Increased longevity and improved health at older ages seen globally present significant challenges such as older adults having a higher prevalence of multimorbidity that often results in simultaneous treatment with multiple drugs (1). Polypharmacy (i.e., using five or more medications) (2) may increase the number of drug-related problems such as adverse effects and interactions, and is frequently associated with an increased use of potentially inappropriate medications (PIMs). Polypharmacy could have negative associations on long-term physical and cognitive functions (3–5).

Higher levels of physical activity (PA; occupational, commuting, leisure-time and household chores) have been significantly associated with a decreased risk of medicine use even after adjusting for sex, age, and economic status (6, 7). In a previous study, both medication use, and lack of PA were strongly associated with poor self-reported health status in older adults (8). More recently, sedentary behavior (SB; any waking activity in a sitting, reclining, or lying posture where energy expenditure is <1.5 metabolic equivalents) has been also considered an important determinant of health-related quality of life in older adults (9). Other studies conducted in a similar population showed that the likelihood of being categorized as highly sedentary increased with poor self-rated health, being overweight or obese, having a history of smoking, and multiple medication use (10, 11). An association between both sedentary time and cardiometabolic risk factors (such as metabolic syndrome and obesity) in older adults and between sedentary time and increased medication use have also been suggested (12).

However, evidence on whether a program aimed at reducing SB and increasing PA may help decrease medication use is yet to be established. This sub-study within the SITLESS Trial evaluated the relationship of an exercise referral scheme (ERS) enhanced by self-management strategies (SMS) aimed at reducing SB—SITLESS intervention- (13) and targeting community-dwelling older adults, on medication use compared to the ERS alone and usual care (UC) in a subsample of the population. Therefore, this study aims to understand how PA interventions in older adults can influence medication use. The main hypothesis established was that increased physical activity is associated with decreased medication needs. A secondary hypothesis was that the medication group that could potentially be reduced would be the medicines that affected the Central Nervous System.

Participants’ characteristics at baseline are presented in Table 1. The sample was composed mostly of older overweight adults (77% female). The waist to hip ratio was on average 0.91 (SD = 0.09) which indicates a high health risk. Participants presented in average 3.2 (SD = 2.1) comorbidities: 52% had high blood pressure (UC 61%; ERS 46%; ERS + SMS 50%), 50% had arthritis (UC 48%; ERS 58%; ERS + SMS 42%), 31% had obesity (UC 13%; ERS 39%; ERS + SMS 39%), 20% chronic pain (UC 17%; ERS 19%; ERS + SMS 23%), 19% diabetes (UC 4%; ERS 19%; ERS + SMS 31%), 15% depressive disorders (UC 26%; ERS 12%; ERS + SMS 8%) and 12% anxiety disorders (UC 9%; ERS 12%; ERS + SMS 15%). The average blood pressure for the sample was 134.4 (SD = 15.9)/80.0 (SD = 12.0). Regarding smoking habits, 10% were active smokers, 23% ex-smokers and 67% non-smokers. Smokers had been smoking for 30.6 years (SD = 17.7) with an average of 13.4 (SD = 7.9) cigarettes per day. The SPPB score indicates that the sample had one or more mobility limitations.

Differences in medicine use were observed in the entire sample and in all subgroups (Table 2) between baseline and after the intervention. However, significant reductions in medicine use were most visible and significant only in the ERS alone group or when collapsing this group with the ERS + SMS group.

The subgroup of patients that had a complete medication record (n = 20) including active ingredients and dosages was analyzed to better understand the type of changes that might have happened (Table 3). Most changes occurred in medicines that were used for acute situations (e.g., vitamins, eye drops or anti-inflammatories).

This study contributes to the evidence base on medication use impacts of interventions aimed at reducing SB and promoting PA in community-dwelling older adults. It was hypothesized that the SITLESS intervention would contribute to a change in medication use patterns. We could observe a reduction for the whole sample and for the ERS groups but could not fully explain all the causes for this reduction. The reduction of medicine burden in older adults can be one solution for the growing problem of polypharmacy, that could have negative associations on long-term physical and cognitive functions (3–5). Exercise interventions could provide a solution for better health outcomes while providing older adults with a longer life and lower medicine burden.

The sample included in this sub analysis showed poor values of body composition (BMI = 29; waist to hip ratio = 0.91; daily step count around 5,000 steps). Interventions targeting SB and PA can lead to improved body composition and might prevent the development of chronic diseases, leading ultimately to lower use of medication by older adults (14, 15). In order to explore this relationship in further detail, more studies should aim to collect movement behavior in different timepoints (preferably with accelerometry), body composition and medication outcomes, and also analyze the interaction between these variables.

The difference in the number of medicines at baseline and after intervention seems to follow a trend of reduction in overall number of medicines taken, as seen in other studies (6). Even though the reduction was seen in the overall sample and in the subgroup ERS alone, the ERS + SMS group did not show a significant reduction. However, when combining these two subgroups [ERS alone + (ERS + SMS)] the change became significant also, suggesting the inability to detect a significant change in all subgroups receiving the main intervention (ERS) could result mainly from the small sample size and not from the intervention per se. It is also reasonable to expect longer-term effects of the SMS sessions enhancing the effects of the ERS, thus only captured later. SMS had been shown to be effective in changing behavior (16), thus to evaluate the effects of the SITLESS intervention in medication use, particularly the enhanced approach using SMS, compared to ERS alone or UC we may need a longer follow up. This gains relevance when the average number at baseline indicates polypharmacy (> 5 medicines) in most groups, which follows a growing trend of polypharmacy in the last years (17). Future studies should aim for obtaining medication use data and whenever possible supplement patient-reported information with electronic health records (18, 19).

