This cross-sectional analysis utilised data from the WHO Integrated Care for Older People (ICOPE) pilot in Lianyungang City. Between March and April 2024, we enrolled adults aged ≥60 years residing locally for ≥6 months (community-dwelling and institutionalised populations) via convenience sampling at community health centres and long-term care facilities. Inclusion required: (1) baseline communication capacity for assessment completion; and (2) family member consent for participation support. Exclusion criteria comprised: (1) acute conditions necessitating urgent intervention; (2) incomplete baseline data; or (3) terminal illness with <6-month life expectancy. Of 509 screened participants, 467 met eligibility criteria and were analysed (Figure 1).

This study protocol was reviewed and approved by the Ethics Committee of Lianyungang First People’s Hospital on December 9, 2022, with ethics number QT-20221118001-02. The study complied with STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. Trained researchers administered pretested structured questionnaires using validated instruments. Rigorous quality control measures included: dual independent data entry, logic consistency validation, and outlier detection protocols throughout data collection.

Intrinsic capacity (IC) was assessed using the WHO ICOPE screening tool (3), comprising six domains: cognition, mobility, nutrition, vision, hearing, and psychological status. The reliability and validity of the tool have been well verified (14). Participants with positive initial screening results underwent further clinical evaluation. Cognition was quantified via the Mini-Mental State Examination (MMSE; score range 0–30), with impairment thresholds adjusted for educational attainment: ≤17 for illiterate individuals, ≤20 for primary school education, ≤22 for middle school, and ≤23 for university-level education.

Mobility assessment employed the Short Physical Performance Battery (SPPB), where scores ≤9 indicated impairment. Nutritional status was evaluated using the Mini Nutritional Assessment-Short Form (MNA-SF), defining malnutrition as MNA-SF scores ≤11 (15).

Those with positive results of visual acuity and hearing were further evaluated by comprehensive geriatric assessment tools commonly used by famous medical institutions in China (e.g., Peking Union Medical College Hospital and Beijing Geriatric Hospital). In-depth evaluation of visual acuity included the following: difficulty in reading/walking/watching TV, or occlusion or defect of vision, distortion of vision. A “no” score of 0 (good vision) was assigned for all three items, a “yes” score of 1 (poor vision) was assigned for 1–2 items, and a “yes” score of 2 (bad vision) was assigned for all three items. An in-depth hearing assessment includes the following questions: complaints about the volume of the television, the need for repetition, and difficulty listening to the phone. A “no” score for all three items was 0 (good hearing), a “yes” score for 1–2 items was 1 point (poor hearing), and a “yes” score for all three items was 2 points (bad hearing) (16).

According to the ICOPE guidelines (WHO, 2019) (17), the Psychological capacity dimension focuses on managing depressive symptoms. Its screening criteria are clear: If an individual has experienced any of the symptoms of “feeling down,” “feeling depressed,” “feeling hopeless” or “lack of interest/pleasure” in the past two weeks, they can be defined as having depressive symptoms.

Calculation of composite IC score: A composite IC score (range: 0–8) was derived by summing the scores assigned to each of the six domains. Cognition, mobility, nutrition, and psychological status (depressive symptoms) were scored dichotomously (0 = no impairment, 1 = impairment). Vision and hearing were scored on a three-level scale (0 = good, 1 = moderate impairment, 2 = severe impairment). IC decline (DIC) was defined as a composite score ≥ 2, while scores <2 denoted normal IC (non-DIC).

Polypharmacy is defined as the prescription of five or more different medications taken daily. The types of medications included in polypharmacy encompass both prescription and over-the-counter drugs (18), while herbal remedies and supplements were excluded. During the baseline survey, the number of medications taken by patients was assessed by asking, “What medications have you been taking daily for the past three months?” to gather information on their recent medication use.

Key sociodemographic and clinical covariates were adjusted for in the analyses. Sociodemographic factors encompassed: age (<80 vs. ≥80 years), marital status (married/unmarried), educational attainment (illiterate/primary school vs. ≥junior high school), residential setting (urban/suburban), nursing home residence (yes/no), and caregiver availability (yes/no). Clinical covariates incorporated frailty status, social functioning (assessed via Social Participation Frequency, SPF), basic activities of daily living (ADL), urinary incontinence (UI), and Charlson comorbidity index (CCI) scores.

