Long-Term Antidepressant Use and Dementia Risk: A Propensity Score-Matched Study of Swiss Older Adults, with an Anticholinergic-Specific Analysis Authors: Melanie A Amrein, Michael P Hengartner, Markus Näpflin, Renato Farcher, Carola A Huber Published: 2026 ABSTRACT AbstractBackgroundAntidepressants are widely prescribed, with increasing use among older adults. However, evidence on their association with dementia remains inconsistent.ObjectiveThis study examines the association between long-term antidepressant use and dementia risk in older adults using a propensity score-matched design, with a separate analysis of long-term tricyclic antidepressant (TCAs) use, given their anticholinergic properties.MethodsA retrospective cohort study was conducted using Swiss healthcare claims data (2013–2023) from older adults aged 65 years and above. Long-term antidepressant users (> 2 years) and long-term TCAs users were separately matched to nonusers using variable ratio propensity score matching. The primary outcome was incident dementia, defined by a recorded dementia diagnosis or antidementia drug prescription. Covariates were derived from inpatient and/or outpatient records, depending on the respective variable.ResultsA final sample of 19,155 long-term antidepressant users were compared with 20,648 propensity score matched nonusers. After adjusting for covariates, long-term antidepressant use was significantly associated with dementia (odds ratio [OR] 1.92, 95% confidence interval [CI] 1.72–2.15, p < 0.001). Long-term TCAs users also had higher odds of developing dementia compared with nonusers (OR 1.73, 95% CI 1.33–2.25, p < 0.001). Sensitivity analyses largely confirmed these results: long-term antidepressant users had higher odds of developing dementia compared with short-term users (OR 1.23, 95% CI 1.11–1.35, p < 0.001), but for TCA, the difference between long-term and short-term users was not statistically significant (OR 1.25, 95% CI 0.99–1.57, p = 0.059).ConclusionsThese findings suggest that long-term antidepressant and long-term TCAs use is associated with an increased risk of dementia in older adults, even after accounting for various comorbidities. This study adds to the ongoing debate on the link between antidepressants and dementia. Future research is needed to further clarify underlying mechanisms and guide clinical practice.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s40801-026-00560-y. Key Points This study examined 19,155 individuals aged 65 years and above who used antidepressants for more than 2 years, defined as long-term users.Long-term antidepressant use was associated with higher odds of developing dementia compared with propensity score matched non-users, after adjusting for multiple covariates.Our findings emphasize the importance of weighting potential cognitive risk when considering prolonged antidepressant treatment in aging populations. Introduction Antidepressants are considered first-line medications for various psychiatric disorders, including depression [1], as well as other conditions such as chronic pain [2]. Antidepressants are among the most frequently prescribed medications globally, with a steady increase in their use over time [3, 4]. The increase may indicate a positive trend, reflecting enhanced awareness, early diagnosis of mental health conditions, and reduced stigma surrounding mental health treatment [5]. However, increasing antidepressant use may also result from extended treatment durations [6] and potential overprescription [7], particularly among older adults [8]. Concerns persists about the safety and efficacy of antidepressants in older populations, as studies suggests an association between certain antidepressants and an higher risk of adverse outcomes, including dementia [9–12]. Dementia, a neurodegenerative disease marked by a cognitive decline in older adults, is a growing public health concern given the aging global population and the current lack of treatments capable of reversing or halting disease progression [13]. Known risk factors include hypertension, hearing loss, diabetes, and smoking [14]. The use of psychopharmaceuticals, particularly in older adults, has been linked to an increased dementia risk, potentially influenced by age-related changes in brain structure and drug metabolism [15, 16]. Polypharmacy in older adults with comorbidities further heightens the risk of adverse drug interactions, potentially accelerating neurodegenerative processes [16]. Yet, the influence of antidepressants on dementia risk remains widely uncertain [17]. Previous studies have yielded inconsistent findings; while some suggest a protective effect [18, 19], others report a worsening effect [20–22], and others find no significant association [23, 24]. Given the high prevalence of antidepressant use among older adults [3], investigating their potential impact on dementia risk