An education intervention for childhood asthma by Aboriginal and Torres Strait Islander health workers: a randomised controlled trial

Children (< 18 years) were referred to the specialist clinics in a primary health care setting on Thursday Island and Horn Island, and in Bamaga, Torres Strait, Queensland, between April 2005 and March 2007, and reviewed by two paediatric respiratory physicians (A B C and I B M) using a standardised protocol for data collection. The children had a provisional diagnosis of asthma, or had been referred by IHCWs for assessment. None had previously been seen by the respiratory team.

Spirometry was performed in the standing position using a noseclip and a spirometer (calibrated daily) approved by the American Thoracic Society. Predicted values of Hibbert and colleagues were used.13 Clinical asthma was defined as repeated episodes of wheeze with dyspnoea that responded to bronchodilators. In children aged 3–6 years, two or more episodes of wheezy illness associated with cough and shortness of breath, and documented amelioration of symptoms and clinical signs after administration of a bronchodilator, supported a diagnosis of asthma.14 Severity of asthma was classified as persistent, frequent episodic, or infrequent episodic, based on the clinical pattern in the past 12 months.14

Before enrolment, all children had an asthma education session with a trained IHCW using the adapted asthma booklets. The intervention group had three additional education sessions with a trained IHCW at 1, 3 and 6 months after the baseline visit, using the same educational resources. Adherence to the study protocol (Box 1) was monitored by checking data collected during these visits. Control group children were not visited by the IHCWs. At 12 months, all children completing the study had a follow-up clinical consultation.

The primary endpoint was the number of unscheduled visits to hospital or a doctor because of asthma exacerbation in the 12 months’ follow-up. Secondary analysis focused on measures of quality of life (QoL)15,16 and the functional severity index for asthma;17 asthma knowledge and use of asthma action plans (AAPs); and school days missed because of wheezing.

Data on all acute exacerbations of asthma requiring treatment in a health facility were collected from the child’s medical records (at health clinics and Thursday Island Hospital) from 12 months before to 12 months after each child’s baseline visit. Data for secondary outcomes were collected at the baseline visit and at the 12 months’ follow-up, and also at each education session in the intervention group. Children and their carers were asked about the written AAP and what to do if asthma symptoms worsen. Carers of children taking regular medication were assessed for basic knowledge of asthma medications (eg, dosage, how the drugs work) and drug delivery technique.11 The functional severity index,17 a measure of functional impairment caused by asthma over a period of 12 months was scored (total score, 0–24; higher scores indicate worse impairment), and children and carers completed the Paediatric Asthma Quality of Life Questionnaire.15,16

Sample size calculations were based on estimates of frequency of asthma exacerbations from our previous experience in this community and determined a priori. Assuming a study power of 90%, a = 0.05, and using a Poisson process, the minimal study sample needed to detect a 33% reduction in unscheduled hospital or doctor visits (three versus two visits) was 54 children in each study arm. For secondary outcome measures (eg, QoL), with 30–40 children in each arm and a = 0.05, the study has 90% power to detect a 20%–25% reduction in total mean score.

Consecutive patients with a confirmed asthma diagnosis were eligible for our study. At the clinic where eligibility was assessed, informed consent was obtained from parents or guardians, and children were randomly allocated to the intervention group (additional education sessions), or to the control group (no additional education sessions) (Box 1). A randomly generated list (using a computer-generated permuted block design) within age strata (< 7 and ≥ 7 years) was used for study allocation; allocation group was revealed after enrolment. Staff collecting data from the medical records during the follow-up did not know the children and were blinded to the study allocation; other clinical study staff and the children were aware of the study allocation.

As study enrolment was slower than anticipated, we modified the intervention allocation based on estimates of asthma prevalence12 and guided by the sample size and power calculations. If a child was allocated to “additional education” but there was no trained IHCW in the child’s community to administer the intervention, we changed the allocation to “no additional education” (n = 7). If “no additional education” was not possible, either because a sibling had been allocated to “additional education” (n = 8) or the IHCW at the community health centre was familiar with the intervention (n = 5), we changed allocation to “additional education”.

We used the Statistical Package for the Social Sciences (version 15.0; SPSS Inc, Chicago, Ill, USA). Baseline characteristics are presented as mean and SD (data normally distributed), median and range (data not normally distributed) and proportion. Where variables had more than two categories, crude significance levels were calculated (χ2 test of the association). For other variables, logistic regression was used to adjust for potential confounders. Outcome data for each group were analysed “per protocol”. The potential confounder (asthma severity at baseline) was incorporated into the multivariate analysis. We used general linear models for normally distributed data, non-parametric tests (Kruskal–Wallis test) for data not normally distributed, and logistic regression models.

As all children had a clinical consultation and an asthma education session at the baseline visit, we also evaluated the effect of this intervention on QoL scores, functional severity index, and asthma exacerbations, comparing outcome measures at baseline with those at 12 months, overall and in the intervention and control groups. Non-parametric tests (Wilcoxon signed rank test) were used.

