Polypharmacy, Potential Drug–Drug Interactions and Medication Non-Adherence in Patients with Multiple Sclerosis: A Longitudinal Study

Abstract

Introduction

Multiple sclerosis (MS) is a chronic neuroinflammatory disease affecting approximately 2.9 million people worldwide. Disease-modifying therapies for MS effectively lower the risk of relapses and delay disability progression, but the increasing medication burden and ongoing adherence challenges complicate disease management. An improved understanding of predictors of medication-related risks is essential to optimize long-term safety, treatment effectiveness, and quality of life in patients with MS.

Methods

In this longitudinal observational study, 206 adults with MS or clinically isolated syndrome were enrolled, of whom 175 completed the 5-year follow-up assessment. Sociodemographic, clinical, and comprehensive medication data were collected at baseline and follow-up through structured interviews and review of medical records. Polypharmacy was defined as the concurrent use of ≥ 5 medications. Potential drug–drug interactions (pDDIs) were systematically identified using the DrugBank database, and medication non-adherence was defined based on patient self-reported missed medication.

Results

Over the 5-year follow-up period, the prevalence of polypharmacy increased from 53.1% to 62.3% (p = 0.024), and pDDI exposure rose from 67.4% to 81.1% (p < 0.001), mainly driven by the greater use of drugs for comorbid conditions and dietary supplements. In contrast, monthly medication non-adherence remained stable (25.7% to 27.3%, p = 0.855). Major interactions accounted for 7.1% of all identified pDDIs. Older age, disability pension status, higher disability levels, coexisting medical conditions, and lower educational attainment were associated with polypharmacy and the presence of pDDIs, whereas non-adherence was linked to prior non-adherent behavior and inpatient care at baseline.

Conclusion

Over time, the prevalence of polypharmacy and pDDIs increased in patients with MS, whereas medication non-adherence emerged as a largely independent risk domain. Polypharmacy and the presence of pDDIs were mainly associated with aging- and disability-related factors, whereas medication non-adherence was more difficult to predict. Our findings emphasize the need for regular medication monitoring that considers both prescribed and non-prescribed drugs, alongside individualized adherence support to mitigate distinct medication-related risks in long-term MS care.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40120-026-00987-z.

Key Summary Points

Why carry out this study?

Patients with multiple sclerosis (MS) often use multiple long-term medications, including disease-modifying therapies, symptomatic treatments, medications for comorbidities, and self-medication.

Increasing medication burden is associated with polypharmacy, potential drug–drug interactions (pDDIs), and challenges in medication adherence.

This longitudinal study assessed changes in these medication-related risks over 5 years in 175 patients with MS.

What was learned from the study?

The prevalence of polypharmacy and pDDIs increased significantly over 5 years, whereas medication non-adherence remained stable.

Regular medication review and individualized adherence support are essential to address distinct medication-related risks in patients with MS.

Introduction

Multiple sclerosis (MS) is a chronic neuroinflammatory disease of the central nervous system (CNS) and a leading cause of non-traumatic neurological disability in young adults [1]. It affects approximately 2.9 million people worldwide [2]. MS is characterized by highly heterogeneous clinical manifestations, with symptoms depending on the location and extent of demyelinating lesions within the CNS [3]. Emerging data indicate that MS involves not only focal inflammation but also chronic low-grade inflammation and neuronal dysregulation linked to smoldering disease activity [4]. Clinically, MS is classified into three phenotypes: relapsing–remitting MS (RRMS), primary progressive MS (PPMS), and secondary progressive MS (SPMS). While RRMS involves episodes of neurological dysfunction (i.e., relapses) with partial or complete recovery, the progressive forms are defined by gradual disability accumulation [3]. Recent work suggests that clinically isolated syndrome (CIS) and early MS represent stages along a disease continuum rather than distinct entities, which has important implications for diagnosis and early treatment decisions [5]. The latest revisions to the diagnostic criteria have incorporated the optic nerve and the kappa free-light chain index [6, 7].

Disease-modifying therapies (DMTs) for MS modulate or suppress immune activity to reduce relapse frequency and slow disability progression, thereby improving the quality of life of patients with MS. DMTs vary widely in their mechanisms of action, clinical efficacy, safety profiles, and routes of administration [1, 8, 9]. High-efficacy DMTs such as B cell-depleting therapies are increasingly used early in the disease course [10]. In older adults, treatment decisions require balancing efficacy against immunosenescence and comorbidity-related risks, underscoring the importance of long-term safety and contributing to increasing care complexity [11, 12]. The clinical management of patients with MS usually involves symptomatic treatments to address a range of manifestations, including pain, spasticity, bladder and bowel dysfunction, fatigue, cognitive impairment, mood disorders, and gait impairment [3, 13]. Further pharmacological interventions encompass acute relapse treatment and the treatment of comorbidities [14]. Additionally, patients with MS frequently use self-medication, including over-the-counter (OTC) drugs and dietary supplements [15, 16].

Consequently, polypharmacy has become a central aspect of the experience of patients with MS [17]. Although definitions vary widely [18], the concurrent use of five or more medications is most commonly applied in adult populations [19]. Between 14% and 76.5% of patients with MS meet the criteria for polypharmacy [17, 20], with estimates varying depending on the study population and methodology, including whether only prescribed (Rx) medications or all medications are considered. Recent studies have demonstrated a rising prevalence of polypharmacy in MS over the past years [21, 22]. Several factors have been associated with an increased likelihood of polypharmacy in MS, including older age, greater disability, a higher number of comorbidities, and lower educational attainment [21, 23, 24]. In patients with MS, polypharmacy has been linked to a higher risk of adverse drug reactions [25], increased risk of future hospitalizations [26], greater healthcare utilization [27], cognitive deficits [28], and worse health-related quality of life [29]. Importantly, polypharmacy increases the risk of potential drug–drug interactions (pDDIs) and may compromise adherence to medication regimens.

Drug interactions may arise through pharmacokinetic or pharmacodynamic mechanisms. Pharmacokinetic interactions involve alterations in drug absorption, distribution, metabolism, or excretion, whereas pharmacodynamic interactions occur when the effect of one drug is modified by another, resulting in enhanced or diminished therapeutic efficacy and/or side effects [30]. Given the substantial medication burden in MS, the risk of pDDIs is considerable. In our previous studies, approximately 65–70% of patients with MS had at least one pDDI in their medication profile [31–34], with prevalence varying depending on the pDDI database used. Severe pDDIs constitute only a relatively small proportion of all identified interactions. However, polypharmacy was associated with an approximately tenfold higher risk of having a severe pDDI (17.7% vs. 1.7%) [32]. Drug interactions may result in unintended drug exposure and adverse therapeutic effects, which may in turn necessitate further pharmacotherapy. Accordingly, the identification of pDDIs in patients with MS is essential for risk assessment and treatment optimization, thereby enhancing safety and improving clinical outcomes.

