Proactive asthma care in childhood: general practice based randomised controlled trial

Recruitment

Primary schoolchildren in the Australian Capital Territory received a health survey between February and April 2000. Children were eligible for recruitment if the returned questionnaire indicated that they had moderate to severe asthma and a usual general practitioner. Box 2 summarises how we determined moderate to severe asthma, defined by National Asthma Council criteria,4 from the survey responses. General practitioners had to have at least three children meeting these criteria to be included in the trial.

Summary of trial timeline and participant retention rates

• Download figure
• Open in new tab
• Download powerpoint

Randomisation

The returned surveys identified usual general practitioners and their practices. We enrolled only one general practitioner in each practice to avoid within practice contamination. We used SPSS version 10 to randomly assign 30 eligible practices to control and intervention groups. Within each practice, we ranked general practitioners according to the number of children with moderate to severe asthma and selected the highest ranking general practitioner from each practice. Twenty four general practitioners (12 in each group) agreed to participate. Of 225 children managed by these 24 general practitioners, 175 (78%) consented and attended the research clinic for baseline assessment at trial entry. One child withdrew immediately after this appointment, leaving 101 children in the intervention arm and 73 in the control arm. No temporal differences in the recruitment rates by study arm occurred.

Intervention

The chief investigator gave a one to one academic detailing session to all general practitioners. We asked control general practitioners to continue their usual paediatric asthma care and intervention general practitioners to administer care according to the 3+ plan. We instructed them in the plan and gave them a 3+ plan patient and practitioner resource kit. We prompted intervention general practitioners when a child's next 3+ visit was due. We offered a supplementary trial information evening separately to each general practitioner group. We classified a 3+ plan as fully complete if over the 12 month trial period the child attended their general practitioner for at least two asthma related visits, one of which was proactive, and if each of the two visits showed at least one different 3+ plan content item. We classified 3+ plans as partially complete if during the same period the child attended their general practitioner for at least two asthma related visits, one of which was proactive.

Allocation concealment and consent

We told participants that the study was examining the effects of reminder systems in proactive care and that some children would be reminded to return to their doctor for asthma care. We based this approach on ethical considerations for cluster randomised trials.7 General practitioners were aware that two study groups existed but were not informed of their allocation. We assessed concealment at trial end by asking general practitioners which arm of the trial they thought they were in. Six control general practitioners said the control group, and the rest did not know. Six intervention general practitioners said the control group, two said the intervention group, and three did not know.

Data collection

Children attended three assessment clinics during the trial. These were at trial entry, five to six months, and trial completion. We administered standard questionnaires, including items from the international study of asthma and allergies in childhood.8 The children then had spirometric tests (Fleisch-type electronic spirometer, Vitalograph alpha 2, model AL0642; Vitalograph, Buckingham, UK) before and after a standard cold air challenge test.9 10

We audited general practitioners' case notes for all patients at trial end to assess the number and type of asthma related visits made by participants to their nominated general practitioner during the trial period. For each participant, general practitioners also gave a verbal report of their written medical records of asthma related consultations, which two independent auditors classified into consultation types (table 1). One auditor was blind to the general practitioners' study group allocation, and the other was not. We used the data of the blinded auditor for all analyses. We carried forward six month pilot audit data for analysis for two of the initial general practitioners, one “control” and one “intervention,” who were unavailable for audit at trial end because of illness or interstate relocation.

Outcome measures

Study endpoints comprised process, clinical, and spirometric measures as shown in tables 2, 3, and 4. We obtained the process and clinical measures by using questionnaires completed by parents. Spirometric measures comprised the percentage reduction and absolute reduction in forced expiratory volume in one second (FEV₁) after cold air challenge.

Statistical methods

We estimated sample sizes by using the approach of Campbell et al,12 assuming that each general practitioner would have between three and seven children with asthma. For a power of 80% and a significance level of 0.05 we needed between 13 and 21 general practitioners (91 and 63 children) to detect a change from a baseline rate for a written asthma action plan of 30% to 70% at 12 months.

