To achieve our objectives, we carried out a descriptive, observational, cross-sectional study, which received approval from the Ethics Committee of the University of Salamanca (Record No. 1066). All procedures adhered to the latest revision of the Declaration of Helsinki (17). Data collection was conducted between January 2025 and July 2025 in Salamanca (Spain).

Informal caregivers of individuals diagnosed with dementia (including Alzheimer’s disease and related dementias) or cerebral palsy were recruited to participate in this study. Recruitment was facilitated through a collaboration agreement with the State Reference Centre for the Care of People with Alzheimer’s Disease and Other Dementias (CREA) of the Institute for the Elderly and Social Services (IMSERSO) in Salamanca, and with the Association of Parents of People with Cerebral Palsy and Related Encephalopathies of Salamanca (ASPACE), Spain.

Eligible participants were adults (≥18 years) who acted as the primary informal caregiver of a person with dementia or cerebral palsy and had at least 1 year of caregiving experience in that role. Caregivers were excluded if they were illiterate or otherwise unable to complete the interview process, or if they served as secondary or formal caregivers.

A non-probability consecutive sampling strategy was used. Participants were recruited among informal caregivers of users attending the collaborating institutions (CREA-IMSERSO and ASPACE) in Salamanca, Spain. Recruitment and data collection took place between January 2025 and July 2025.

Potential participants were contacted by institutional staff via invitation letters and follow-up phone calls. A total of 100 caregivers were contacted, and 80 underwent eligibility screening. Of these, 73 met the inclusion criteria and completed the scheduled interview comprising the final analytical sample (n = 73, complete data). The remaining 7 caregivers did not attend the scheduled interview and were excluded from the study.

Each participant was individually invited to a room equipped for testing under controlled environmental conditions (e.g., lighting, temperature, noise). An identification code was assigned to each participant to preserve anonymity and ensure confidentiality. Subsequently, a digital questionnaire comprising three sections was completed: (1) an information sheet outlining the study, including a digital informed consent form; (2) a structured set of questions to gather relevant sociodemographic data (e.g., age, gender); and (3) standardized questionnaires for the assessment of the following study variables:

Given the observational design and consecutive recruitment from a finite pool of eligible caregivers attending the collaborating institutions during the study period, the sample size was determined pragmatically by feasibility. A sensitivity power analysis was conducted to evaluate whether the achieved sample was sufficient to detect associations of interest. With n = 73, a two-tailed α = 0.05 and 80% power, the study was able to detect a minimum Pearson correlation of approximately |r| = 0.32, corresponding to a moderate effect size.

Statistical analysis was performed using IBM SPSS Statistics software (version 28). A statistical significance level of p < 0.05 was established for all tests.

A total of 73 informal caregivers participated in the study. The mean age of caregivers was 61.1 ± 13.6 years, and the average weekly caregiving time was 82.2 ± 52.7 h. The mean functional dependence of the care recipient was 2.75 ± 0.7, indicating a moderate level of dependence. The mean ZBI total score was 38.6 ± 13.6, suggesting a moderate-to-high caregiver burden. The average mental health component of the SF-36 was 60.7 ± 19.9, reflecting an overall satisfactory mental well-being. Table 1 details the sociodemographic characteristics of the sample.

Spearman correlation analyses revealed significant associations between caregiver burden and several independent variables. Higher burden scores were correlated with poorer mental health (r = −0.55, p < 0.001), and higher anxiety (r = 0.48, p < 0.001) and depression (r = 0.34, p = 0.003) scores. In addition, greater musculoskeletal pain intensity was associated with higher levels of perceived burden with a statistical significance of p < 0.05 in all regions evaluated. These findings indicate that both physical and psychological dimensions are relevant contributors to caregiver stress. However, no statistically significant correlation was observed between patient dependence and caregiver burden.

After conducting descriptive analyses, hierarchical multiple regression models were performed with the ZBI score as the dependent variable.

To examine whether caregiver mental health mediated the relationship between patient dependence and caregiver burden, the Baron and Kenny causal-step approach was applied (Figure 1).

Dependence was marginally associated with caregiver burden (β = 0.20, p = 0.089), indicating a weak positive relationship. However, dependence was not statistically significantly associated with caregiver mental health (β = −0.04, p = 0.713), failing to meet the prerequisite for mediation.

Mental health was strongly and inversely associated with caregiver burden (β = −0.54, p < 0.001), suggesting that poorer mental health corresponded to higher burden levels. When both variables were included in the same model, the direct effect of dependence on burden decreased slightly (β = 0.18, p = 0.075) but remained nonsignificant.

Therefore, the indirect effect through mental health was not supported by the data. The results indicate that although caregiver mental health is a strong correlate of burden, it does not act as a mediator between patient dependence and burden. Instead, mental health showed an independent association, largely explaining variance in burden irrespective of the patient’s functional status.

Given the absence of mediation, a moderation model was tested to explore whether caregiver mental health moderated the relationship between patient dependence and caregiver burden.

In Model 1, patient dependence and caregiver mental health were entered as main effects and explained 32.7% of the variance in caregiver burden (R² = 0.327, F (2, 68) = 16.49, p < 0.001).

In Model 2, the interaction term (Dependence × Mental Health) was added; however, it did not significantly improve the model fit (ΔR² = 0.024, ΔF (1, 67) = 2.467, p = 0.121). The interaction term was not statistically significant (β = 0.203, p = 0.121), indicating that caregiver mental health did not moderate the relationship between patient dependence and caregiver burden.

Graphically (Figure 2), the slope of the dependence-burden relationship was steeper among caregivers with poorer mental health, whereas it was flatter among those with higher mental health scores.