Non-adherence to treatment remains a clinically relevant challenge in patients with MS, with reported rates varying widely depending on the study population and adherence definition applied. Notably, most studies focus exclusively on adherence to DMTs rather than overall medication adherence. Published adherence rates to DMTs in patients with MS range from 27.0% to 93.8% according to a scoping review [35]. A Canadian research group reported rates of optimal adherence to DMTs of approximately 75–80% [26, 36, 37]. These rates were higher than those observed for medications used to treat other chronic conditions, such as ACE inhibitors (54.7%) [37]. Factors associated with non-adherence include younger age, lower perceived treatment efficacy, adverse effects, and treatment complexity [35, 38, 39]. Adherence in MS is further influenced by disease severity and personality traits [40]. Non-adherence has been linked to increased relapse risk and higher healthcare utilization [41]. Medication adherence is therefore a key determinant of long-term treatment outcomes in MS, highlighting the need for individualized adherence support strategies in routine clinical care [42].

Polypharmacy, pDDIs, and medication non-adherence are increasingly recognized as interconnected challenges in MS care. However, these aspects have typically been studied in isolation. Moreover, existing studies have largely relied on cross-sectional designs [20], limiting insights into how these factors evolve over time. In particular, longitudinal analyses of how the risk of pDDIs changes over time in patients with MS have not been reported. Potential predictors of changes in polypharmacy status, pDDI exposure, and adherence patterns over time also remain to be explored. Consequently, as the MS population ages [43] and medication complexity increases, there is a need for real-world data to better understand the temporal dynamics and shared determinants of medication use, pDDI risk, and adherence in MS.

In this study, we aimed to investigate longitudinal changes in polypharmacy, pDDI exposure, and medication non-adherence in patients with MS. In addition, we sought to identify sociodemographic and clinical factors associated with these medication-related risks and to examine predictors of their changes over time. By integrating these domains, our study aims to provide a more comprehensive understanding of how medication use patterns and treatment-related risks evolve alongside disease progression, thereby informing personalized care strategies in MS.

Methods

Patient Enrollment and Inclusion Criteria

This single-center, longitudinal, observational cohort study was conducted between August 2019 and August 2025 at the MS outpatient clinic of the Ecumenical Hainich Hospital in Mühlhausen, Germany. The hospital provides dedicated outpatient and inpatient care.

Inclusion criteria were age ≥ 18 years, a CIS or a confirmed diagnosis of MS according to the revised McDonald criteria from 2017 [44], cognitive ability to participate in the study, and willingness to participate.

The study comprised two assessment time points:

1. Baseline: August 2019–June 2020

2. Follow-up: October 2024–August 2025

Ethical approval was obtained from the ethics committee of the University of Rostock (permit number A 2019-0048) and the Medical Chamber of Thuringia (reference number 23117/2019/88). The study was conducted in accordance with the Declaration of Helsinki and the European General Data Protection Regulation. All participants provided written informed consent prior to inclusion.

Clinical and Sociodemographic Data Collection

A multimodal strategy was employed to maximize participation. Accordingly, the patients were contacted using the method most appropriate to their circumstances, either by postal questionnaire, structured telephone interview, or face-to-face interview conducted at the hospital or during a home visit. The data were collected using standardized questionnaires and through a review of available medical records (Fig. 1).

Fig. 1.

Overview of the longitudinal study design. The study was conducted at the Ecumenic Hainich Hospital in Mühlhausen (Germany). A total of 206 patients with MS or a CIS were included at baseline, and 175 patients participated in the follow-up assessment approximately 5 years later. At both time points, data were collected through in-person interviews, telephone interviews, postal questionnaires, and review of medical records, with a focus on current medication use and medication adherence. The reasons for dropouts (n = 31) are indicated in the diagram. Three main domains were investigated: A, polypharmacy (purple); B, presence of pDDIs (blue); and C, medication non-adherence (orange). The analysis of pDDIs was conducted using the DrugBank database [47]. Finally, the baseline and follow-up data were compared to assess temporal changes in patient medication status. CIS clinically isolated syndrome, MS multiple sclerosis, pDDI potential drug–drug interactions

Sociodemographic variables included age, sex, educational level, occupational status, pension status, partnership status, and smoking status. Among the clinical data collected were disease duration, type of care received, and disease course, categorized as CIS, RRMS, SPMS, or PPMS [45]. The patients’ degree of disability was assessed using the Expanded Disability Status Scale (EDSS), which ranges from 0 (normal neurological examination) to 10 (death due to MS) [46]. Coexisting medical conditions, including MS-related symptoms and non-MS-related diagnoses, were assessed based on patient history and medical records.

Medication Data Collection and Outcomes

At both study time points (baseline and follow-up), the complete medication profile of each patient was recorded, including both prescribed therapies and self-medication. All medications were classified according to three dimensions: (1) therapeutic objective, (2) prescription status, and (3) dosing schedule. With regard to therapeutic objective, the medications were categorized as DMTs, drugs used to treat MS-related symptoms, or drugs used to treat comorbidities. According to prescription status, medications were classified as Rx drugs or OTC medications. OTC medications were defined as drugs that can be obtained without a physician’s prescription and included analgesics as well as vitamins, minerals, and other dietary supplements. Finally, medications were classified according to dosing schedule as long-term or on-demand drugs. Long-term medications were defined as drugs intended for continuous use over extended periods, whereas on-demand medications were taken intermittently or only when needed for acute symptoms.

Polypharmacy was defined as the concurrent use of ≥ 5 medications in accordance with the most commonly used definition [18]. Rx polypharmacy was defined as the concurrent use of ≥ 5 prescribed medications excluding OTC medications.

Potential drug interactions were identified using the DrugBank Drug Interaction Checker (version 5.1.13, released January 2025) [47, 48]. The interaction dataset in DrugBank is derived from drug labels and primary literature and curated by the DrugBank team [47]. For each patient, the complete medication profile was systematically screened for pairwise pDDIs. The medications were assigned to their respective active ingredients prior to analysis to ensure compatibility with the DrugBank database. For combination products, each active ingredient was entered separately into the interaction checker. Identified interactions were categorized according to the DrugBank severity classification as minor, moderate, or major. Minor interactions are considered to have limited clinical relevance. Moderate interactions may or may not result in substantial changes for a patient. Major interactions should prompt consideration of additional monitoring or treatment modification by a healthcare professional [47]. The patients were then classified in two ways. First, according to the presence or absence of at least one pDDI of any severity in the medication profile. Second, according to the presence or absence of at least one major pDDI.