We compared the children in the intervention and control groups with respect to study outcomes by using two by two contingency tables with a χ² test or Fisher's exact test. Table 2 summarises the differences in baseline measures for trial outcomes. The total numbers differ slightly by variable in tables 3 and 4, because not all parents reported on all variables. We then used generalised estimating equations with fixed and random components with a logit link and a binomial error structure to provide odds ratios and associated confidence intervals for the effect of the intervention on the questionnaire outcomes examined.13 Inclusion of a term for the baseline value of each outcome ensured that the models were adjusted for any differences in baseline values that arose by chance through the randomisation process. We examined spirometric measures of lung function by using a generalised estimating equation regression model, adjusting for the intervention and baseline values as fixed effects and general practitioners as random effects. We also estimated the within general practitioner correlations for each outcome. They ranged from 0.16 (completion of 3+ plan) to −0.05 with a mean of −0.002. We used Stata 6 for intention to treat analyses.

Baseline demographics and participation rates

Table 2 summarises the baseline characteristics of the participating general practitioners and children. The two groups were similar, but the children in the intervention group had, on average, greater reductions in FEV₁ after cold air challenge (-2.2%) than control children (-0.7%). Four out of 69 (6%) control children and 13/92 (14%) intervention children had missing spirometric outcome data at the end of the trial (P = 0.2). Eight children were lost to follow up, 6/101 (6%) in the intervention group and 2/73 (3%) in the control group. Four of these children formally withdrew from the trial (all intervention)–one had moved interstate, and the remaining three failed to attend three scheduled visits and did not subsequently return mailed clinic questionnaires.

Outcome measures at 12 months

Table 3 shows that intervention children attended their general practitioners for asthma related consultations more often than control children, and for intervention children these consultations were significantly more likely to be proactive. Twenty eight per cent of the children in the intervention group fully completed the 3+ plan, significantly more than the 3% in the control group. An additional 20% of the intervention children partially completed the 3+ plan. The intervention group had increased use of written asthma action plans at 12 months. The number needed to treat (benefit) for one additional written asthma action plan was 5 (95% confidence interval 3 to 41) children.14 Neither symptom-free day scores nor the three components of this composite score (wheeze, night time symptoms, and restriction of activities) differed between groups (table 4). However intervention children had fewer occurrences of speech-limiting wheeze at trial end (table 3). Intervention children were more likely to use pressurised metered dose inhalers with a spacer and less likely to use relievers more than four days each week. Emergency department attendances tended to occur less often in intervention children, although the difference did not reach statistical significance. Spirometric measurements indicated that the intervention children were significantly less reactive to cold air than were control children (table 4). This persisted after further adjustment for clinic temperature and humidity and presence of an upper respiratory infection.

Study strengths and limitations

The findings are strengthened by the randomised controlled design, relatively unusual in studies looking at the organisation of healthcare delivery.17 Blinding is difficult to achieve in studies such as this,7 but less than half of general practitioners correctly identified their group allocation at trial end, indicating that the concealment approach used was effective. Retention rates were high. The study endpoints provided a balance between process, clinical, and spirometric data, and the study method was designed to approximate as closely as possible the “real world” of general practice. The intervention was not “forced.” General practitioners, parents, and children were free to make decisions about whether a visit actually took place. Even when a practice was prompted to remind a child to return, the decision to actually recall that child was made by the doctor. If a recall was made, parents had a free choice about whether to attend. The intention to treat analyses considered all intervention children as a group regardless of how much of the 3+ plan they were exposed to. After intervention general practitioners had one face to face academic detailing session and a dinner meeting discussing the 3+ plan, 52% of the intervention children received no aspects of the 3+ plan and 20% received only parts of it. This moderate uptake is consistent with the reported effectiveness of outreach visits and local opinion leaders as a means of influencing practice.18 The effect is important, however, given that 89% of the control children were not exposed to the 3+ plan at all.

Explanations for the limited uptake of the 3+ plan by the intervention general practitioners are speculative. They include framing the study question as general practitioners' recall rather than the 3+ plan in order to approximate blinding, the mild nature of some children's asthma being appropriately managed by less structured care, and the difficulty of implementing the 3+ plan approach for all patients despite its adoption by general practitioners. The last possibility is likely given that all but two intervention general practitioners completed the 3+ plan for at least one child in their care, yet the overall completion rate for the 3+ plan was low.