is critical. Research on the antidepressant–dementia relationship is challenging to conduct since depression itself is a potential dementia risk factor [24–26]. Depression may also even represent an early symptom or prodrome of dementia in some individuals, making it challenging to disentangle causality in the antidepressant–dementia relationship [27–29]. Large datasets with extended follow-up periods are needed to rigorously explore this association. Prior studies have employed various methodological approaches, including propensity score matching, to control for confounding factors and strengthen causal inferences when analyzing large observational datasets [12, 30]. Among different classes of antidepressants, tricyclic antidepressants (TCAs) warrant particular attention owing to their strong anticholinergic properties. Anticholinergic drugs block acetylcholine, a neurotransmitter essential for cognitive function, and have been associated with an increased risk of dementia in older adults [15, 31, 32]. TCAs may contribute to cognitive impairment and neurodegenerative processes over time. Studies have demonstrated that cumulative exposure to anticholinergic antidepressants is significantly associated with a higher dementia risk, with effects persisting up to 20 years post-exposure [32]. Given these findings, it is crucial to analyze TCAs separately from other antidepressants to determine whether their impact on dementia risk is driven by their anticholinergic burden or other pharmacological properties. This study aims to deepen our understanding of the risks of antidepressant use in older adults. Specifically, we will assess the relationship between long-term antidepressant use and the incidence of dementia in the older Swiss population. We conduct a longitudinal, propensity score matched analysis of healthcare claims data from a large, representative sample. In addition to analyzing all long-term antidepressant users, we will conduct a separate analysis of TCAs to account for their strong anticholinergic effects and their potential role in dementia risk. Methods Data This observational retrospective study was based on healthcare claims data from one of the largest health insurances in Switzerland. Basic insurance is mandatory in Switzerland, and in 2023, the Helsana Group covered 14.5% (1,296,439) of the Swiss population. The claims data include detailed information on insured individuals, such as sociodemographic characteristics, outpatient and inpatient services, inpatient diagnoses (ICD‑10‑GM), prescribed medications, laboratory tests, and healthcare costs. Study Design and Study Population To evaluate the association between antidepressant use and incident dementia, we conducted a retrospective cohort study using data from 1 January 2013 to 31 December 2023. The study population comprised individuals aged 65 years and older who were continuously enrolled during the study period. Antidepressants were identified using World Health Organization’s [32] Anatomical Therapeutic Chemical (ATC) classification (see Online Resource 1 for detail information). The index date was defined as the first antidepressant prescription, preceded by a one-year wash-out period during which no antidepressant prescriptions were allowed. Individuals with dementia diagnoses or antidementia drug prescriptions during the wash-out period or treatment phase were excluded to reduce the influence of preexisting documented dementia [12]. The duration of antidepressant use was operationalized as the number of quarters in which at least one antidepressant was recorded during the treatment phase [3]. The treatment phase extended from the index date for 2–3 years (11 quarters) and was followed by a one-year outcome phase. Long-term antidepressant use was defined as prescriptions in nine or more quarters during the treatment phase, corresponding to a duration of more than 2 years and allowing up to a 6 month interruption [33]. Individuals without antidepressant use were defined as having no antidepressant prescriptions and no diagnosed or treated dementia at any time before the outcome phase (see Online Resource 2). Additionally, we evaluated the association between long-term tricyclic antidepressant (TCA) use and incident dementia compared with nonusers. TCAs were identified using ATC codes (see Online Resource 1). The same index date definition, wash-out period, treatment structure, and exclusion criteria were applied as described above. Outcome We assessed dementia incidence using ICD-10 codes F00, F01, F02, F03, G30, G31.0, G31.1, or G31.82, and the prescription of antidementia medications (ATC code N06D) within the outcome phase. The dataset includes only ICD-10 diagnosis from inpatient stays. Covariates Patient sociodemographic and morbidity characteristics were used both for matching and as covariates in the analyses. Sociodemographic variables were