research team’s involvement with program implementation, we likely identified several areas critical to bridging the gap between efficacy and effectiveness. If greater than 60% calendar adherence is needed to induce asthma control, then supervised therapy alone might be insufficient without also addressing medication adherence at home. Although relying on self-reported asthma for program eligibility is simple and inexpensive, it might lead to overtreatment in a subset of students with mild asthma who are unlikely to benefit. Increasing medication adherence alone might not be sufficient to overcome the social and environmental circumstances in which children with asthma live, learn, and play because socioeconomic status, environmental exposures, obesity, and stress all make it difficult to maintain asthma control. Lastly, because the SAMS study enrolled a unique population of low-income Latino students who were predominantly of Mexican origin, generalizability to other settings and populations cannot be assumed. Furthermore, only 39% of eligible children enrolled in the SAMS study. In Arizona 11% of children live with at least 1 parent who is an unauthorized immigrant, with a greater proportion concentrated among border communities like Tucson.39 Even if both parents are legal residents, extended family members might be unauthorized. Such families tend to be less likely to interact with government programs, nonprofit organizations, and the health care system.40 In conclusion, the SAMS study demonstrated that supervised therapy is feasible in schools with dedicated health assistants, school nurses, or both. With minimal interruption of their work

10 GERALD ET AL

routine or the students’ education, school health personnel ensured near 100% adherence with daily ICS administration when children were present at school. However, this might still be insufficient to improve asthma outcomes because of countervailing forces, such as frequent school absences, poor at-home adherence, and detrimental social and environmental exposures. We acknowledge the assistance of Aimee Snyder and Ashely Lowe, 2 competent, caring, and detail-oriented public health doctoral students. We also acknowledge Sean McKenzie, a Master’s in Public Health student, who helped create a developmentally appropriate K-2 version of the Open Airways for Schools curriculum. We also thank the administrators and staff of the TUSD who allowed us to work with their students and welcomed our interruptions of their daily routine. We owe a debt of gratitude to the numerous school nurses and health assistants who provided care to the children enrolled in our study. Lastly, we thank other University of Arizona students who helped us collect data and provide the Open Airways in School asthma education program as part of their applied public health training.

Clinical implications: School-supervised use of a once-daily ICS regimen achieved 53% calendar adherence; however, it did not improve asthma control, school absences, or health care use compared with usual care. REFERENCES 1. Wu AC, Butler MG, Li L, Fung V, Kharbanda EO, Larkin EK, et al. Primary adherence to controller medications for asthma is poor. Ann Am Thorac Soc 2015;12:161-6. 2. Krishnan JA, Bender BG, Wamboldt FS, Szefler SJ, Adkinson NF Jr, Zeiger RS, et al. Adherence to inhaled corticosteroids: an ancillary study of the Childhood Asthma Management Program clinical trial. J Allergy Clin Immunol 2012;129: 112-8. 3. Williams L, Peterson E, Wells K, Ahmedani B, Kumar R, Burchard E, et al. Quantifying the proportion of severe asthma exacerbations attributable to inhaled corticosteroid nonadherence. J Allergy Clin Immunol 2011;128:1185-91. 4. Gupta RS, Weiss KB. The 2007 National Asthma Education and Prevention Program asthma guidelines: accelerating their implementation and facilitating their impact on children with asthma. Pediatrics 2009;123(suppl 3):S193-8. 5. American Academy of Allergy, Asthma & Immunology. Pediatric Asthma: Promoting Best Practice a Guide for Managing Asthma in Children. Rochester (NY): American Academy of Allergy, Asthma & Immunology; 1999. 6. Lynn J, Oppenheimer S, Zimmer L. Using public policy to improve outcomes for asthmatic children in schools. J Allergy Clin Immunol 2014;134:1238-44. 7. Gerald JK, Grad R, Bailey WC, Gerald LB. Cost-effectiveness of school-based asthma screening in an urban setting. J Allergy Clin Immunol 2010;125: 643-50.e12. 8. Noyes K, Bajorska A, Fisher S, Sauer J, Fagnano M, Halterman JS. Cost-effectiveness of the School-Based Asthma Therapy (SBAT) program. Pediatrics 2013;131:e709-17. 9. McEwen M, Johnson P, Neatherlin J, Millard MW, Lawrence G. School-based management of chronic asthma among inner-city African-American schoolchildren in Dallas, Texas. J Sch Health 1998;68:196-201. 10. Millard MW, Johnson PT, McEwin M, Neatherlin J, Lawrence G, Kennerly DK, et al. A randomized controlled trial using the school for anti-inflammatory therapy in asthma. J Asthma 2003;40:769-76. 11. Halterman JS, Szilagyi PG, Yoos HL, Conn KM, Kaczorowski JM, Holzhauer RJ, et al. Benefits of a school-based asthma treatment program in the absence of secondhand smoke exposure: results of a randomized clinical trial. Arch Pediatr Adolesc Med 2004;158:460-7. 12. Gerald LB, McClure LA, Mangan JM, Harrington KF, Gibson L, Erwin S, et al. Increasing adherence to inhaled steroid therapy among schoolchildren: randomized, controlled trial of school-based supervised asthma therapy. Pediatrics 2009;123:466-74. 13. Halterman JS, Szilagyi PG, Fisher SG, Fagnano M, Tremblay P, Conn KM, et al. Randomized controlled trial to improve care for urban children with asthma: results of the School-Based Asthma Therapy trial. Arch Pediatr Adolesc Med 2011;165: 262-8. 14. Halterman JS, Fagnano M, Tajon RS, Tremblay P, Wang H, Butz A, et al. Effect of the School-Based Telemedicine Enhanced Asthma Management (SB-TEAM)

J ALLERGY CLIN IMMUNOL nnn 2018

15. 16.