Medication adherence was assessed through structured patient interviews conducted at baseline and follow-up. The patients with MS were asked how often they miss or intentionally omit medication on a weekly and monthly basis. Non-adherence was defined as missing at least one scheduled medication and classified as monthly non-adherence (≥ 1 missed medication per month) or weekly non-adherence (≥ 1 missed medication per week). No distinction was made regarding which specific medication was missed or whether doses were omitted intentionally or unintentionally. Participants who indicated missing at least one medication were additionally asked about the reasons for not taking their medication and encouraged to provide one or more reasons. Responses were subsequently grouped into four categories: forgetting to take the medication, experiencing side effects, considering the medication unnecessary, or the medication not being available at the time of intake. The analyses of adherence were restricted to patients who were taking at least one medication and who provided a valid response to the adherence question.

Statistical Analysis

The data were entered and organized in Microsoft Excel (Microsoft 365, Version 2511). All statistical analyses were performed using the R statistical software environment (version 4.5.1) and cross-checked using jamovi (version 2.7.9), JASP (version 0.95.4), and PSPP (version 2.0.1).

Descriptive statistics were used to summarize patient characteristics and medication-related variables at baseline and follow-up. Continuous variables are presented as means and standard deviations, whereas categorical variables are reported as counts and percentages. When missing values occurred, percentages were calculated based on valid (non-missing) observations.

Longitudinal within-subject changes between baseline and follow-up were assessed using paired statistical tests. Paired t tests were applied for continuous variables under the assumption of approximate normality, whereas paired categorical data were analyzed using McNemar’s chi-squared test. All statistical tests were two-tailed, and p values < 0.05 were considered statistically significant. When multiple related hypotheses were tested, p values were adjusted using the false discovery rate (FDR) method [49].

Potential predictors of medication-related outcomes at follow-up were first identified using univariable binary logistic regression. Sociodemographic, clinical, and medication data collected at baseline were considered as candidate predictors. The main outcomes at follow-up were polypharmacy based on the total number of medications taken, presence of pDDIs of any severity, and medication non-adherence per month. In addition, transition outcomes between baseline (2019/20) and follow-up (2024/25) were analyzed. These included transition from no polypharmacy to polypharmacy, from absence to presence of pDDIs, and from medication adherence to non-adherence. For the transition analyses, only patients without the respective condition at baseline were included. These patients were then compared according to whether they developed the condition by follow-up or remained without it. Additional analyses were conducted using stricter outcome definitions, namely Rx polypharmacy, presence of at least one major pDDI, and weekly medication non-adherence, considering both outcome status at follow-up and corresponding transitions from baseline. Univariable logistic regression models were fitted for each baseline variable and each outcome. When zero cell counts in contingency tables indicated complete or quasi-complete separation, Firth’s bias-reduced logistic regression was applied to avoid overestimation of effect sizes. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs). Corresponding p values were calculated and adjusted for multiple testing using the FDR method [49].

Additionally, multivariable binary logistic regression models were constructed to identify independent predictors of each outcome while limiting overfitting. The sociodemographic and clinical variables at baseline were considered as candidate predictors for this analysis. Feature selection was performed using a bootstrap-based stability selection procedure with elastic-net regularization [50, 51]. The elastic net combines two types of penalization controlled by the parameters α and λ. The mixing parameter was set to α = 0.5, corresponding to a balanced elastic net with equal contributions of the least absolute shrinkage and selection operator and ridge penalties. The penalty parameter λ was estimated using tenfold cross-validation, and the largest λ whose cross-validated deviance remained within one standard error of the minimum was selected (λ_1se). This approach favors parsimonious models. To evaluate the stability of predictor selection, 1000 bootstrap samples were generated by resampling the study population with replacement. Within each bootstrap sample, an elastic-net logistic regression model was fitted and predictors with non-zero coefficients at the selected λ were recorded. For each predictor, the selection frequency across bootstrap models was calculated. Predictors selected in at least 50% of bootstrap samples were retained and entered into a final unpenalized multivariable logistic regression model. Model performance was evaluated using classification accuracy based on predicted probabilities with a threshold of 0.5. Stability selection results were visualized using lollipop charts showing the selection frequency of candidate predictors across bootstrap samples. Final regression results were displayed as forest plots showing ORs and 95% CIs with corresponding p values.

Results

Baseline Characteristics of the MS Study Cohort

At baseline, a total of 206 patients with MS participated. Of the initial cohort, 31 patients were lost to follow-up for the following reasons: declined to participate (n = 16), deceased during the period (n = 8), could not be contacted (n = 4), and became ineligible (n = 3). Consequently, 175 patients (85.0%) completed the follow-up assessment and constituted the final study cohort (Fig. 1).

At study entry (2019/20), the cohort had a mean age of 45.8 ± 12.7 years, and most participants were women (70.3%) (Table 1). The average schooling duration was 10.2 ± 1.0 years, and most participants (69.7%) were trained as skilled workers. Notably, only 25.1% of the study population were employed full-time. Nearly half of the participants (46.9%) were not in the workforce, and 31.4% received a disability pension. Clinically, most patients had a CIS (n = 12) or RRMS disease course (n = 118), which were grouped together as CIS/RRMS (74.3%). Patients with progressive disease courses were older on average (CIS/RRMS 42.6 ± 11.6 years, SPMS 54.0 ± 12.0 years, PPMS 58.5 ± 6.8 years). The patients had a mean EDSS score of 3.2 ± 2.4 at a mean disease duration of 10.1 ± 8.7 years. They were predominantly managed in the outpatient setting (95.4%). Coexisting medical conditions were common, affecting 78.9% of the cohort.

Table 1.

Characteristics of the patients with MS (n = 175) at baseline (2019/2020)

	Statistics
Sociodemographics
Age (years), mean ± SD	45.8 ± 12.7
Sex, n (%)
Female	123 (70.3)
Male	52 (29.7)
Schooling (years), mean ± SD	10.2 ± 1.0
Educational level, n (%)
No training	9 (5.1)
Skilled worker	122 (69.7)
Technical college	28 (16.0)
University	16 (9.1)
Occupational status, n (%)
Not in workforce	82 (46.9)
Part-time work	49 (28.0)
Full-time work	44 (25.1)
Pension status, n (%)
Not receiving pension	106 (60.6)
Old-age pension	14 (8.0)
Disability pension	55 (31.4)
Living in a partnership, n (%)
No	35 (20.0)
Yes	140 (80.0)
Smoking status, n (%)
Never smoker	55 (31.4)
Former smoker	52 (29.7)
Current smoker	68 (38.9)
Clinical characteristics
Disease course, n (%)
CIS/RRMS	130 (74.3)
SPMS	34 (19.4)
PPMS	11 (6.3)
EDSS score, mean ± SD	3.2 ± 2.4
Disease duration (years), mean ± SD	10.1 ± 8.7
Patient care, n (%)
Outpatient	167 (95.4)
Inpatient	8 (4.6)
Coexisting medical condition, n (%)
No	37 (21.1)
Yes	138 (78.9)