Although the graphical pattern is suggestive of a potential buffering trend, the interaction term did not reach statistical significance.

Correlations between caregiver burden (ZBI) and psychological variables were moderate (HADS-Anxiety: r = 0.488; HADS-Depression: r = 0.342; SF-36 Mental Health: r = −0.553), suggesting related but non-redundant constructs. Multicollinearity diagnostics showed acceptable VIF values (range 1.026–2.407) and tolerance values above 0.40, indicating absence of problematic construct overlap. The maximum condition index was 22.17, below commonly accepted thresholds.

Combining the hierarchical, mediation, and moderation analyses provides a nuanced understanding of the mechanisms underlying caregiver burden:

Physical symptom associations: Caregiver physical pain contributes directly to burden levels, representing the objective dimension of care demands.

The findings of this study highlight the pivotal role of caregiver mental health as a factor independently associated with perceived burden of caring for people with dementia or cerebral palsy, beyond the objective characteristics of caregiving, such as patient dependence or the physical demands of care. Clinically, this suggests that interventions aimed solely at improving functional outcomes in care recipients or reducing the physical load of caregiving may be insufficient to alleviate caregiver distress. Instead, a multidimensional approach that integrates psychological assessment and mental health support should be incorporated into caregiver care programs. Routine screening for anxiety and emotional well-being in informal caregivers could facilitate early identification of individuals at higher risk of burden and promote timely psychosocial or therapeutic interventions.

A multidimensional approach to caregiver support—integrating emotional, psychosocial, and practical components—has been widely emphasized in dementia care guidance and evidence syntheses (34). Therefore, we do not present this concept as novel. Instead, the contribution of the present study is to provide an integrated multivariable assessment of factors associated with caregiver burden by jointly modelling care-recipient functional severity indicators (dependency/disability) together with caregivers’ region-specific musculoskeletal pain and mental health. This framework helps identify which caregiver-related, potentially modifiable factors show the most robust associations with burden in our sample and may inform prioritization of screening and targeted support.

These implications are particularly relevant given that our caregiver profile reflects an aging caregiver population. Older caregivers may have lower physical reserve, more chronic comorbidities, and a higher risk of musculoskeletal problems, which can compound role-related strain. In practice, this supports routine screening of caregiver mental health and musculoskeletal symptoms within dementia and disability services, with clear referral pathways to psychosocial support and rehabilitation/ergonomics training (e.g., safe transfer techniques, activity pacing, and home-based exercise). From a policy perspective, older caregivers may benefit disproportionately from accessible respite services, home support, and assistive devices/home adaptations that reduce physical load. Service delivery should also consider barriers common in older adults (time constraints, mobility limitations, and potential digital exclusion), favoring flexible options (in-person/community-based programs, telephone support) and integration across health and social care.

From a scientific standpoint, the potential buffering pattern observed graphically highlights the importance of considering psychological resilience in models of caregiver strain, even though the moderation effect did not reach statistical significance. The results also support the inclusion of mental health variables as central components in exploratory models of caregiving outcomes, providing an evidence base for holistic public health policies designed to sustain informal caregiving networks and prevent caregiver burnout.

This study has several limitations that should be acknowledged. First, its cross-sectional design precludes causal inferences regarding the directionality of the observed associations. Longitudinal studies are needed to determine whether poorer mental health leads to higher burden, or whether chronic burden progressively deteriorates psychological well-being. Second, the sample size was modest and limited to a specific geographical area, which may restrict the generalisability of findings to other populations or caregiving contexts. Although the sample size was modest, the number of predictors was restricted and consistent with recommended subject-to-variable ratios. Furthermore, multicollinearity diagnostics did not indicate instability of regression coefficients. Third, the absence of a control group of non-caregivers prevents comparisons that could contextualize the magnitude of caregiver-specific effects.

An additional limitation of this study is the inclusion of informal caregivers of people with dementia and of people with cerebral palsy within a single analytical framework. Although both conditions involve substantial caregiving demands, the caregiving populations may differ in terms of age, familial role, and life-course context. While our analyses focused on caregiver-related factors that cut across diagnoses, these differences may limit the generalizability of the findings to specific caregiving groups.

Unmeasured and residual confounding is also likely. Several key care-recipient indicators were caregiver-reported, including dependence grade and disability percentage (based on the most recent official social services assessment), which may introduce measurement error and attenuate associations with burden. In addition, we did not include objective clinical severity measures or standardized assessments of behavioral and cognitive symptoms (e.g., neuropsychiatric/behavioral disturbances in dementia), communication problems, or functional performance measures that are known to influence caregiver strain. Finally, other potentially relevant caregiver-level confounders were not measured (e.g., comorbidity burden, sleep quality, coping strategies, social support, and socioeconomic constraints), which limits our ability to fully adjust the multivariable models.

Finally, caregivers were recruited through two institutions using a non-probability consecutive sampling approach. Although eligibility screening and consecutive recruitment were applied within a predefined time window, caregivers connected to institutional services may differ from those not engaged with such resources, which may limit representativeness and introduces potential selection bias.

Future research should adopt longitudinal or mixed-methods designs to explore the temporal dynamics between mental health, physical strain, and caregiver burden, thereby clarifying causal mechanisms. Interventional studies assessing the effectiveness of integrated programs that combine psychological support, stress management, and physical health promotion are warranted to confirm the clinical utility of these findings. Additionally, biomarkers of stress, sleep quality, or inflammation could also provide an objective understanding of the psychophysiological impact of caregiving. Finally, future models should examine resilience, coping strategies, and social support as potential protective factors capable of moderating the burden experienced by informal caregivers for people with dementia and cerebral palsy.