assessed at the index date, whereas morbidity characteristics were derived from claims data recorded during the 1-year period prior to the index date. Sociodemographic covariates included age (continuous), sex (female/male), language region (German-, French-, or Italian-speaking regions of Switzerland), and area of residence (urban, midland, or rural). Morbidity-related covariates comprised nursing home stay, the number of chronic conditions (used for matching based on the CCI and PCG), and specific chronic conditions operationalized as binary indicators on the basis of PCG classifications. Because diagnostic information is not comprehensively available in claims data, PCGs were used as a proxy to capture morbidity on the basis of prescribed medications. PCGs represent a well-established pharmaceutical classification system commonly applied in settings with limited clinical diagnostic data. Included chronic conditions comprised acid-related disorders, bone disease, cancer, cardiovascular disease, diabetes, epilepsy, glaucoma, gout, human immunodeficiency virus (HIV), hyperlipidemia, inflammatory disease, iron deficiency, migraine, Parkinson’s disease, pain, respiratory illness, rheumatism, thyroid disorders, and tuberculosis (see Online Resource 3 for detailed information). Statistics To balance differences in baseline characteristics between long-term antidepressant users and nonusers, we applied variable ratio propensity score matching using a nearest neighbor algorithm, allowing up to 12 control individuals per treated individual (mean matching ratio ≈ 3). Exact matching was performed on a predefined set of key covariates, including index month and year, age, sex, nursing home residence, depression diagnosis, and the use of antipsychotics, anxiolytics, hypnotics, or sedatives. In addition, a propensity score was estimated using logistic regression on the basis of further covariates described in the methods section and was used as the distance measure for nearest neighbor matching. Specifically, individuals were weighted to achieve balance with respect to language, area of residence, the number of somatic and psychiatric hospitalizations, number of chronic conditions and number of specific conditions (see Online Resource 3 for detailed information). The distribution of propensity scores before and after matching is presented in Online Resource 4. On the basis of the matched sample, logistic regression models were used to estimate the association between long-term antidepressant use and incident dementia. All models were adjusted for the covariates used in both the exact matching and propensity score estimation, assessed during the 1-year period prior to the index date or at the index date. The adjustment was performed to account for potential residual imbalance after matching and to obtain more robust and consistent estimators. To account for the matched study design, robust standard errors were used for all post matching regression analyses. Reduced models are presented for readability and a more concise presentation and include covariates most strongly associated with the outcome, based on estimates from the fully adjusted models. Associations are reported as odds ratios (OR) with 95% confidence intervals (CI). A p-value of < 0.05 was considered statistically significant. Additional analyses were conducted to assess the robustness of the findings by comparing long-term with short-term antidepressant use, defined as a prescription duration of less than 1 year (see Online Resource 1 for detailed information), both in the overall cohort and in analyses restricted to TCAs. These sensitivity analyses applied the same propensity score–based approach as the main analyses. A landmark design was used, in which individuals were classified at a fixed time point and required to be alive and without a dementia diagnosis at that time, with follow-up for outcomes beginning thereafter. Furthermore, we additionally conducted analyses that incorporated individuals with incident dementia occurring during the treatment phase. All analyses were conducted using the statistical software R, version 4.4.1 [37] and R packages MatchIt, final-fit [38, 39]. All analyses were conducted using the statistical software R, version 4.4.1 [33] and R packages MatchIt, finalfit [34, 35]. Results We identified 22,000 individuals who initiated long-term antidepressant use between January 2013 and December 2023. Totally 2845 individuals were excluded owing to dementia during the treatment phase, yielding a final sample of 19,155. A matched control group included 22,000 weighted cases with no antidepressant prescriptions, with 1352 exclusions for dementia during the treatment phase, yielding 20,648 in the no-use control group. Post-matching baseline characteristics of the two groups, including demographic factors, are presented in Table 1. Propensity score