CIS clinically isolated syndrome, EDSS Expanded Disability Status Scale, MS multiple sclerosis, PPMS primary progressive multiple sclerosis, RRMS relapsing–remitting multiple sclerosis, SD standard deviation, SPMS secondary progressive multiple sclerosis

Changes in Polypharmacy, pDDI Exposure, and Medication Non-Adherence Over Time

From baseline (2019/20) to follow-up (2024/25), a significant increase in the prevalence of polypharmacy was observed (Fig. 2). The proportion of patients with MS who had polypharmacy, defined based on all medications, increased from 53.1% to 62.3% (p = 0.024). Similarly, the proportion of patients with Rx polypharmacy, reflecting the use of prescribed drugs only, rose from 36.6% at baseline to 48.0% at follow-up (p = 0.004).

Fig. 2.

Proportion of patients with MS with polypharmacy, at least one pDDI, or medication non-adherence at the two time points. The stacked bar charts show the percentage of patients with polypharmacy, pDDIs, and medication non-adherence at baseline (2019/20) and follow-up (2024/25). a Prevalence of polypharmacy (≥ 5 medications), with lighter segments indicating polypharmacy based on all medications (including over-the-counter drugs) and darker segments indicating Rx polypharmacy. b Proportion of patients with one or more pDDIs, with lighter segments representing patients with ≥ 1 pDDI across all severity levels and darker segments representing patients with ≥ 1 pDDI of major severity. c Frequency of self-reported non-adherence to medication, with lighter segments indicating monthly non-adherence and darker segments weekly non-adherence. Percentages are provided above each bar. Brackets indicate comparisons between the time points, with p values obtained from McNemar tests. Polypharmacy and the presence of pDDIs were significantly more frequent at follow-up, whereas the proportion of non-adherent patients did not change significantly. MS multiple sclerosis, pDDI potential drug–drug interactions, Rx prescription

Consistent with the increased medication use, the proportion of patients with at least one pDDI in their medication profile, irrespective of severity, increased significantly from 67.4% at baseline to 81.1% at follow-up (p < 0.001). In contrast, the proportion of patients with at least one major pDDI remained relatively stable over the study period, increasing slightly from 21.7% to 24.6% (p = 0.522).

Despite the higher medication load and increased pDDI exposure, medication adherence remained largely unchanged between baseline and follow-up. Monthly non-adherence increased marginally from 25.7% to 27.3% (p = 0.855), and weekly non-adherence rose from 8.2% to 12.1% (p = 0.181), with neither change reaching statistical significance.

Detailed Patterns of Medication Use, pDDIs, and Medication Non-Adherence

A more detailed longitudinal analysis of the medication-related characteristics is presented in Table 2. From baseline (2019/20) to follow-up (2024/25), the mean total number of medications per patient increased from 5.1 ± 2.8 to 6.8 ± 4.3 (p < 0.001). This increase was mainly attributable to a higher use of comorbidity drugs, rising from 2.2 ± 2.1 to 3.5 ± 3.4 (p < 0.001). Prescribed medications accounted for the majority of drugs taken at both time points, with the mean number per patient increasing from 3.9 ± 2.7 at baseline to 5.0 ± 3.5 at follow-up (p < 0.001). The mean number of long-term and on-demand medications increased by 0.8 and 0.9, respectively (both p < 0.001). In parallel, the mean total number of pDDIs per patient increased markedly, more than doubling from 5.9 ± 9.2 to 11.9 ± 17.8 (p < 0.001). Increases were observed across all pDDI severity categories (minor, moderate, and major). Although major pDDIs represented a smaller proportion of interactions overall, the patients with MS had a significantly higher mean number at follow-up (0.7 ± 1.7) compared with baseline (0.4 ± 0.8) (p = 0.008). Regarding reasons for medication non-adherence, forgetting to take medication was the most frequently reported barrier, affecting approximately 82% of non-adherent patients at both time points. Other reasons, such as side effects or a perceived lack of necessity for the medication, were reported less often and showed no significant changes over time.

Table 2.

Medication counts, pDDI counts, and reasons for medication non-adherence by study year

Medication data	2019/2020	2024/2025	p value
Number of medications by category, mean ± SD
All medications	5.1 ± 2.8	6.8 ± 4.3	< 0.001*†t
Disease-modifying drugs	0.7 ± 0.5	0.7 ± 0.5	0.764t
Symptomatic drugs	2.2 ± 1.9	2.6 ± 2.0	0.018*†t
Comorbidity drugs	2.2 ± 2.1	3.5 ± 3.4	< 0.001*†t
Prescribed drugs	3.9 ± 2.7	5.0 ± 3.5	< 0.001*†t
Over-the-counter drugs	1.2 ± 1.1	1.8 ± 2.1	< 0.001*†t
Long-term drugs	4.7 ± 2.7	5.5 ± 3.3	< 0.001*†t
On-demand drugs	0.4 ± 0.7	1.3 ± 2.0	< 0.001*†t
Number of pDDIs by severity, mean ± SD
All pDDIs	5.9 ± 9.2	11.9 ± 17.8	< 0.001*†t
Minor pDDIs	2.2 ± 4.3	4.7 ± 7.8	< 0.001*†t
Moderate pDDIs	3.4 ± 5.2	6.5 ± 9.6	< 0.001*†t
Major pDDIs	0.4 ± 0.8	0.7 ± 1.7	0.008*†t
Reasons for not taking drugs as scheduled,a n (%)
All non-adherent patients (per month)b	44 (25.7)	41 (27.3)	0.855M
Forgot to take medicationc	36 (81.8)	34 (82.9)	0.458M
Experienced side effectsc	2 (4.5)	3 (7.3)	1.000M
Considers medication unnecessaryc	4 (9.1)	2 (4.9)	1.000M
Medication was not availablec	2 (4.5)	0 (0.0)	0.480M

A total of 175 patients with MS were included in the analysis. This table provides a detailed comparison of the medication-related data between the two study time points. DrugBank’s Drug Interaction Checker [47] was used to identify pDDIs for each patient

FDR false discovery rate, MS multiple sclerosis, pDDI potential drug–drug interaction, SD standard deviation

^†Significant association (p < 0.05)

*Significant after FDR correction

^aMultiple responses were possible

^bPercentages refer to valid responses only, with 4 and 25 missing values at baseline and follow-up, respectively

^cPercentages refer to those patients who reported not taking a medication at least once a month