To check whether bias was introduced by per-protocol analysis instead of intention-to-treat analysis, we conducted multivariate analysis including the variables “treatment allocation as per protocol”, “randomisation” (randomised v non-randomised), the interaction term “randomisation”*“treatment allocation as per protocol”, and “asthma severity at baseline”, and for each outcome variable. As P > 0.05 applied to the interaction term in all models run (ranging from 0.093 for carers’ knowledge of asthma medication to 0.999 for carers’ knowledge of how preventers and relievers work), we proceeded with per-protocol analysis without adjusting for randomisation. As siblings from the same families were included in the study (11 children), we repeated the analysis including one randomly chosen child from each family. The results of subgroup analysis were very similar to those of the whole sample, so we have only presented data for the whole cohort.

We received support from the Torres Strait Regional Health Council and the Torres Strait and Northern Peninsula Area Health Service District. Ethics approval was given by the Queensland Institute of Medical Research Human Research Ethics Committee and the Children’s Health Services District Ethics Committee. We provided study results to communities through written reports to councils, face-to-face presentations, and local radio interviews. A flyer with a summary of the findings in clear English and the key points in Torres Strait Creole was given to those who attended a community meeting in 2008.

The intervention and control groups were comparable at baseline (Box 2 and Box 3) except for asthma severity, which appeared to be worse in the control group. There were no significant differences between the intervention and control groups in ethnicity, or parents’ highest level of education. One in 10 children possessed an AAP, two-thirds of the carers could not name their child’s medication, half did not know the dosage, and most could not explain how asthma medications worked (Box 3).

No significant differences were observed between the groups in the number of unscheduled hospital and doctor visits caused by asthma exacerbation in the 12 months before the intervention (median [range], 1.0 [0–7] for both groups). The median (range) total score for the functional severity index was 8.5 (0–20) for the intervention group compared with 9.0 (0–24) for the control group (P = 0.01). However, there were no significant differences in baseline QoL scores between the intervention and control groups.

The mean (SD) time between the baseline visit and the final consultation was 13.7 (3.15) months for the intervention group and 14.8 (4.95) months for controls (P = 0.32). The intended number of encounters with study personnel was two visits for controls (baseline and 12 months’ follow-up) and five for the intervention group (baseline, three extra education sessions, and 12 months’ follow-up). The median (range) number of actual extra encounters was 2 (0–4) for the intervention group and 0 (0–2) for controls.

Half the children (52%) had no asthma episode that required a visit to hospital or a doctor in the follow-up period (Box 4). The median number (range) of unscheduled hospital or doctor visits was 1.0 (0–4) for the intervention group and 0 (0–4) for controls (P = 0.25). There were eight asthma-related hospital admissions (four in each group). The mean difference between the number of episodes of asthma for the intervention group (mean, 1.0) and controls (mean, 0.7) was 0.30 (95% CI, 0.22–0.39).

In the intervention group, compared with the control group, there was a significant improvement in asthma knowledge, and more carers knew where their child’s AAP was kept and were able to interpret it (Box 5); more intervention group carers could accurately recall their child’s medication dosage (difference, 37%; 95% CI, 19%–54%), knew how β2 agonists worked (difference, 55%; 95% CI, 38%–72%), and knew how preventers and relievers worked (difference, 31%; 95% CI, 11%–51%). They also reported having used their AAP (difference, 32%; 95% CI, 4%–60%) and could describe it accurately (difference, 29%; 95% CI, 9%–49%).

Children in both groups reported missing school days because of wheezing in the previous year (difference, 4%; 95% CI, − 25% to 33%); however, control group children missed more days than those in the intervention group (all school-aged children in the intervention group missed between 0 and 6 days, whereas 21% of those in the control group missed 7–14 days; difference, 21% (95% CI, 5%–36%). The median (range) total score for the functional severity index was 5.0 (0–22) for the intervention group and 5.0 (0–20) for the control group (P = 0.65). The median (range) QoL scores for carers were 6.7 (1.9–7.0) for the intervention group and 6.8 (2.7–7.0) for the control group (P = 0.86); for children, QoL scores were 7.0 (2.5–7.0) and 6.9 (3.5–7.0), respectively (P = 0.57).

Among the whole group, there was a decrease in the number of unscheduled hospital or doctor visits from the 12-month period before (median [range], 1.0 [0–7]) to the 12-month period after study entry (median, 0.0 [0–4]; P = 0.038) (Box 4). Likewise, the median (range) functional severity index score for the total cohort was reduced from 9.0 (0–22) to 5.0 (0–24); (P < 0.01). The median (range) QoL scores for carers increased from 5.2 (1.9–7.0) to 6.8 (1.9–7.0); (P < 0.01), and the QoL scores for children increased slightly from 6.0 (2.6–7.0) to 6.9 (2.5–7.0); (P = 0.053). These results were similar when the intervention and control groups were analysed separately.