matching balanced the groups effectively. In the long-term antidepressant use group, mean age was 76.0 years (SD ± 7.4), with 72.2% female, and 10.2% (n = 1954) residing in nursing homes and 0.6% (n = 119) of the individuals had a depression diagnosis from an inpatient stay. Additionally, 8.5% (n = 1637) had at least one antipsychotic prescription, and 62.3% (n = 11,938) had at least one prescription for an anxiolytic, hypnotic or sedative. In the no-use control group, mean age was 76.2 years (SD ± 7.4), 72.2% were female, 10.4% (n = 2148) were residing in nursing homes, 0.7% (n = 135) of the individuals had a depression diagnosis from an inpatient stay, 8.5% (n = 1765) had at least one antipsychotic prescription, and 61.0% (n = 12,603) had at least one prescription for an anxiolytic, hypnotic, or sedative. Table 1Baseline characteristics for long-term antidepressants use group and no-use control group after weighting.VariablesNo-use control groupLong-term antidepressants use groupExact matching variables SexWomen14,914 (72.2)13,833 (72.2)Men5734 (27.8)5322 (27.8) Age (at index date)Mean (SD)76.2 (7.4)76.0 (7.4) Nursing homeNo18,500 (89.6)17,201 (89.8)Yes2148 (10.4)1954 (10.2) Diagnosis depressionNo20,513 (99.3)19,036 (99.4)Yes135 (0.7)119 (0.6) AntipsychoticsNo18,883 (91.5)17,518 (91.5)Yes1765 (8.5)1637 (8.5) Anxiolytics, hypnotics or sedativesNo8045 (39.0)7217 (37.7)Yes12,603 (61.0)11,938 (62.3)Demographic characteristics LanguageGerman14,662 (71.0)13,685 (71.4)French3776 (18.3)3474 (18.1)Italian2210 (10.7)1996 (10.4) AreaUrban13,905 (67.3)12,758 (66.6)Midland3962 (19.2)3784 (19.8)Rural2780 (13.5)2613 (13.6)Clinical characteristics Somatic hospitalizationNo7572 (36.7)6344 (33.1)Yes13,076 (63.3)12,811 (66.9) Psychiatric hospitalizationNo20,317 (98.4)18,266 (95.4)Yes331 (1.6)889 (4.6) Comorbidities Comorbidity-index (CCI)Mean (SD)0.2 (0.8)0.2 (1.0) Acid related disordersNo14,560 (70.5)13,240 (69.1)Yes6088 (29.5)5915 (30.9) Bone diseaseNo18,838 (91.2)17,399 (90.8)Yes1810 (8.8)1756 (9.2) CancerNo20,062 (97.2)18,604 (97.1)Yes586 (2.8)551 (2.9) CVDNo4299 (20.8)4328 (22.6)Yes16,349 (79.2)14,827 (77.4) DiabetesNo17,560 (85.0)16,293 (85.1)Yes3088 (15.0)2862 (14.9) EpilepsyNo19,826 (96.0)18,284 (95.5)Yes822 (4.0)871 (4.5) GlaucomaNo18,355 (88.9)16,865 (88.0)Yes2293 (11.1)2290 (12.0) GoutNo19,753 (95.7)18,302 (95.5)Yes895 (4.3)853 (4.5) HIVNo20,638 (100.0)19,140 (99.9)Yes9 (0.0)15 (0.1) HyperlipidemiaNo12,472 (60.4)11,736 (61.3)Yes81,76 (39.6)7419 (38.7) Inflammatory DiseaseNo20,458 (99.1)18,945 (98.9)Yes190 (0.9)210 (1.1) Iron deficiencyNo19,476 (94.3)18,014 (94.0)Yes1172 (5.7)1141 (6.0) MigrainesNo20,380 (98.7)18,917 (98.8)Yes268 (1.3)238 (1.2) ParkinsonsNo19,622 (95.0)18,096 (94.5)Yes1026 (5.0)1059 (5.5) PainNo13,803 (66.9)12,699 (66.3)Yes6845 (33.1)6456 (33.7) Respiratory illnessNo17,597 (85.2)16,218 (84.7)Yes3051 (14.8)2937 (15.3) RheumatismNo14,287 (69.2)13,161 (68.7)Yes6361 (30.8)5994 (31.3) Thyroid disorderNo18,224 (88.3)16,884 (88.1)Yes2424 (11.7)2271 (11.9) TuberculosisNo20,622 (99.9)19,121 (99.8)Yes26 (0.1)34 (0.2)Data are mean (SD) or n (%)CVD cardiovascular diseases Baseline characteristics for long-term antidepressants use group and no-use control group after weighting. Data are mean (SD) or n (%) CVD cardiovascular diseases In the long-term antidepressant use group, 4.9% (n = 933) developed dementia in the outcome phase, with 534 individuals receiving antidementia treatment and 446 receiving a dementia diagnosis. In comparison, in the no-use control group, 2.6% (n=530) of individuals developed dementia, with 310 receiving antidementia treatment and 241 receiving a dementia diagnosis. Dementia risk was depending on the age range of individuals, for example, in the long-term antidepressant use group, 2.3% of individuals between 65–70 years of age developed dementia, while 7.3% of individuals older than 85 years of age developed dementia. Long-Term Antidepressants Use versus No-Use Control Group Table 2 presents the results of a logistic regression analysis examining the association between long-term antidepressant use versus no-use control group and dementia risk, controlling for significant covariates. Individuals in the long-term antidepressant use group had significantly higher odds of dementia compared with those in the propensity score matched no-use control group (OR 1.92, 95% CI 1.72–2.15, p < 0.001), controlled for significant covariates. Older age (OR 1.06, 95% CI 1.05–1.07, p < 0.001), Italian-speaking (OR 1.23, 95% CI 1.04–1.44, p = 0.012) compared with German-speaking individuals, antipsychotics (OR 1.46, 95% CI 1.23–1.72, p < 0.001), somatic hospitalization (OR 1.31, 95% CI 1.17–1.47, p < 0.001), psychiatric hospitalization (OR 1.47, 95% CI 1.13–1.88, p = 0.003), and comorbid iron deficiency (OR 1.28, 95% CI 1.04–1.55, p = 0.016), were also significantly associated with higher odds for dementia. Online