^MMcNemar's chi-squared test

^tPaired t test

A detailed overview of medication use is provided in Supplemental Table 1, listing all drug compounds recorded in the cohort for which DrugBank entries were available, along with their frequency of use and involvement in pDDIs. In total, 285 distinct drugs were identified. Cholecalciferol was the most frequently used compound, with its use increasing from 70.3% of patients (n = 123) at baseline to 91.4% (n = 160) at follow-up. Although cholecalciferol was also the drug most frequently involved in pDDIs overall (34.3% at baseline and 51.4% at follow-up, p < 0.001), its involvement in major pDDIs remained relatively low (1.1% and 7.4%, respectively). Over time, the use of other dietary supplements was likewise more frequently associated with pDDIs, namely magnesium (11.4% to 30.9%) and cyanocobalamin (8.0% to 20.0%) (both p < 0.001), neither of which was associated with major pDDIs in the patients’ medication profiles. Overall, 16 drug compounds were significantly more often involved in pDDIs at follow-up than at baseline (p < 0.05). These included DMTs such as ofatumumab (0.0% vs. 6.3%, p = 0.003), treatments for MS-related symptoms such as baclofen (7.4% vs. 12.6%, p = 0.016), and medications for comorbid conditions such as candesartan (3.4% vs. 9.1%, p = 0.009). The drug most frequently involved in major pDDIs was ocrelizumab (10.3% of patients at baseline). Carbamazepine was used less commonly (2.3% at baseline and 5.1% at follow-up) but was associated with major pDDIs in all patients taking this drug.

An overview of all identified pairwise pDDIs and their frequency among the patients with MS at baseline and follow-up is provided in Supplemental Table 2, including the severity classification assigned in DrugBank. Overall, 1564 distinct pDDIs were identified within the cohort, of which 111 (7.1%) were classified as major. At baseline, the most frequent interaction was the combination of ocrelizumab and prednisolone, affecting 16 patients (9.1%). This combination is classified as a major pDDI in DrugBank because of its potential to increase immunosuppressive effects, thereby increasing the risk of serious infection. At follow-up, the most frequent interaction was between cholecalciferol and metoprolol, also affecting 16 patients (9.1%). This interaction is classified as moderate in DrugBank based on a potential effect on metoprolol metabolism. Overall, eight pDDIs were identified significantly more often at follow-up compared with baseline (p < 0.05). These comprised the following combinations: cholecalciferol with metoprolol, venlafaxine and carbamazepine, metamizole with torasemide, pantoprazole and metoprolol, as well as cladribine with cyanocobalamin and magnesium. However, individual pDDIs affected relatively few patients, and increases in their prevalence over time were small, reaching up to 4.6 percentage points.

Associations of Baseline Characteristics with Medication-Related Outcomes at Follow-Up

Several sociodemographic and clinical characteristics at baseline were associated with the primary medication-related outcomes at follow-up in univariable logistic regression analyses (Table 3). Older age was significantly associated with polypharmacy at follow-up (OR 1.048 per year). Moreover, patients receiving a disability pension had higher odds of polypharmacy (OR 6.101), whereas those who were employed part-time (OR 0.272) or full-time (OR 0.199) had lower odds. More years of schooling were associated with lower odds of polypharmacy as well (OR 0.664 per year). Among the clinical variables, SPMS (OR 4.672), higher neurological disability (OR 1.381 per one-point increase in the EDSS score), longer disease duration (OR 1.068 per year), and having at least one coexisting medical condition (OR 8.187) were significantly associated with polypharmacy. Similar associations were observed for the presence of pDDIs at follow-up: age, schooling years, employment status, disability pension, EDSS score, and having a coexisting medical condition showed significant associations in the same direction as for polypharmacy. In addition, having a university degree was associated with lower odds of having a pDDI (OR 0.085). Non-adherence at follow-up was not associated with baseline sociodemographic characteristics but was associated with clinical factors, including PPMS (OR 0.100) and EDSS score (OR 0.854). The baseline medication-related characteristics were predictive of the corresponding outcomes at follow-up. For instance, polypharmacy at baseline was strongly associated with polypharmacy at follow-up (OR 9.781). However, polypharmacy and the presence of pDDIs at baseline were not significantly associated with medication non-adherence at follow-up. Sex, partnership status, and smoking status were not associated with any of the medication-related outcomes.

Table 3.

Associations of baseline patient characteristics with polypharmacy, presence of pDDIs, and medication non-adherence at follow-up

Characteristics as of 2019/20	Polypharmacy in 2024/25 (n = 109 yes vs. n = 66 no)			Presence of pDDIs in 2024/25 (n = 142 yes vs. n = 33 no)			Non-adherence to medication in 2024/25 (n = 41 yes vs. n = 109 no)
	OR	95% CI	p valueLR	OR	95% CI	p valueLR	OR	95% CI	p valueLR
Sociodemographic data
Age (years)	1.048	1.021–1.076	< 0.001*†	1.032	1.000–1.064	0.049†	0.972	0.945–1.001	0.059
Sex: male (ref: female)	1.074	0.549–2.102	0.835	0.966	0.423–2.205	0.935	0.513	0.215–1.224	0.132
Schooling (years)	0.664	0.479–0.922	0.014*†	0.572	0.391–0.838	0.004*†	1.286	0.885–1.867	0.187
Educational level (ref: no training)
Skilled worker	0.506	0.101–2.545	0.409	0.263	0.015–4.702	0.271F	2.333	0.274–19.867	0.438
Technical college	0.514	0.089–2.963	0.457	0.151	0.008–2.920	0.117F	4.667	0.495–43.961	0.178
University	0.171	0.026–1.111	0.064	0.085	0.004–1.720	0.038F†	3.111	0.281–34.419	0.355
Occupational status (ref: not in workforce)
Part-time work	0.272	0.126–0.591	0.001*†	0.542	0.208–1.414	0.210	1.143	0.497–2.632	0.753
Full-time work	0.199	0.089–0.443	< 0.001*†	0.331	0.131–0.836	0.019†	0.703	0.275–1.797	0.462
Pension status (ref: not receiving pension)
Old-age pension	2.596	0.766–8.798	0.125	0.732	0.211–2.543	0.623	0.682	0.176–2.644	0.580
Disability pension	6.101	2.632–14.143	< 0.001*†	2.927	1.050–8.162	0.040†	0.909	0.408–2.027	0.816
Living in a partnership: yes (ref: no)	1.128	0.529–2.408	0.755	0.667	0.237–1.874	0.442	1.299	0.510–3.307	0.583
Smoking status (ref: never smoker)
Former smoker	1.345	0.619–2.922	0.455	1.333	0.509–3.490	0.558	0.389	0.144–1.047	0.062
Current smoker	1.515	0.728–3.151	0.266	1.302	0.533–3.182	0.562	0.798	0.354–1.799	0.586
Clinical data
Disease course (ref: CIS/RRMS)
SPMS	4.672	1.702–12.830	0.003*†	2.967	0.846–10.404	0.089	0.535	0.207–1.382	0.180F
PPMS	2.148	0.545–8.464	0.274	2.871	0.353–23.360	0.324	0.100	0.006–1.754	0.027F†
EDSS score	1.381	1.190–1.601	< 0.001*†	1.355	1.123–1.635	0.002*†	0.854	0.729–1.000	0.050†
Disease duration (years)	1.068	1.023–1.115	0.003*†	1.052	0.997–1.110	0.066	1.009	0.970–1.050	0.660
Patient care: inpatient (ref: outpatient)	1.864	0.365–9.518	0.454	0.684	0.132–3.551	0.651	11.676	1.264–107.812	0.030†
Coexisting medical condition: yes (ref: no)	8.187	3.539–18.941	< 0.001*†	5.423	2.376–12.375	< 0.001*†	1.159	0.452–2.975	0.759
Medication data
Polypharmacy: yes (ref: no)	9.781	4.739–20.185	< 0.001*†	9.126	3.324–25.058	< 0.001*†	0.858	0.418–1.760	0.675
Presence of pDDIs: yes (ref: no)	5.678	2.857–11.282	< 0.001*†	7.306	3.165–16.864	< 0.001*†	0.593	0.279–1.263	0.176
Non-adherence to medication: yes (ref: no)	1.028	0.504–2.096	0.939	1.476	0.560–3.888	0.431	9.754	4.171–22.814	< 0.001*†

