Participants representing a broad range of severity of asthma were recruited from two sources. The first source was adults who attended the outpatient specialist referral clinic of a large metropolitan teaching hospital. At their initial clinic visit, patients were asked to complete a questionnaire that collected information about their medical history, treatment and whether they consented to being contacted about research projects. The second source was a database of adults who had their name recorded on the research volunteer database at the Woolcock Institute of Medical Research. This database contains the names of adults who had participated in previous research studies and who agreed to be contacted in the future. It also contained the names of adults who had contacted the Institute in response to an advertisement or a media item and who had consented to be entered into the volunteer database.
Eligible participants were adults who self-reported both doctor diagnosed asthma and prescribed daily inhaled corticosteroid medication by pressurised metered dose inhaler (pMDI), and who agreed to participate in research and who were fluent in English. All participants provided informed written consent. The protocol was approved by the University of Sydney Human Ethics Committee (Ref 01/08/44).
Potential participants were contacted by telephone or mail and informed that we were conducting research investigating the way people manage their asthma. Participants were told they would be sent questionnaires and a device that measures the dose of inhaler medicine used. All participant questions seeking further information or clarification were answered honestly and completely. However, we did not offer any mention of adherence or compliance. Participants who agreed to participate and who met the study criteria were enrolled and mailed the study materials with instructions to wait for our next telephone call to connect the supplied device to their inhaler. Seven days after the recruitment telephone call participants were again telephoned to ensure the package containing the adherence questionnaire and electronic monitor had arrived and to assist them to attach the electronic monitor to their prescribed daily corticosteroid inhaler. At this time, they also provided information about their asthma and asthma management and returned these study materials via mail.
The analyses presented here represent a sub-study within a larger study of factors associated with non-adherence to prescribed daily inhaled corticosteroid medication. As part of the larger study, participants also completed questionnaires that measured anxiety, depression, asthma knowledge, personality, perceived involvement in care, optimism, and asthma severity.
Measurements of adherence
The Inhaler Adherence Questionnaire  is a 6-item Yes-No self-completed questionnaire that takes less than 1 min to complete. A yes is scored 0 and a no answer is scored 1 with a higher score indicating inhaler adherence.
The inhaler adherence electronic monitor we used was the DoserCT (Meditrack, Hudson, MA)  which attached to the canister of the corticosteroid pMDI and recorded the number of actuations per 24-h with sufficient memory for 45 days. The DoserCT has the added advantage of being able to operate in blind mode so that no actuation information is displayed by the device. Because the DoserCT was attached to the pMDI it did not matter whether the participant did or did not use a spacer. The Doser has been compared with the MDI Log (Medtrac Technologies; Lakewood, CO); the SmartMist (Aradigm Corporation; Hayward, CA) and the Nebulizer Chronolog (Medtrac Technologies; Lakewood, CO) and has been shown to be sufficiently accurate to monitor adherence [16, 17]. At enrolment participants recorded their prescribed inhaled corticosteroid regimen. During the 6 weeks of electronic monitoring participants made a diary entry only to record changes to their regimen either initiated by their doctor, as outlined in an asthma action plan or self-initiated. The DoserCT and diary were returned by mail.
Electronically monitored adherence definition
Although there is no consensus definition of adherence, the definition used in this study combined the measurement capability of the DoserCT with results reported from previous studies. Electronically monitored adherence was measured using DoserCT data from days 8 to 42, to exclude days when the DoserCT was in transit and to discard any enrolment or learning effect . Participants were classified as adherent if they used the prescribed number of actuations per day, or more, for at least 70% of days. The definition of adherence as 70% of days having used the prescribed or greater than prescribed dose of inhaled corticosteroid was based on previous research [18, 19] and our clinical experience. Pragmatically, selecting an adherence cut-off point of 70% allows for participants who regularly take their prescribed medication but miss the occasional dose, or who miss doses while waiting to get a prescription filled or who forget to take their medication with them when away from home to still be defined as adherent.
Participants in this study were drawn from a larger study assessing factors associated with non-adherence to prescribed daily inhaled corticosteroid medication. Assuming a non-adherence prevalence estimated of 50%, a minimum sample size of 40 participant were required to achieve a minimum power of 80% for detecting a change in the percentage value of sensitivity of a screening test from 0.50 to 0.80, based on a target significance level of 0.05. With the same assumptions, detecting a change in sensitivity of the screening test from 0.50 to 0.70 required a minimum sample size of 98 .
All data were analysed using the SAS statistical package for Windows . Spearman’s rank order correlation is a non-parametric test used to assess both the direction and the strength of association between two variables . The internal consistency of the IAQ was tested using Cronbach’s alpha . Cronbach’s alpha reflects how closely a set of items in a questionnaire are related and can be interpreted as a measure of scale reliability. The validity of the IAQ as a measure of adherence was quantified as sensitivity and specificity in relation to the objectively measured adherence using the electronic monitor. Sensitivity and specificity were measured each cut point on the IAQ score and the area under the curve (AUC) was estimated. The area under the curve provides information about the diagnostic accuracy of the test. An AUC value of 0 reflects a perfectly inaccurate test and a value of 1 indicates a perfectly accurate test .