Resource 5 presents the results of the logistic regression analysis, examining the association between long-term antidepressant use versus no-use control group and dementia risk, controlling for all covariates, which did not show significant effects for other covariates. For sensitivity analysis, we investigated the risk for the long-term antidepressant use group compared with the no-use control group including all individuals with dementia diagnosis or anti-dementia treatment during the treatment phase, still showing higher odds for long-term antidepressant use group (OR 2.26, 95% CI 2.10–2.44, p < 0.001; see Online Resource 6). Table 2Results of the logistic regression investigating dementia risk by comparing long-term users with no-users controlling for significant covariates.No dementiaDementiaOR (95% CI, p value)GroupNo-use control group20,118 (97.4)530 (2.6)–Long-term antidepressants use group18,222 (95.1)933 (4.9)1.92 (1.72–2.15, p < 0.001)AgeMean (SD)75.6 (7.2)78.6 (6.9)1.06 (1.05–1.07, p < 0.001)AntipsychoticsNo61,048 (96.3)2332 (3.7)–Yes4240 (94.5)246 (5.5)1.46 (1.23–1.72, p < 0.001)LanguageGerman47,646 (96.2)1885 (3.8)–French11,030 (96.6)387 (3.4)0.78 (0.67–0.90, p = 0.001)Italian6611 (95.6)307 (4.4)1.23 (1.04–1.44, p = 0.012)AreaUrban43,842 (96.0)1818 (4.0)–Midland12,777 (96.4)477 (3.6)0.98 (0.86–1.12, p = 0.798)Rural8669 (96.8)283 (3.2)0.77 (0.65–0.91, p = 0.003)Somatic hospitalizationNo22,226 (96.6)771 (3.4)–Yes43,061 (96.0)1808 (4.0)1.31 (1.17–1.47, p < 0.001)Psychiatric hospitalizationNo63,605 (96.2)2485 (3.8)–Yes1682 (94.7)94 (5.3)1.47 (1.13–1.88, p = 0.003)GoutNo62,416 (96.2)2484 (3.8)–Yes2871 (96.8)95 (3.2)0.73 (0.54–0.96, p = 0.034)Iron deficiencyNo61,759 (96.3)2398 (3.7)–Yes3528 (95.1)180 (4.9)1.28 (1.04–1.55, p = 0.016)N = 39,803 Results of the logistic regression investigating dementia risk by comparing long-term users with no-users controlling for significant covariates. N = 39,803 Long-Term versus Short-Term Antidepressants Use Group Figure 1 presents the results of a logistic regression model assessing dementia risk by comparing individuals taking antidepressant less than 1 year (short-term antidepressants use group) with those using them for over 2 years (long-term antidepressants use group), controlling for significant covariates. In the short-term antidepressant use group (n = 21,523), 3.7% (n = 805) developed dementia in the outcome phase. The results indicate higher odds for individuals in the long-term compared with short-term antidepressants use group (OR 1.23, 95% CI 1.11–1.35, p < 0.001). The covariates age, psychoses, somatic hospitalization, psychiatric hospitalization, the comorbidities gout and iron deficiency were also significantly associated with dementia (see Fig. 1). Online Resource 7 presents the results of the logistic regression analysis, examining the association between long-term versus short-term antidepressant use group and dementia risk. Forest plot of a logistic regression model outcomes for the association between groups of antidepressant use (long-term use versus short-term use) and dementia Long-Term Tricyclic Antidepressants Users Altogether 3579 (18.7%) of long-term antidepressant users had at least one prescription of TCA (see Online Resource 8) and 1564 (8.2%) of long-term antidepressant users had only TCAs prescriptions during the whole study period (monotherapy). There was a small sex difference in monotherapy, as 8.6% of women and 7.1% of men received long-term tricyclic antidepressants only. To assess the association between long-term TCAs use and dementia risk, we compared long-term antidepressant users with at least one TCAs prescription (n = 3579) with a propensity score-matched no-use control group (n = 3795). Online Resource 8 showing baseline characteristics of long-term TCAs use group and the corresponding no-use control group. They were well-balanced across, for example, taking antipsychotics (long-term group = 9.4% and control group = 9.3%) or whether individuals are living in a nursing home (long-term group = 6.6% and control group = 6.5%). In the long-term tricyclic antidepressant use group 4.2% developed dementia and in the no-use control group 2.5% developed dementia (see Online Resource 9). Our analysis revealed significantly higher odds for developing dementia for individuals in the long-term tricyclic antidepressant use group compared with those in the no-use control group (OR 1.73, 95% CI 1.33–2.25, p < 0.001), adjusted for significant covariates such as age (OR 1.07, 95% CI 1.05–1.09, p < 0.001). Further we compared 21,523 of short-term tricyclic antidepressant users with 3579 long-term tricyclic antidepressant users (see Online Resource 10). Our analysis showed no significant effect (OR 1.25, 95% CI 0.99–1.57, p = 0.059). Discussion In this propensity score-matched longitudinal case-control study of adults aged 65 years and older, we found that