This table reports ORs and p values from logistic regression models run separately for each baseline characteristic. Polypharmacy was defined as the concurrent use of ≥ 5 medications, presence of pDDIs as at least one potential interaction regardless of severity, and non-adherence as not taking a medication at least once per month

CI confidence interval, CIS clinically isolated syndrome, EDSS Expanded Disability Status Scale, FDR false discovery rate, MS multiple sclerosis, OR odds ratio, pDDI potential drug–drug interaction, PPMS primary progressive multiple sclerosis, ref reference, RRMS relapsing–remitting multiple sclerosis, SD standard deviation, SPMS secondary progressive multiple sclerosis

^†Significant association (p < 0.05)

*Significant after FDR correction

^FFirth's correction was applied

^LRBinary logistic regression

Further univariable logistic regression analyses revealed consistent associations across additional outcome definitions (Supplemental Table 3). Transition to polypharmacy between baseline and follow-up was associated with a subset of the sociodemographic and clinical factors that were also related to polypharmacy status at follow-up, with employment status, progressive disease course, EDSS score, disease duration, and coexisting medical conditions showing associations in the same direction. Likewise, university education and pension status were negatively associated with the development of pDDIs, while no baseline variable was related to the onset of medication non-adherence during follow-up. When applying stricter definitions of the status outcomes, the same predictors remained significant for Rx polypharmacy as for overall polypharmacy, with associations in the same direction. However, SPMS was additionally associated with higher odds of major pDDIs (OR 3.316) and lower odds of weekly medication non-adherence (OR 0.092) at follow-up. Overall, similar patterns were observed when analyzing transitions in the medication-related outcomes between baseline and follow-up and when applying stricter outcome definitions. However, as some of these analyses were based on small and unevenly distributed patient subgroups, these findings should be considered exploratory.

To identify independent predictors of medication-related outcomes, multivariable logistic regression with bootstrap-based stability selection was performed. For polypharmacy at follow-up, having at least one coexisting medical condition showed the highest selection stability (99.2%) and was strongly associated with the outcome (OR 6.822) (Fig. 3). EDSS score (selection frequency 86.1%, OR 1.236) and disability pension status (81.7%, OR 2.695) were likewise retained as stable predictors in the final model. For the presence of pDDIs at follow-up, only coexisting medical conditions crossed the predefined stability threshold, with a selection frequency of 56.7% and an OR of 5.423. Similarly, inpatient care at baseline was the only variable stably selected as a predictor of medication non-adherence (selection frequency 50.7%, OR 11.676). In these cases, as no additional predictors were retained, the findings from the multivariable procedure were consistent with the univariable results (Table 3). Further multivariable analyses using stricter outcome definitions yielded limited additional information. For Rx polypharmacy at follow-up, a similar predictor pattern to that for overall polypharmacy emerged, with the additional inclusion of age as an independent explanatory variable (OR 1.031). In contrast, no predictor reached the stability selection threshold for major pDDIs or weekly medication non-adherence. Likewise, in the analyses of transitions in medication-related outcomes between baseline and follow-up, no variable met the criterion for inclusion in multivariable models.

Fig. 3.

Predictors of polypharmacy, presence of pDDIs, and medication non-adherence identified by the multivariable procedure. Results of the association analyses of sociodemographic and clinical characteristics at baseline (2019/20) with three medication-related outcomes in patients with MS (n = 175). Dependent variables were a having polypharmacy (total medication), b having one or more pDDIs (of any severity), and c being non-adherent to medication (per month) at the follow-up time point (2024/25). Upper panels (lollipop charts): Selection probabilities for candidate predictors as estimated from bootstrap-based feature selection, combining penalized multivariable logistic regression via a balanced elastic net (α = 0.5 and λ_1se from cross-validation) and stability selection [50]. The horizontal lines represent the frequency with which a variable was retained across 1000 bootstrap resamples. Only the top 5 variables are shown in each panel. The red vertical lines denote the 50% stability selection threshold. Colored lines indicate variables that passed this threshold, whereas gray lines represent unstable variables that did not reach it. Lower panels (forest plots): ORs and 95% CIs for the predictors of polypharmacy, presence of pDDIs, and medication non-adherence in the final unpenalized logistic regression models. Models shown in b, c are effectively univariable, as only a single predictor was retained. Model accuracy is provided in the plot headers. CI confidence interval, EDSS Expanded Disability Status Scale, OR odds ratio, pDDI potential drug–drug interaction

Discussion

Medication-related risks represent a central challenge in contemporary MS care, as long-term DMTs are often combined with symptomatic treatments and medications for comorbid conditions [14, 25]. In this longitudinal cohort study, we assessed changes in polypharmacy, the presence of pDDIs, and medication non-adherence over 5 years in patients with MS and explored baseline patient characteristics associated with these outcomes. A particular strength of our study lies in the comprehensive and integrative evaluation of multiple medication-related domains across the entire treatment regimen, including prescribed drugs and self-medication, in a real-world MS population. Our findings indicate that medication use patterns and related risks evolve over time and are shaped by clinical and sociodemographic factors, underscoring the importance of regular medication reviews and individualized adherence support in routine MS care.