Frailty was assessed using the K-FRAIL scale (19), comprising five criteria: fatigue, resistance, ambulation, illnesses, and weight loss (20). Each criterion was scored dichotomously (0/1), with total scores categorised as: robust (0), pre-frail (1–2), or frail (≥3).

Social Participation Function (SPF) was evaluated through a validated instrument assessing five domains (21): appropriate social adaptation (score 0), adaptation to simple environments with undetectable cognitive issues at initial contact (score 1), social withdrawal with passive interaction, impaired initiation, and inappropriate remarks increasing deception vulnerability (score 2), limited interactions with unclear speech/inappropriate expressions (score 3), markedly impaired social engagement (score 4), total SPF scores defined functional levels: intact (0–2), mild (3–7), moderate (8–13), or severe impairment (14–20).

Activities of daily living (ADL) were measured using the modified Barthel index (MBI; range 0–100), classifying dependency as: complete independence (100), mild (91–99), moderate (61–90), severe (21–60), or total dependence (≤20).

The International Consultation on Incontinence Questionnaire-Urinary Incontinence Short Form (ICIQ-UI SF) assesses the severity of urinary incontinence (UI) symptoms. It comprises three questions: Question 1 quantifies the frequency of urine leakage, Question 2 evaluates the amount of leakage, and Question 3 measures the impact of UI on daily life. The total ICIQ score ranges from 0 to 21. A score of 0 indicates no incontinence; 1–5: mild incontinence; 6–12: moderate incontinence; 13–21: severe incontinence (22).

Comorbidity burden was quantified using the Charlson comorbidity index and Cumulative Illness Rating Scale-Geriatrics (CIRS-G), supplemented by clinical expertise. Seventeen disease categories (e.g., myocardial infarction, chronic pulmonary disease, malignancies) were graded 1–6 based on severity. Total scores ≥1 indicated comorbidity presence, with higher scores reflecting greater burden (0 = no comorbidity) (23).

Descriptive statistics characterised baseline participant characteristics: normally distributed continuous variables as mean ± standard deviation (SD), non-normally distributed variables as median (interquartile range, IQR), and categorical variables as frequency (percentage). Group comparisons used independent samples t-tests or Mann–Whitney U tests for continuous variables, and χ² or Fisher’s exact tests for categorical variables.

Multivariable logistic regression examined polypharmacy (≥5 medications)-intrinsic capacity (IC) associations. To address confounding and validate robustness, covariates were selected through: univariate analysis with p < 0.05 (Supplementary Table 1); ≥10% change-in-estimate criterion; clinically relevant variables per expert consensus. Four adjusted models generated odds ratios (ORs) with 95% CIs. Subgroup analyses assessed associations across strata. We employed the nonparametric Kruskal–Wallis H test to compare differences in medication counts among the high-, medium-, and low-level intrinsic capability (IC) groups. Given minimal missing data (0–6%), case deletion was applied. All analyses were conducted using the statistical software packages R (http://www.r-project.org, The R Foundation) and Free Statistics software versions, with a two-sided p-value <0.05 considered statistically significant.

The cohort comprised 467 older adults (Figure 1), stratified into non-polypharmacy (<5 medications, n = 434) and polypharmacy (≥5 medications, n = 33) groups. The prevalence of DIC was 67.9% (317/467). Compared with the non-polypharmacy group, the polypharmacy group demonstrated significantly higher proportions of comorbidities, frailty, urban residency, vision disorder and married status (p < 0.05; Table 1). IC scores were marginally elevated in polypharmacy users (mean ± SD: 4.0 ± 1.5 vs. 3.5 ± 1.5; p = 0.054), with significantly increased incidence of IC decline (84.8% vs. 66.6%; p = 0.03). No significant intergroup differences were observed in: sex distribution, age stratification, BMI, Mini-Mental State Examination (MMSE) scores, malnutrition, depression, SPF, sleep quality, and ADL levels (all p > 0.05).