long-term antidepressant use of more than 2 years was associated with an increased risk of all-cause dementia after adjustment for a wide range of comorbidities. This finding was robust in secondary analyses, comparing long-term antidepressant users with short-term antidepressant users, thus controlling for an exposure-dependent relationship. The increased risk of dementia remained consistent in a separate propensity score-matched analysis focused exclusively on tricyclic antidepressants (TCAs). As we excluded individuals with antidepressant use within 1 year prior to the observation period, and the inpatient diagnosis of depression was equally balanced between groups, our findings should be less affected by both protopathic bias and indication bias owing to depression. Our findings are consistent with prior studies [12, 20–22, 27, 36]. For example, two meta-analyses reported that antidepressant use increases dementia risk in elderly populations compared with nonusers [37, 38], though high heterogeneity among included studies was noted, particularly regarding adjustments for confounders such as depression. Notably, recent studies explicitly examining the antidepressant–dementia association in depressed individuals support our findings. For instance, Peakman et al. [39] observed an increased dementia risk after 2.5 years of antidepressant use in older individuals with depression. Similarly, Wang et al. [40] found a higher dementia risk within 2 years of antidepressant use compared with psychotherapy in older adults with depression. However, evidence remains controversial, as other studies have found no association between antidepressant use and dementia risk [18, 19, 23, 41]. One of the primary critique points of observational studies is their susceptibility to selection bias. Therefore, we used propensity score matching to control for the inpatient diagnosis of depression and 21 medical comorbidities in a general population sample of older Swiss adults. Our findings contribute to the ongoing debate on the role of antidepressants in dementia by demonstrating a 1.92-fold increased odd developing dementia among long-term antidepressant users compared with nonusers in a large Swiss cohort. Descriptive analysis revealed that 1564 (8.2%) of long-term antidepressant users aged 65 years and older were prescribed only TCAs for more than 2 years, and 3579 (18.7%) had at least one TCA prescription in addition to other antidepressants. TCAs are generally not recommended as first-line treatments in this age group owing to their anticholinergic effects and the potential for adverse reactions such as falls and fractures [42]. In a separate analysis, we found that long-term tricyclic antidepressant use longer than 2 years was associated with a 1.73-fold increased odds of incident dementia after adjustment for multiple comorbidities. This association between anticholinergic antidepressants and dementia aligns with previous longitudinal cohort studies indicating that anticholinergic medications negatively affect cognitive performance [15, 31, 32, 43, 44]. These findings highlight the urgent need for greater awareness among prescribers regarding the potential long-term cognitive risks associated with these drugs. To assess dose–response relationship for anticholinergic drugs, such as TCAs, we compared long-term and short-term tricyclic antidepressant use. The odds ratio is almost identical to the odds ratio for general antidepressant use comparing long-term versus short-term use (OR 1.23). However, it did not reach the level of statistical significance owing to lack of statistical power in this subgroup analysis. We defined short- and long-term antidepressant use according to guidelines of depression (see 3) to estimate duration of drug exposure. However, other studies have used standardized daily doses, which emphasize the cumulative amount of drug consumed rather than exposure duration [31]. Future studies should apply both approaches, for instance, examining daily doses across different time trajectories, which may enhance our understanding of the antidepressant–dementia relationship. Previous studies have explained the mechanism of anticholinergic drugs on cognitive functioning, primarily by inhibiting the action of acetylcholine at its receptor sides [43]. However, given that the effect of any antidepressant drug was similar to TCA use, our results suggest that other mechanisms may also contribute to the observed association between long-term antidepressant use and dementia risk. Long-term antidepressant use may directly influence neurodegenerative processes through alterations in brain chemistry or plasticity [45, 46], especially in the context of age-related vulnerabilities [15]. Indirect factors, such as polypharmacy and its associated risks, may also play a role, given the high prevalence of comorbidities and concurrent medication