Our MS cohort (n = 175) had a mean age of 45.8 years at baseline, with 70.3% being female. This closely reflects the demographic characteristics reported by the German MS Registry, which includes more than 40000 patients (mean age 48.0 years, 70.9% female) [52]. Clinical features such as disease duration, neurological disability, and distribution of disease courses were likewise comparable to the registry data [52]. The overall demographic profile of our cohort is also comparable to data from other national MS registries, including the Danish MS Registry (mean age 54.2 years, 68.5% female) [53] and the Swedish MS Registry (mean age 52.3 years, 70.3% female) [54]. This suggests that our study population broadly reflects patients with MS seen in routine care. However, the findings should be interpreted in the context of the German healthcare setting. Moreover, medication use patterns are known to be influenced by regional variations in prescribing practices [55], and the prevalence of self-medication varies substantially between countries [56]. Consequently, the generalizability of our results to other healthcare systems and MS populations may be limited.

In our cohort, the prevalence of overall polypharmacy and Rx polypharmacy increased by 9.2 and 11.4 percentage points over 5 years, reaching 62.3% and 48.0% at follow-up. These rates fall within the upper range of previously reported polypharmacy rates in MS populations (14–76.5%) [17, 20], which may be explained by our inclusion of both prescribed and non-prescribed medications. A recent Australian study using a stricter definition of Rx polypharmacy based on repeated medication dispensing within 1 year likewise demonstrated a rise in polypharmacy prevalence over time (23% to 29% between 2017 and 2022) [21]. In our cohort, the use of drugs for comorbid conditions increased most markedly, whereas the number of symptomatic drugs rose moderately and DMT use remained stable over the study period. This likely reflects the accumulation of coexisting medical conditions and symptomatic treatment needs with advancing age and disease duration, in line with previous studies by our group and others linking multimorbidity to higher medication load [20–22, 24]. In particular, comorbidities such as type 2 diabetes and hypertension have been associated with a higher risk of polypharmacy in MS [21]. In our cohort, antihypertensive medications such as candesartan and ramipril were among the drugs used significantly more often at follow-up. The most frequently used compounds were cholecalciferol, cyanocobalamin, and magnesium, consistent with reports showing that vitamins and dietary supplements are among the most commonly used medications in MS, with cholecalciferol taken by 60–73.5% of patients [57, 58]. Our findings suggest that the use of these supplements has increased in recent years in patients with MS.

Several baseline characteristics were associated with polypharmacy status and transitions over time in the logistic regression analyses. Older age and the presence of coexisting medical conditions were among the strongest predictors of polypharmacy. In addition, variables related to aging and disease progression, including longer disease duration, higher neurological disability, progressive disease course, and disability pension status, were associated with higher odds of polypharmacy, whereas higher educational attainment was associated with lower odds. These findings are consistent with those of our previous cross-sectional analyses [59, 60] and with reports from other research groups on factors associated with polypharmacy in MS [17, 21, 23, 58]. From a clinical perspective, these results underscore the importance of awareness of increasing medication burden in aging patients with MS and support the need for regular, structured medication reviews aimed at minimizing unnecessary pharmacotherapy and promoting non-pharmacological management strategies where appropriate [11, 61].

In parallel with the increase in medication use, the prevalence of pDDIs in our cohort rose significantly by 13.7 percentage points, whereas the proportion of patients with at least one major pDDI remained largely unchanged (+ 2.9 percentage points). To our knowledge, longitudinal changes in pDDI prevalence in MS have not previously been reported. Notably, the majority of identified pDDIs were non-severe. In addition, interactions detected by database screening are initially theoretical rather than necessarily associated with adverse outcomes. In a hospital-based study, most patients had at least one pDDI (70.1%), but fewer than one-third of interactions were considered clinically relevant (27.0%), and only 0.9% of patients experienced actual harm [62]. Accordingly, override rates for DDI alerts in clinical practice are high (55–98%) [63]. However, high-priority alerts are also commonly overridden, which can increase the risk of adverse drug events [64]. In our cohort, the increasing pDDI exposure was partly driven by the greater use of dietary supplements, particularly cholecalciferol. While clinical evidence for vitamin D-related interactions remains limited, clinicians should be aware that OTC products may interact with prescribed drugs [31, 65, 66], underscoring the importance of systematically documenting self-medication. Finally, several interactions involved DMTs for MS. For example, the combination of ocrelizumab and prednisolone was classified as a major pDDI because of additive immunosuppressive effects. In routine practice, however, corticosteroids are commonly administered as premedication to reduce infusion-related reactions and improve the tolerability of ocrelizumab treatment [67]. Potential interactions were also identified for several other DMTs, such as cladribine, reflecting its immunosuppressive properties and the possibility of transporter-mediated pharmacokinetic interactions [68].

The strongest predictor of having at least one pDDI at follow-up was polypharmacy at baseline (OR 9.126). Beyond medication-related variables, the presence of a coexisting medical condition emerged as the most robust predictor (OR 5.423), while univariable analyses additionally showed positive associations with older age and higher EDSS scores and negative associations with educational level. These findings are consistent with our previous cross-sectional analyses on determinants of pDDI exposure in MS [32, 33]. More specifically, we previously found significantly higher pDDI prevalence rates among patients with MS who had cardiovascular, neurological, psychiatric, and orthopedic comorbidities compared with those without these comorbidities [31]. In a Spanish MS cohort, pDDIs were predominantly attributed to pharmacodynamic mechanisms such as additive CNS depression and increased gastrointestinal toxicity, indicating that interaction risk in patients with MS is often linked to symptomatic treatments and medications for comorbid conditions [39]. Studies in non-MS populations likewise identified multimorbidity and polypharmacy as important factors associated with increased exposure to pDDIs [69], whereas higher educational attainment has been linked to a lower likelihood of severe pDDIs [70]. These findings underscore the importance of proactively considering pDDI risk in aging patients with accumulating comorbidities and increasingly complex medication regimens. Structured, multidisciplinary medication reviews that incorporate clinical context and patient-specific factors may help to improve the evaluation of database-detected interaction alerts and thereby enhance patient safety [71].

Despite the increasing prevalence of polypharmacy and pDDIs, medication non-adherence remained stable over 5 years, with monthly and weekly rates rising by only 1.6 and 3.9 percentage points. Most non-adherent patients (approx. 80%) reported simply forgetting to take their medication. Most previous studies in MS have focused specifically on adherence to DMTs, with a systematic review reporting adherence rates ranging from 52% to 92.8% [72]. This wide variability primarily reflects differences in adherence definitions and assessment methods [73]. In our study, adherence was assessed by retrospective self-report, which is known to result in an overestimation of adherence due to recall bias and social desirability bias [73]. On the other hand, our assessment was not restricted to DMTs, as adherence is clinically relevant across all therapies used. Notably, a Canadian study reported higher adherence to DMTs than to chronic non-MS medications [37], suggesting that adherence challenges in patients with MS may be particularly relevant for concomitant therapies. Targeted strategies can help reduce both unintentional and intentional non-adherence, for example through simplified medication schedules, patient education on reminder strategies, and counseling about treatment-related concerns [74].

