The study population comprised adult patients, aged 18 to 70 years, with a clinical diagnosis of asthma, and fulfilled the criteria for difficult asthma (see Chapter 2.1.2). The inclusion criteria was set to be broad in order to ensure that the research study would be inclusive and open to as many patients as
possible, based on the fact that there are relatively fewer people with difficult asthma than mild to moderate asthma.
There is known to be a certain degree of diagnostic uncertainty with patients attending difficult asthma clinics, as many may have an alternate diagnosis (Barnes and Woolcock, 1998), or significant co-morbidities that contribute to their symptoms (Heaney et al., 2003). Therefore it was decided that all recruited patients must have a diagnosis of asthma confirmed by a consultant respiratory physician based on subjective and objective criteria, such as demonstration of airflow obstruction on spirometry (British Thoracic Society and Scottish
Intercollegiate Guidelines Network, 2014). Patients with other co-morbidities that could contribute to their respiratory symptoms were excluded because other respiratory conditions such as bronchiectasis or COPD may not respond to an intervention focussed upon asthma education and treatment.
It was considered that this research study should attempt to recruit patients of all severities from mild to severe asthma. However many asthma patients will
remain well controlled and require limited intervention from healthcare professionals, and so it may be unlikely that an interventional study would achieve significant impact in people with well-controlled asthma. Indeed, some community pharmacy studies have targeted patients with mild to moderate asthma and have failed to demonstrate a significant impact on asthma control after pharmacist interventions comprising education, inhaler technique training and smoking cessation advice (Mehuys et al., 2008).
Consequently, it was decided that it was most appropriate to target patients who are labelled as having ‘difficult asthma’, defined by the BTS/SIGN asthma guidelines as persistent symptoms and/or frequent exacerbations despite treatment at step 4 or step 5 (British Thoracic Society and Scottish
Intercollegiate Guidelines Network, 2014). It was thought appropriate to include only patients with difficult asthma despite previous studies that have failed to demonstrate a significant impact after intervention from community pharmacists in patients at high risk of asthma exacerbations (Charrois et al., 2006). This is because compliance with the intervention was poor in that study, and it is possible that the lack of specialist training, knowledge and skills of community pharmacists may have adversely affected the study.
A number of sampling methods to recruit patients were considered including selecting patients from within the difficult asthma clinic itself, during admission to hospital because of an exacerbation of asthma, or from GP practices. There is however no specific register of difficult asthma patients in primary or
secondary care from which to identify potential suitable patients for the study.
Additionally, advertising in the local newspaper for potential patients was not considered, as the cost of this was prohibitive.
Since the primary setting for the study was in the local difficult asthma clinic, recruitment from patients referred to the difficult asthma clinic had advantages over the other methods, as this provided an easily accessible cohort of patients thought to have difficult asthma. In addition to ease of recruitment, this would allow ease of monitoring and follow-up, and has been shown to have generally good response and retention rates within single clinic settings (Bowling, 2002).
Hospital admissions allowed another source for identifying suitable patients, but would require screening of all admissions to respiratory and medical admission wards to identify patients who potentially met the inclusion criteria. This was thought to be too labour intensive to be a practical method, and suitable patients tended to be referred to the difficult asthma clinic anyway, and
consequently the screening of hospital admissions could duplicate the effort to identify the same patients.
Screening of patients registered with GP practices was considered a method that could potentially highlight large numbers of patients that could be invited for the study, but it was decided that it was impossible to screen all asthma patients in each of the 112 GP practices within the three Leeds Clinical Commissioning Groups. A cluster sampling method (Bowling, 2002) selecting specific GP practices would have been required to screen for suitable patients, but this would have risked creating sampling errors if practices were not representative of the general patient population. In addition, some GP practices may have been better at managing asthma than others, so it could be harder to
demonstrate a positive response to the intervention in this scenario. Similarly, if the majority of patients are recruited from GP practices where asthma is not managed so well, there could be a potential for a greater impact of the intervention in the sample than would be expected for the whole population.
Consequently, it was decided that the most appropriate sampling method for the study was to identify patients from the hospital difficult asthma clinic.
4.3.1 Sample size
On the 7-point scale of the ACQ, a change in score of 0.5 is the smallest that can be considered clinically important. In the original validation study, the ACQ was able to detect a change in asthma control with a mean change (± standard deviation (SD)) of 0.73 ± 0.54 (Juniper et al., 1999b, Juniper et al., 2000).
A sample size calculation can be based on the ability to detect a difference of 0.5 in the mean ACQ score, assuming a SD of 0.54. Therefore for a power of 80% and a two-tailed significance level of 5%, the required study population is 38 patients. However, allowing for an attrition rate of 25%, observed in other
studies of Pharmacist intervention in asthma patients, the sample size required for this study is 52 patients.
This is based on a Comparison of means (unpaired data) sample size calculation:
Where:
The approximate number of patients per group The standard deviation
The difference in mean response between the intervention and control groups
A function based on the study power of 80% (=7.849)
Therefore:
Therefore = 19 patients per group.