use in the elderly population [16]. Although we controlled for various comorbidities, further research should explore the interactions between multiple medications within a longitudinal case–control design to clarify the mechanisms underlying the negative effect of antidepressants on dementia risk. A major strength of this study is the use of a large, nearly representative dataset with detailed prescription and diagnostic records, enabling robust longitudinal analyses. The application of propensity score matching enhanced the methodological strength by reducing confounding effects. The comparison between long-term and short-term users facilitated the examination of varying levels of exposure, thereby indicating an exposure–response relationship. Nonetheless, notable limitations remain. First, the observational nature of the study precludes causal inference, and residual confounding may persist despite matching. Additionally, lifestyle factors (e.g., smoking, physical activity), genetic predisposition, and psychosocial variables were not accounted for, which could influence dementia risk. Second, dementia ascertainment may have been incomplete. Prodromal neuropsychiatric symptoms may occur several years before formal diagnosis, and antidepressant initiation may therefore partly reflect underlying disease in its early stages. While antidementia medication records allowed partial capture of dementia cases beyond the inpatient setting, outpatient diagnoses were unavailable, and inpatient diagnoses likely reflect later disease stages. In addition, medication-based ascertainment likely captures specific dementia subtypes, potentially underrepresenting conditions such as frontotemporal dementia. Furthermoe, similar to dementia ascertainment, potential confounding related to depression should be considered, as depression represents the primary indication for antidepressant use. Depression was identified using inpatient diagnoses only, likely capturing more severe cases while missing individuals treated exclusively in outpatient settings. Therefore, we cannot exclude that these factors may have influenced the observed associations. Last, restricting the cohort to individuals continuously enrolled throughout the study period resulted in a survivor population, as individuals who died were excluded. However, this restriction was a direct consequence of the study’s explicit focus on long‑term exposure, which required continuous enrolment and sufficient follow‑up. This may have influenced the observed outcomes and limited the generalizability of the findings to the broader older adult population. Conclusions Our study highlights an association between long-term antidepressant use and an increased dementia risk in older adults. This finding applies to both use of TCAs and other antidepressant drug classes, indicating that anticholinergic effects are not the sole putative causal mechanism. Our findings emphasize the importance of weighting potential cognitive risk when considering prolonged antidepressant treatment in aging populations. Further research is essential to explore the underlying mechanisms of long-term antidepressant use and to guide clinical decision-making regarding antidepressant use in aging population. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 161 KB) Acknowledgements Not applicable. Funding The Helsana Health Insurance AG provided support in the form of salaries for the authors (M.A.A., M.N., R.F., C.A.H.) and had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Declarations Ethics Approval and Consent to Participate According to the Swiss national ethical and legal regulations, ethical approval from the local ethical committee of the canton Zurich was not needed for the study as all data were de-identified. Written informed consent from the study population was waived because of the retrospective nature of this study. Consent for Publication Not applicable. Availability of Data Owing to Swiss law restrictions, data are not publicly available. Insurance claims data from Helsana Group underlie protection and privacy restrictions but are available upon reasonable request from Helsana Department of Health Sciences (C.A.H.). Competing Interests The authors declare that they have no competing interests. Authors' Contributions M.A.A. contributed to conception and design, analysis, and interpretation of data, and preparation of the manuscript; M.P.H. contributed to interpretation of data and preparation of the manuscript; M.N. acquisition of data, contributed to conception and design, analysis, and preparation of the manuscript; R.F. contributed to interpretation of data and preparation of manuscript; C.A.H. contributed to conception and design and preparation of the manuscript. All authors read and approved the final version. References Associated Data Supplementary Materials Supplementary file1 (DOCX 161 KB)