In the regression analyses, inpatient care and non-adherence at baseline emerged as the strongest predictors of non-adherence at follow-up, whereas patients with progressive disease course and higher disability levels showed better adherence. In contrast, neither polypharmacy nor the presence of pDDIs predicted subsequent non-adherence. These findings suggest that medication non-adherence represents a distinct domain of medication-related risk. Previous research has indicated that polypharmacy can influence adherence in opposing directions, potentially reducing adherence due to increased regimen complexity while also promoting treatment routines in some patients [40, 75]. The association between inpatient care at baseline and later non-adherence was based on a small inpatient subgroup and reflects care status at a single time point rather than general healthcare utilization, and should therefore be interpreted cautiously. One possible explanation may be that patients who were hospitalized received intensified counseling, increasing their awareness of missed doses and thereby influencing subsequent self-reporting of medication adherence. The finding that higher EDSS scores were associated with lower odds of non-adherence is consistent with previous reports [40, 76], possibly reflecting a reduced perceived need for continuous treatment in less disabled patients, although evidence across studies is inconsistent [72]. Overall, predicting individual non-adherence is challenging, as multiple factors contribute, including socio-economic context, continuity of care, patient–provider relationships, health literacy, and personality traits [40, 77]. In patients at risk of skipping medication doses, long-term treatment outcomes might be improved by targeted education on treatment benefits and risks combined with structured adherence support [74].

Several limitations should be acknowledged. This study was conducted within a specific German healthcare setting, limiting the generalizability of the results to regions with different prescribing practices and self-medication patterns. Polypharmacy was defined quantitatively based on medication counts, following the most frequently used numerical definition in the literature [18]. However, this definition does not capture medication appropriateness and may overestimate or underestimate the clinical relevance of medication burden in individual patients. Newer concepts emphasize the value of qualitative assessments to distinguish between appropriate and unnecessary polypharmacy [78], which was beyond the scope of the present study. Regarding the assessment of pDDIs using DrugBank, it should be noted that considerable variability exists between drug interaction databases [33]. Furthermore, the occurrence of clinically relevant pDDI-related adverse outcomes was not systematically assessed. Such an assessment would require a detailed clinical evaluation of whether adverse events were attributable to specific drug interactions rather than to individual drug effects, comorbidities, or MS-related symptoms. Moreover, no information was available on whether the treating physicians were aware of the identified pDDIs and whether they had already implemented preventive strategies such as dose adjustments or modified intake schedules. Medication adherence was assessed by patient self-report and is therefore subject to recall and reporting bias. In addition, some estimates were based on small subgroup sizes, particularly for rare categories and transition outcomes, and should therefore be interpreted with caution. Despite these limitations, the integrative evaluation of polypharmacy, pDDIs, and non-adherence within a unified longitudinal framework represents a key strength of our study and provides real-world insights into evolving medication-related risks in routine MS care. Future research may build on these findings by incorporating multicenter designs, repeated follow-up assessments, and objective adherence measures to further refine risk stratification and support targeted strategies to optimize the pharmacological management in patients with MS.

Conclusion

Medication-related risks represent an important challenge in the management of patients with MS, as DMTs are increasingly combined with symptomatic therapies and medications for comorbid conditions. In this real-world cohort study, we jointly examined polypharmacy, pDDI exposure, and medication non-adherence over a 5-year period. The prevalence of polypharmacy and pDDIs increased significantly over time, largely reflecting disease progression and the aging-related accumulation of comorbidities requiring more complex treatment. In contrast, medication non-adherence remained relatively stable and appeared to constitute a largely independent risk domain, influenced by prior adherence behavior and showing lower rates among patients with higher disability levels. These findings emphasize the clinical value of regular, structured medication reviews that consider the full range of prescribed drugs as well as OTC medications and dietary supplements. At the same time, targeted adherence support remains essential to sustain treatment effectiveness. Overall, this integrative longitudinal perspective provides clinically meaningful insights into the evolving medication-related challenges faced by patients with MS and underscores the importance of patient-centered approaches to optimize long-term treatment safety and outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Niklas Frahm, Jörg Richter, Felicita Heidler, and Uwe K. Zettl conceptualized the study. Avinash M. Suntah, Bassel Barhoum, Julia Baldt, Barbara Streckenbach, and Felicita Heidler collected the data. Avinash M. Suntah and Michael Hecker analyzed the data and prepared the figures and tables. Avinash M. Suntah and Niklas Frahm verified the underlying data. Avinash M. Suntah and Michael Hecker interpreted the data and drafted the original manuscript. Jonas E. Langenberger, Niklas Frahm, Felicita Heidler, and Uwe K. Zettl provided important intellectual content. Uwe K. Zettl supervised the research. All authors have read and approved the final version of the manuscript.

Funding

No funding or sponsorship was received for this study or publication of this article. The Rapid Service Fee was funded by the authors.

Data Availability

The data supporting the findings of this study are available within the article and its supplementary material. Additional data are available from the corresponding author upon reasonable request.

Declarations

Conflicts of Interest

Michael Hecker received speaking fees and travel funds from Bayer HealthCare, Biogen, Merck Healthcare, Novartis, and Teva. Niklas Frahm received travel funds for research meetings from Novartis. Felicita Heidler received speaking fees and travel funds from Bayer HealthCare, Biogen, Bristol Myers Squibb, Janssen, Merck Healthcare, Neuraxpharm, Novartis, Roche, Sanofi Genzyme, and Teva. Uwe K. Zettl received research support, speaking fees, and travel funds from Alexion, Almirall, Bayer HealthCare, Biogen, Bristol Myers Squibb, Janssen, Merck Healthcare, Novartis, Roche, Sanofi Genzyme, and Teva as well as the European Union, BMBF, BMWi, and DFG. Avinash M. Suntah, Bassel Barhoum, Jonas E. Langenberger, Julia Baldt, Barbara Streckenbach, and Jörg Richter declare that they have no conflicts of interest.

Ethical Approval

Ethical approval was obtained from the ethics committee of the University of Rostock (permit number A 2019-0048) and the Medical Chamber of Thuringia (reference number 23117/2019/88). The study was conducted in accordance with the Declaration of Helsinki and the European General Data Protection Regulation. All participants provided written informed consent prior to inclusion.