17.

18.

19.

20. 21.

22. 23.

24.

25. 26. 27. 28.

29.

30.

31.

32.

33.

34.

35. 36.

37.

38. 39.

40.

program on asthma morbidity: a randomized clinical trial. JAMA Pediatr 2018; 172:e174938. Ellenberg SS. The stepped-wedge clinical trial: evaluation by rolling deployment. JAMA 2018;319:607-8. Ji P, DuBois DL, Flay BR, Brechling V. ‘‘Congratulations, you have been randomized into the control group!(?)’’: issues to consider when recruiting schools for matched-pair randomized control trials of prevention programs. J Sch Health 2008;78:131-9. Open Airways for Schools. Available at: http://www.lungusa.org/lung-disease/ asthma/in-schools/open-airways/open-airways-for-schools-1.html. Accessed August 31, 2018. Juniper EF, Gruffydd-Jones K, Ward S, Svensson K. Asthma Control Questionnaire in children: validation, measurement properties, interpretation. Eur Respir J 2010; 36:1410-6. Nguyen JM, Holbrook JT, Wei CY, Gerald LB, Teague WG, Wise RA, et al. Validation and psychometric properties of the Asthma Control Questionnaire among children. J Allergy Clin Immunol 2014;133:91-7, e1-6. B€urkner P-C. An R package for Bayesian multilevel models using Stan. J Stat Software 2017;80(1). The PEW Charitable Trust. The precarious state of family balance sheets. 2015. Available at: http://www.pewtrusts.org/;/media/assets/2015/01/fsm_balance_ sheet_report.pdf. Accessed August 31, 2018. Chronic absenteeism in the nation’s schools. Available at: https://www2.ed.gov/ datastory/chronicabsenteeism.html. Accessed August 31, 2018. Engelkes M, Janssens HM, de Jongste JC, Sturkenboom MC, Verhamme KM. Medication adherence and the risk of severe asthma exacerbations: a systematic review. Eur Respir J 2015;45:396-407. Lasmar L, Camargos P, Champs NS, Fonseca MT, Fontes MJ, Ibiapina C, et al. Adherence rate to inhaled corticosteroids and their impact on asthma control. Allergy 2009;64:784-9. Jentzsch NS, Camargos P, Sarinho ES, Bousquet J. Adherence rate to beclomethasone dipropionate and the level of asthma control. Respir Med 2012;106:338-43. Busse W, Morgan WJ, Taggart V, Togias A. Asthma outcomes workshop: overview. J Allergy Clin Immunol 2011;129(suppl):S1-8. Kelly ML, Ackerman PD, Ross LF. The participation of minorities in published pediatric research. J Natl Med Assoc 2005;97:777-83. Gerald JK, Gerald LB, Vasquez MM, Morgan WJ, Boehmer SJ, Lemanske RF Jr, et al. Markers of differential response to inhaled corticosteroid treatment among children with mild persistent asthma. J Allergy Clin Immunol Pract 2015;3: 540-6.e3. Statistical portrait of Hispanics in the United States. Available at: http://www. pewhispanic.org/2016/04/19/statistical-portrait-of-hispanics-in-the-united-stateskey-charts/; 2016. Accessed August 31, 2018. Chetty R, Stepner M, Abraham S, Lin S, Scuderi B, Turner N, et al. The association between income and life expectancy in the United States, 2001-2014. JAMA 2016; 315:1750-66. Pastor PN, Reuben CA, Duran CR. Reported child health status, Hispanic ethnicity, and language of interview: United States, 2011-2012. Natl Health Stat Report 2015;(82):1-10. Akinbami LJ, Moorman JE, Simon AE, Schoendorf KC. Trends in racial disparities for asthma outcomes among children 0 to 17 years, 2001-2010. J Allergy Clin Immunol 2014;134:547-53.e5. Beck AF, Huang B, Auger KA, Ryan PH, Chen C, Kahn RS. Explaining racial disparities in child asthma readmission using a causal inference approach. JAMA Pediatr 2016;170:695-703. Balfanz R, Byrnes V. Chronic Absenteeism: Summarizing What we Know from Nationally Available Data. Baltimore: Johns Hopkins University Center for Social Organization of Schools; 2012. Moonie SA, Sterling DA, Figgs L, Castro M. Asthma status and severity affects missed school days. J Sch Health 2006;76:18-24. Echeverria SE, Velez-Valle E, Janevic T, Prystowsky A. The role of poverty status and obesity on school attendance in the United States. J Adolesc Health 2014;55: 402-7. Meng YY, Babey SH, Wolstein J. Asthma-related school absenteeism and school concentration of low-income students in California. Prev Chronic Dis 2012;9: E98. Cicutto L, Gleason M, Szefler SJ. Establishing school-centered asthma programs. J Allergy Clin Immunol 2014;134:1223-31. Passel J, Cohn D. Unauthorized immigrant totals rise in 7 states, fall in 14: decline in those from Mexico fuels most state decreases. Pew Research Center’s Hispanic Trends Project. Washington (DC): Pew Research Center; 2014. Holguin F, Mannino D, Anto J, Mott J, Ford E, Teague W, et al. Country of birth as a risk factor for asthma among Mexican Americans. Am J Respir Crit Care Med 2005;171:103-8.