Due to the limited sample, we could not explore in detail the causes for the change in medicine use. A more robust sample would have allowed to use a regression model that could provide some clues on the reason for the change in medicine use. Medication management is a complex process that involves multiple healthcare professionals and the patient as the main player. Multiple challenges are known to arise when managing polypharmacy (e.g., transition of care, duplication of therapy, etc.) (20). Future studies could consider the continuum in patient centered care and whenever full integration of electronic records is not possible, increasing the number of contact points with the healthcare system, including with the local pharmacy. This could lead to more detailed and robust information to explore the effects of any intervention on medication patterns (21).

Lastly, the mixed-methods approach taken to explore qualitatively the medication classes most associated with decreases in the overall number of medicines can provide some clues on possible short-term effects of interventions targeting SB. We observed that medications used for acute conditions were those more frequently removed, including painkillers. However, to a lower extent, there was also removal of medications whose effectiveness is questionable (e.g., vitamins) or removal of potentially unsafe medications, including interactions (e.g., acenocumarol + aspirin) or potentially inappropriate medications (e.g., antidepressants with fall-inducing potential). Exercise interventions can support older adults in the prevention of falls and fall-related injuries (22, 23). Considering this, paying special attention to medicines that can affect balance or have a fall-inducing potential can have a protective effect on older adults and ultimately reduce fall risk and improve quality of life. In the current study, we could not fully explore changes in therapeutic subgroups, where changes could be expected, such as fall-risk increasing drugs (FRIDs, which include, e.g., drugs causing sedation, e.g., antipsychotics, drugs causing hypotension, e.g., loop diuretics, drugs with anticholinergic effects, e.g., tricyclic antidepressants, drugs causing bradycardia, e.g., antiarrhythmics, and other associated classes), mostly because of the limited sample size of the sub-analysis but also because of the pitfalls in recording of medication data. However, it has been shown that exercise is the single intervention demonstrated to significantly reduce the rate of falls in community-dwelling individuals (24), which is often associated to deprescribing following medication review in multifactorial interventions (25).

In general, the total quantitative difference might be low; however, qualitatively, the changes observed are aligned with pharmaceutical care principles, by focusing on necessity (e.g., pain-killers patient #11 and #4), effectiveness (e.g., supplements patient #8) and safety (e.g., CNS-acting medicine—patient #15 and #16; or interactions—patient #6). These findings suggest that for exploring the full potential of SB interventions on medication patterns, mixed methods are particularly useful as a change in number might not be clinically significant (26). The full potential of such analysis was however not reached as additional information on indication and duration of exposure would be needed for type 2 medication review. In such reviews, it would be possible to identify according to well established criteria, e.g., if removal of proton-pump inhibitors (patient #18) could also be justified by excessive duration. The benefits of type 3 medication review are even higher, but would also imply having additional information on comorbidities and laboratory values; these would allow judging if, e.g., removal of statins (patient #8 and #15) could be justified by effectiveness, as the most recent criteria consider there is no evidence to support the use of statins for primary prevention of cardiovascular disease in diabetes mellitus (27).

Many of the alterations in medication may be justified not only by the intervention but also by changes in the clinical condition, for which access to biochemical values would be valuable (e.g., to judge if the use of painkillers is done in the absence of tissue damage) and their absence constitutes one of the limitations of this study. A multidisciplinary approach to medication tracking using both data registries from physicians and pharmacists, and information from caregivers could add more accuracy to the reasons for changes in medication use. Another limitation of this study is the limited sample size with information on medicines used compared to the original sample (n = 446). This difference in sample size does not allow for a robust interpretation of the available data and only allows this study to provide an exploratory view of this topic. The reduction in the number of medicines could also be biased by the collection of data, as participants might not have recalled all the medicines after the intervention. A thorough data collection for medicines, including those as main variables of the study, could be a first step into increasing accuracy and more associations between exercise and medication use.

Although this retrospective post-hoc analysis showed possible relationships between lifestyle interventions and medication use, the hypothesis established could not be confirmed. We have identified that exercise interventions can have a modest impact in medication use (i.e., number) and identified some potential medicine groups that can be highlighted in future research (e.g., FRIDs, namely those causing sedation, causing hypotension, or with anticholinergic effects). More robust and exhaustive health information systems, strengthened by interprofessional efforts are needed to better capture the impact of lifestyle interventions on medication use patterns.

The data analyzed in this study is subject to the following licenses/restrictions: dataset owned by the original article owner. Requests to access these datasets should be directed to MG-G, mariagg@blanquerna.url.edu.

The studies involving humans were approved by the Ethics and Research Committee of Ramon Llull University (Fundació Blanquerna, Spain; reference number: 131400IP). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

RV, FAC, RM, and MG-G: conceptualization and writing — original draft preparation. RV, FAC, MG-G, MR-F, MaD, MiD, and EM: methodology. OS-N, JCM, DR, MaD, MiD, PC, MT, and MG-G: data collection. RV, FAC, and RM: formal analysis. RV, FAC, RM, MaD, EM, OS-N, JCM, DR, MiD, PC, MT, MR-F, and MG-G: writing — review and editing. FAC, RM, and MG-G: supervision. All authors contributed to the article and approved the submitted version.