Adherence to inhaled corticosteroids by asthmatic patients: measurement and modelling

Background Poor adherence to inhaled corticosteroids (ICS) is known as the main cause for therapeutic failure in asthma treatment and associated morbidity. To improve adherence, targetted and effective interventions need to be developed ideally based on using longitudinal follow-up of a large study cohort to establish patterns and influences on adherence. Objective To develop an annual measure of asthma patients’ adherence to ICS using primary care prescribing data over consecutive annual intervals, and to statistically model ICS adherence controlling for a range of patient factors. Setting A retrospective cohort study between 1997 and 2010 using United Kingdom general practice prescribing data on asthma patients aged between 12 and 65 years, without a diagnosis of chronic obstructive pulmonary disease. Method Patient’s ICS prescriptions are used to calculate the ‘number of days prescribed during calendar year’ divided by ‘number of days in the interval’ to form an annual prescription possession ratio (PPR) for each patient. Several definitions of PPR are considered and compared when calculating numerator and denominator. Adherence, measured by the preferred PPR, is then modelled to estimate the effect of asthma exacerbation, severity, control and other patient factors on adherence. Main outcome measure PPR, being a proxy measure for adherence. Results Annual PPR by all strategies gave a similar frequency profile. ICS were either over- or under-prescribed for over half of the follow-up time. Adherence was lower in younger patients, those newer to the study timeframe, those with less severe asthma, those with good control, with lower previous adherence, and who had not previously experienced an exacerbation. Conclusion The chosen PPR simulated clinical use of ICS most closely; including overlapping days, excess days passed to the next interval, considering gaps in the denominator, with censoring at 100 %. The PPR is a useful measure for signalling or measuring adherence changes over time. The modelling results identified many characteristics which would indicate which asthma patients and at what points in their treatment cycle they would be at increased risk of low adherence.