Two aspects of the study potentially reduced its power. Firstly, the study relied on survey results to identify children with moderate to severe asthma. Although we used criteria taken from existing guidelines,4 the children's spirometric assessments at trial entry produced mean changes in FEV₁ after cold air challenge of less than 3% in both arms, much less than the 9% identified by Nicolai et al as the 95th centile cut-off point between the reference population and children with excess bronchial hyperreactivity.9 Secondly, the children's asthma status changed over time: although all children satisfied the criteria in March 2000, the same questionnaires given at trial entry (May 2000), round 2 (October 2000), and round 3 (July 2001) identified 76%, 64%, and 53% of children as having moderate to severe asthma. Thus, this study is likely to have included children with mild disease who needed neither additional general practitioner visits nor completion of the 3+ plan for optimal management.

Measurement of process

Written asthma plans represent one facet of optimal asthma care and are a specific content item in the 3+ plan.2 In eastern Australia 21.7% of children reported having a written asthma action plan in 1993, an increase from 16.7% in 1990.19 Here, at trial end, 44% of intervention children had a written asthma action plan, less than the 70% used to estimate sample sizes. However, this is still an important outcome, with five children needing the intervention to provide one additional written asthma action plan.

Measurement of completion of the 3+ plan was complex and needed three criteria to be satisfied. Consultations had to occur in which asthma was considered; at least one of these consultations had to be proactive; and at least two different content items from the 3+ plan had to be covered. This approach led to the apparent paradox that the 3+ plan could be completed in two visits for the purposes of this study. The 3+ plan was designed for delivery over three or more consultations to allow for a usual consultation being about 10-15 minutes long, rather than because of evidence that three or more consultations were beneficial. General practitioners may have taken more time on occasions and delivered more than one consultation's “worth” of 3+ plan materials. The 3+ plan approach was nationally promoted during the course of the trial, and this could explain the fact that some control children received it. The idea of regular review is part of optimal asthma management, so allowing completion of the 3+ plan in a single consultation was inappropriate. Proactive consultations allowed time for observation of inhaler technique, review of a written asthma action plan, or development of a child's understanding of asthma. The measured clinical and spirometric changes reflect both full and partial exposure to the intervention. In keeping with trends in delivering primary care, the 3+ plan allows for content items to be delivered by nurses, asthma educators, or both, where such resources are available. The focus of this trial was on general practitioners undertaking proactive care, rather than on assessing the contributions made by different members of the primary healthcare team.

Several possibilities may explain the excess of reactive visits seen in the intervention group. These children may have had more symptoms as a result of chance. An imbalance between the control and intervention groups in terms of reactive care could have existed at the start of the trial. This could not be assessed, as we did no baseline audit of the general practice records. Parents of intervention children may have had a greater awareness of asthma symptoms, because of the proactive consultations, and therefore a lower threshold for presenting. The classification of consultation type is unlikely to have introduced bias, as the auditor was blind to trial group status.

Child related outcomes

Although most of the clinical outcomes showed beneficial effects in the intervention group, two important clinical indicators of childhood asthma control showed no change in the study–the symptom-free day scores and the number of days missed from school with asthma. For both these items, parents were asked to recall the information at the time of clinic visits. The approach chosen may have lacked sufficient sensitivity to detect true differences. We used the same method for speech-limiting wheeze and emergency department visits, but these dramatic events may be more likely to be accurately recalled. Alternatively, the intervention may have had a stronger impact on more severe asthma. The direction of change for all clinical measures favoured the intervention arm. Intervention children were less likely to use pressurised metered dose inhalers alone and more likely to use pressurised metered dose inhalers with a spacer device. Non-significant trends favouring the intervention children occurred: the daily use of preventer drugs increased, quality of life factors such as activity restriction because of wheeze improved,20 and the frequency of episodes of wheeze decreased. Part of the reduction in symptom scores and visits to the emergency department could plausibly result from improved adherence to preventer drug regimens using optimised delivery systems.

We chose cold air spirometry as the challenge test because it is acceptable to children and represents a “natural” wintertime exposure.9 21 The magnitude of the reduction in reactivity to cold air was small for intervention children. The spirometric test results have little clinical importance in isolation but give useful construct validity, providing reassurance that the effect of the intervention was not just dissuading parents from presenting to an emergency department or encouraging parents to under-report symptoms.

Conclusion

These results suggest that the 3+ visit plan combined with recall in children with asthma results in improved asthma management with a reduction in respiratory morbidity. Confirming the effect of the 3+ plan combined with recall on emergency department presentations and examining the effects of this approach in adolescent and adult populations are important future studies.