Table 2 presents the multivariable regression analyses, which indicate a positive association between the number of medications and decline in intrinsic capacity (IC). When treated as a continuous variable, each additional medication was associated with greater odds of IC decline in the crude model (OR = 1.29, 95% CI: 1.10–1.52, p = 0.002). After sequential adjustment for demographic characteristics (Model 1: OR = 1.28, 95% CI: 1.09–1.51, p = 0.003), socioeconomic factors (Model 2: OR = 1.36, 95% CI: 1.15–1.62, p < 0.001), functional status indicators (Model 3: OR = 1.30, 95% CI: 1.08–1.55, p = 0.005), and comorbidities (Model 4: OR = 1.22, 95% CI: 1.01–1.47, p = 0.04), these associations remained statistically significant.

When stratified by medication threshold (with <5 medications as the reference group), polypharmacy (≥5 medications) was significantly associated with a decline in intrinsic capacity (IC) in the unadjusted model (OR = 2.81, 95% CI: 1.06–7.43, p = 0.037). This association was maintained in Model 1 (OR = 2.82, 95% CI: 1.05–7.56, p = 0.039) and further amplified in Model 2 (OR = 3.65, 95% CI: 1.33–9.98, p = 0.012). However, in Model 4, after adjusting for comorbidities, the odds ratio decreased to 2.31 (95% CI: 0.77–6.88, p = 0.134), no longer reaching statistical significance.

In Supplementary Table 2, we demonstrated a significant association between medication burden and vision impairment. Each additional medication increased the odds of vision impairment by 43% (OR = 1.43, 95% CI: 1.24–1.63, p < 0.001). Likewise, polypharmacy was significantly associated with elevated odds of vision impairment (OR = 3.76, 95% CI: 1.83–7.75, p < 0.001). Concurrently, a positive trend was observed between increasing medication count and impairments in locomotion, nutritional risk, and hearing, although these associations failed to reach statistical significance.

Figure 2 demonstrates significant differences in the distribution of medication numbers among different intrinsic capacity (IC) level groups (Kruskal–Wallis H = 12.80, p < 0.05). The high IC group (<2 points) had the lowest median number of medications (1.6), followed by the medium IC group (2–3 points) with 1.9 medications, and the low IC group (≥4 points) had the highest (2.2 medications), demonstrating an increasing trend in medication numbers with decreasing IC levels.

We conducted subgroup analyses to evaluate potential effect modifiers in the medication count-diminished intrinsic capacity association. Participants were stratified by age (<80 vs. ≥80 years), sex (female/male), marital status (married/unmarried), education level, residential area, and institutional care status. Forest plots were generated to visualise trend consistency across strata. Notably, no significant interaction effects were detected for any subgroup variable (all interaction p-values >0.05; Figure 3).

Methodological strengths of this study include its integration with the World Health Organization (WHO) Integrated Care for Older People (ICOPE) pilot program. By rigorously adhering to the ICOPE protocol and including both community-dwelling and nursing home residents, we established a dose–response relationship between the number of medications and intrinsic capacity (IC). However, the following limitations should be noted. The cross-sectional design precludes causal inference regarding the relationship between polypharmacy and IC. Due to the post-hoc nature of the subgroup analyses, the relatively small sample size in the polypharmacy subgroup limited the statistical power of some analyses, and the results warrant cautious interpretation. Potential multicollinearity among covariates (such as comorbidity burden) may have been present; the observed effect of polypharmacy was attenuated after adjusting for comorbidities in the multivariate regression analysis. Furthermore, reliance on self-reported medication data (susceptible to recall bias) and the lack of prescription details (e.g., drug classes, potentially inappropriate medications) may have influenced the findings. Although we adjusted for key covariates including age, comorbidities, and functional status, the potential impact of unmeasured confounders must be considered. Factors such as social support, lifestyle and diet, and medication adherence could concurrently influence both prescribing patterns and IC trajectories (40). Future research should employ multi-center, prospective, large-sample cohort designs, incorporate a broader range of variables, and track IC trajectories over time to further elucidate the causal relationship and underlying mechanisms between polypharmacy and intrinsic capacity.