The trial was undertaken during 2002 with approval from Grampian Research Ethics Committee.
Recruitment of practices
We invited, by letter, all 59 general practices in the Aberdeen, and Banff and Buchan regions of Grampian, Scotland to participate in the study. All non-responding practices were telephoned and a personal invitation issued. Participating practices gave their fully informed consent to taking part in all aspects of the development programme.
Randomisation
Using random number tables, consenting practices were centrally randomised to early or delayed intervention groups by a researcher not involved in the trial. It was not possible to blind the practices as the intervention was an active process. The researchers were aware of allocation, though the audit data were submitted to a different centre (Tayside Centre for General Practice, Dundee) to that of the research team organising the educational intervention in Aberdeen.
Procedure
Each enrolled practice undertook audits of their management of acute asthma at baseline, 6-months and 12-months using previously piloted methodology[20]. We provided full instructions on how to undertake the audits and a study helpline was available for support. The educational intervention included feedback of audit data and attendance at a workshop during which a practice development plan was formulated by participants. The intervention was provided immediately after the baseline audit to the early intervention group, and immediately after the 6-month audit to the delayed group.
The Acute Asthma Professional Development Programme
The baseline audit
Practices undertook a critical event analysis of acute attacks occurring over a 3-month period in adults and children aged 5 years and over. An attack was defined as "an acute deterioration of asthma for which the patient seeks urgent medical advice"[8]. Participants were advised to identify prospectively all acute attacks occurring during the audit period using computer databases, discharge letters, out-of-hours service slips, visit requests, prescriptions for courses of oral steroids and nebuliser use. Data about the management of the attacks was collected retrospectively by a member of the practice team from patients' written and computer records. The previously piloted critical event analysis form (figure 1) was designed to collect data on objective assessment of severity, treatment provided by health professionals and follow-up within 6 weeks post exacerbation[20]. Actions not recorded were assumed not to have been done.
The educational intervention
• Feedback of audit results: The results of the baseline audit were fed back to participating practices with anonymised comparative data from the other practices taking part in the programme, as well as the standards set by current asthma guidelines. Figure 2 provides an example of the feedback which was posted to practices prior to the workshop.
• Multi-disciplinary interactive workshop: A 3-hour workshop held in the University of Aberdeen, facilitated by two of the researchers (a general practitioner (HP) and respiratory nurse (GH)), was attended by representatives of the participating practice teams (normally a GP and practice nurse; the practice manager or senior receptionist was also invited). Amalgamated audit results from the group were discussed and used to tailor discussion towards specific aspects of acute asthma care identified as falling short of guideline standards. Case studies were used to facilitate discussion of practical aspects of acute asthma management highlighting key deficiencies reported in published literature[8]. Recording of objective assessment of attacks was emphasised.
• A list of suggested references (selected because of their relevance to primary care management of acute asthma), resources (e.g. guideline summary charts, web-sites of professional organisations, training organisations, equipment manufacturers) and practical materials (e.g. examples of asthma action plans) were provided to all participants and discussed in the workshops.
• Formulation of a development plan: Time was allocated at the end of the workshop for practice teams to formulate a practice-specific acute asthma development plan. Using an outline proforma, (Figure 3) the practice teams reflected on their current performance, identified aspects of care they wished to improve and made practical plans to overcome barriers and institute change.
Follow-up audits
The audits completed at 6-months and 12-months were undertaken using the same methodology as at baseline and were fed back to all practices. The 6-month audit provided a review of progress for the early intervention group, and baseline assessment for the delayed intervention group. The time scale of 6 months had proved to be feasible in our pilot study[20].
Outcome measures
Our primary outcome measure was the proportion of acute episodes with a recording of a peak flow compared to the patients' best (or predicted if best was not known) at the 6-month audit. This measurement is recommended by current asthma guidelines as a basis for the classification of severity of acute attacks and for determining appropriate emergency treatment[1]. Our pilot study demonstrated an improvement in the proportion of patients with a peak flow compared to best/predicted from the baseline prevalence of 21% to 61% at 6 months[20].
Other outcome measures, reflecting the recommendations of the guideline in force at the beginning of the trial,[21] from the critical event analysis were considered in three domains: assessment (recording of peak flow, 'peak flow compared to best/predicted', respiratory rate, heart rate, ability to speak), management (provision of oxygen, bronchodilation. systemic steroids actually administered, steroids prescribed, inhaled steroids, referral to hospital) and follow-up (provision of advice, follow up consultation, self-management education). Combined scores for each domain were calculated by summing the questions which were answered 'yes' to give a score out of 5, 6 or 3 for assessment, management and follow-up respectively.
Sample size calculation and statistical analysis
In order to take account of the clustering by practice, the sample size calculation needed to include an estimate of the intracluster correlation coefficient (ICC). Our pilot work suggested that we could expect three acute episodes per full time general practitioner during each three month period and we estimate an average of 5 general practitioners per practice. Using a conservative estimate for the ICC of 0.05, to detect a change of 30% in the proportion with a peak flow compared to best/predicted from a baseline prevalence of 21% with 80% power and 5% two-sided significance level would require 5 practices per arm (ie. an estimated 25 general practitioners and 75 clinical episodes in each group). To allow for withdrawals we decided to recruit at least 20 practices in total.
We analysed associations between categorical data using the chi-square test. Differences in normally distributed combined scores between early and delayed intervention groups were examined with independent sample t-tests.
Because randomisation was at practice level, the statistical analysis had to adjust for the effect of clustering. Multilevel modelling was inappropriate since the outcomes were practice level data across time and not multiple measurements across time on the same individuals. Therefore, the primary analysis was based on the practice summary measures. Analysis of covariance was used to examine differences in outcomes between groups at 6-months and 12-months after adjustment for baseline differences and practice effects. Practices who recorded no exacerbations during an audit cycle were excluded from the analysis at that time point.