As physicians, we take great interest in the adherence of our patients to prescribed medications. This is based on our belief that the benefits of treatment will be greater in patients who take their medication regularly and less among those who do not. C Everett Koop is quoted as having said: “Drugs don’t work in patients who don’t take them”. The practice of faithfully complying with a medication regimen, however, may be just as important as the efficacy of the medication itself. In this month’s issue of Thorax, Vestbo et al1 report that, among patients with chronic obstructive pulmonary disease (COPD) participating in the TORCH trial,2 adherers experienced significantly better survival and a lower risk of hospital admission due to exacerbations than non-adherers (see page 939). These effects, however, were independent of treatment group.

In fact, a growing body of literature suggests that adherence to drug therapy in clinical trials—both medication and placebo regimens—predicts better outcomes. Many of these data have been generated from cardiovascular trials.3 In 1980 a report from the Coronary Drug Project significantly challenged traditional scientific thought regarding the relationship between adherence and disease outcomes.4 This randomised double-blind placebo controlled trial examined the effects of clofibrate on the survival of men who had experienced a myocardial infarction.5 Five-year mortality was similar for those treated with medication (20%) and those treated with placebo (21%). When analysed by adherence patterns, however, good adherers (defined as those taking at least 80% of the prescribed medication) had significantly lower 5-year mortality than poor adherers (15% vs 25%). Surprisingly, however, these findings were nearly identical in the placebo group (15% vs 28% mortality). Initially, much discussion focused on the potential for bias in post hoc analyses. As more data have accumulated, however, the scientific community has begun to realise that, across disciplines and treatment assignments, adherent behaviour in clinical trials is associated with better outcomes.3

While it has been hypothesised that improved outcomes for adherers may be due to selection bias with good adherers having less severe disease, this is probably not the case.4 The improved outcome for adherers in the study by Vestbo et al remained after adjustment for disease severity as measured by forced expiratory volume in 1 s and dyspnoea measured by the MRC dyspnoea scale.1 Similar results have been reported in other trials.4 In fact, the Lung Health Study reported that better adherence was associated with more severe airways obstruction.6 Another hypothesis is that adherence to drug treatment may reflect a patient’s biological, social and psychological make-up which may all influence outcomes. Adherence may be a surrogate marker for other healthy behaviours such as taking medications for comorbid conditions, eating more healthily or taking safety precautions such as wearing seatbelts or sunscreen. Alternatively, however, the adherence itself and the expectation of treatment effect may activate behaviours that are then partly responsible for improved outcomes.7 For instance, the act of adhering to a treatment regimen may instil a sense of well-being and reduce anxiety about a chronic disease or may lead to altered health habits such as exercise or smoking cessation.

Very few data exist regarding adherence to inhaled medication in COPD. It was previously reported that only about 37% of patients with chronic lung disease are fully adherent with medical treatments in general.8 Long-term adherence with inhaled medications was reported from the Lung Health Study.6 Self-reported data suggested that nearly 70% of patients adhered to the medication regimen, dropping off slightly over the next 18 months. Adherence confirmed by canister weights, however, showed that only 48% of patients had good adherence at 1 year. Two other studies of nebulised treatments in COPD suggest that only about half of patients take their medications regularly.9 10 In the current study, 80% of subjects demonstrated good adherence (defined as adherence to study medications of >80% over the entire period the subject was in the study). While this is a good percentage within the context of a clinical trial, it probably does not reflect “real life” patient behaviour.

One would hope that the patients who need medication most might also be the group that is most adherent. It is therefore disappointing that, in the study by Vestbo et al,1 no association was found between disease severity defined by GOLD stage and adherence. In fact, poor adherers actually had higher MRC dyspnoea scores. However, it should also be noted that adherence in this study was determined by the percentage of returned drug left unused. What this analysis does not take into account is the significantly greater number of patients who dropped out of the placebo arm compared with the combination therapy arm (44.2% vs 34.1%), which also could be interpreted as non-adherence in the broadest sense of the word. These data suggest that adherence in this more global sense is probably less than the reported 80%. They also suggest that symptomatic improvement or drug efficacy may have had an impact on study participation even if the analysis of medication adherence among those who stayed in the trial did not differ between treatment and placebo arms.

In summary, this study indicates that adherence to inhaled therapy in COPD within the context of a clinical trial is good and associated with improved outcomes. Ultimately, however, if this is due to qualities intrinsic to the adherer, this actually leaves little room for the physician to modify outcomes other than to prescribe the treatment itself. If, however, the act of adhering actually activates other positive behaviours, then we as physicians should be even more aggressive about creating structured treatment regimens (simulating a clinical trial) for our patients. In a study evaluating metered dose inhaler compliance in COPD, use of an electronic medication monitor with feedback on the accuracy of medication use improved compliance.11 This type of methodology could be used to improve adherence in COPD outside a clinical trial. Clearly, further research is needed to better understand the factors driving the phenomenon of improved outcomes in adherers. While it is disappointing that neither treatment with active drug nor disease severity had an impact on adherence, study participation was better in the combination treatment arm of the TORCH study which suggests that symptomatic improvement may positively affect the likelihood of compliance of a patient with COPD in the more global sense. This would be good news for our patients as the current analysis also suggests that the association between adherence and mortality risk reduction with combination therapy was even greater when only those in the treatment arm were analysed.