We contribute to the literature in 3 ways: (1) We show that medication adherence – measured by the PDC – varies substantially among COPD patients; (2) high medication adherence potentially halves the risk of hospitalized exacerbations; and (3) utilizing real-world (insurance) data yields accurate figures for medication adherence.

We grouped the PDC of COPD patients into five categories. With regard to the distribution, categories 1 and 2 include 6280 COPD patients (48% of total sample). Thus, a large number of COPD patients have a PDC of less than 40%. This finding is critical because it implies that almost half of COPD patients fail to take their medication on four out of seven days. Lastly, almost one fourth of our sample falls into category 5. This finding supports the literature as in clinical practice only around 30% of COPD patients are found to be adherent to their medication prescriptions12.

When comparing the characteristics of the patients grouped within the category, especially the share of hospitalized exacerbations, prescription of short-acting medication, and the number of premium reductions are striking. The lowest rate of exacerbations is present in category 5 with 225 out of 3081 (7%) COPD patients, whereas the first category includes the highest rate of exacerbations with 619 out of 2944 (21%) COPD patients. This relation states a first indication for an association of PDC and hospitalized exacerbations. Additionally, the rate of short-acting medication is highest in category 1 with 28%. Comparing this share with categories 4 and 5, the prescription rate is clearly lower with only 12% in category 4 and 19% in category 5. Short-acting medications are prescribed in case of an acute worsening of the disease symptoms to prevent or treat an exacerbation2. Therefore, the share of prescribed short-acting medication further supports the assumption that COPD patients in category 1, i.e., with low medication adherence, are more likely to suffer from exacerbations.

Furthermore, the share of premium reduction recipients among the COPD patients is remarkably high. Premium reductions are granted to people living in modest economic circumstances, where the family and financial situation is decisive for receiving premium reductions24. In our sample, around 39% to almost 43% of all COPD patients – depending on the corresponding category – receive premium reduction. Compared to the average Swiss population, approximately 28% of all insured persons received a premium reduction in 202025. The higher share of premium reduction recepients is a sign for lower socioeconomic status among COPD patients – as already highlighted by Eisner et al. 5.

In addition, potential influencing factors on hospitalized exacerbations were analyzed. First, higher medication adherence lowers the risk for acute exacerbations, which is in line with the literature21,22,23,26. Our analysis shows that the risk for hospitalized exacerbations can be decreased ~51% in case of high adherence (category 5) compared to non- respectively low adherence (category 1). However, also COPD patients in categories 2 to 4 benefit from lower risks for exacerbations (between 28%-48% less likelihood) compared to COPD patients belonging to category 1. Similar results were also retrieved in RCTs where, for example, Vestbo et al. 11. showed that adherence to inhaled medication is significantly associated with reduced risk of death and admission to hospital due to exacerbations in COPD. Furthermore, Torres-Robles et al. 27. found that medication adherence improves clinical outcomes in COPD patients and that interventions on increasing medication adherence show positive results. These insights shows that COPD patients would already benefit if general practitioners, outpatient specialists, and other medical professionals could manage supporting patients in moderately increasing their medication adherence. Potential approaches for increasing COPD patients’ medication adherence are individualized educational interventions28,29, reminders, motivational strategies, shared decision making or direct feedback on medication use29. These approaches could also be incorporated into digital health interventions (for example, Spielmanns et al.30).

Medication adherence is categorized into three main stages: initiation, implementation, and discontinuation31. The focus of this study is on the intersection of the two latter stages – implementation and discontinuation – by calculating the PDC. Although most COPD patients do not take their medications as prescribed (i.e., no full implementation), they do also not completely quit taking their medications (i.e., no full discontinutation). This inconsistent behavior of COPD patients leads to the assumption that they face certain challenges in their adherence habit such as personal beliefs about the effectiveness of the prescribed medication, patient-physician relationship, or insecurities in the right usage of the medication14,15,16,17. Rising awareness of these challenges in medication adherence plus knowing the factors (such as higher age, receiving premium reduction, or prior exacerbations) being significantly associated to a higher risk of exacerbations, health policy initiatives should sustainably strengthen COPD patients’ medication adherence, e.g. when implementing chronic care programs and/or digital health assistants.

The limitations of this study are twofold: firstly, due to the missing patient diagnosis in the dataset, the number of COPD patients included in the dataset are approximated with two inclusion criteria: taking prescribed long-acting medication and age over 40 years. This means that, on the one hand, we might underestimate the number of COPD patients as patients might not take any medication yet or they stopped taking medication. On the other hand, we might include asthma patients in our sample because there might be asthma patients older than 40 years taking long-acting prescriptions. Both issues should be marginal, however, and not strongly influence our results. Moreover, the advantage of claims data enabling us to observe patients intersectorally is more essential than the limitation of potentially underestimating the number of COPD patients or including single patients with a different diagnosis.

Secondly, due to the empirical setting of this study, the number of observed variables is limited. Two areas with potentially unobserved control variables include the COPD patient’s personal motivation and financial situation. Personal motivation is difficult to quantitatively replicate based on the used dataset but may have an impact on both medication reserve and the likelihood of exacerbation. Furthermore, high personal motivation might positively influence COPD patients’ health literacy and their body awareness. These two factors potentially reduce the exacerbation risk. However, in our study, we only focus on the association of the PDC with exacerbation likelihood. Therefore, we might overestimate this effect as we neglect the association of motivation with exacerbation likelihood. Additionally, the financial situation may influence the utilization of healthcare services, as there is some cost sharing in the Swiss insurance system between insured persons and insurances. We aimed to consider this by including premium reduction as control variable. However, it is only a proxy, since the variable premium reduction only adjusts for lower income groups. If there is no premium reduction, the exact income level is still unknown.

For future research, a better understanding of COPD patients’ needs on potential supporting mechanisms to promote higher medication adherence has to be established. Through a deepened understanding of current needs or challenges faced by COPD patients, researchers may be able to develop supportive tools to empower COPD patients in their daily lives to deal with the challenges of COPD and to increase their medication adherence. On the one hand, the risk of hospitalized exacerbations can be decreased, whereas on the other hand, the COPD patient’s health-related quality of life should be conserved through a potential slowdown of the progression of the disease.

To conclude, medication adherence according to prescription halves the risk for hospitalized exacerbations. Generally, medication adherence is low meaning that there is vast potential to improve patient-relevant outcomes without “inventing” a new treatment but by supporting COPD patients. There are two main levers to do this: (1) Strengthening health literacy so that patients understand the importance of regular medication intake and (2) structured, physician-led support programs such as disease management programs or chronic care programs. The first lever is patient-focused whereas the second physician-focused, yet both can be supported digitally. Additionally, COPD patients’ support needs should be further analyzed and focused on in future research, especially in the context of digital support tools. Future policy changes in Switzerland and elswhere should strengthen the incentive structures for physicians and COPD patients to use disease management programs and to finance digital health interventions.