The recent nationwide study investigating factors linked to high caregiver burden among informal caregivers in psychiatry presents crucial findings that demand both recognition and critical engagement. The authors' documentation that 54.3% of caregivers experience a high burden is a stark figure, powerfully underscoring a systemic and often neglected dimension of mental healthcare (Thygesen et al. 2025). While the study makes a significant descriptive contribution, several methodological and interpretative choices merit deeper examination to ensure the robustness of its conclusions and their utility in guiding targeted interventions. A primary concern centres on the transparency and interpretation of the reported adjusted odds ratios (aORs). The manuscript notes adjustments were made in multivariate models but omits a full specification of the covariates included. This lack of detail is consequential. For example, the conclusion that parental caregiver status is an independent factor associated with burden—with aORs of 1.80 for outpatient and 3.81 for inpatient settings—cannot be fully evaluated without knowing which potential confounders were accounted for. Critical variables such as the caregiver's own psychological distress, caregiving duration, co-residency, patient functional impairment or weekly hours of direct care may disproportionately characterise parental caregivers. Their omission raises the possibility that the identified ‘parental’ effect acts as a proxy for these unmeasured, burdensome caregiving contexts rather than representing the intrinsic burden of parenthood itself. The lower odds of burden among non-first-degree relatives further complicates this relationship. As VanderWeele argues, rigorous confounder selection is paramount for valid causal inference from observational data, even after adjustment (VanderWeele 2019). The authors' insightful thematic analysis suggests parents are impacted across all four burden domains; this very pervasiveness could represent the mediating pathway through which parental status operates. A more nuanced analytical approach, such as mediation analysis, would help disentangle whether parental status exerts influence through these specific challenges or remains an independent risk factor after they are accounted for. Second, the acknowledged cross-sectional design fundamentally constrains the interpretation of the cited associations. Capturing burden and its correlates at a single timepoint renders the determination of temporality and causality speculative. A key question remains unanswered: does caring for a patient with an eating disorder, for instance, lead to higher burden, or are caregivers with pre-existing vulnerabilities—such as higher anxiety—both more likely to experience severe burden and to have a care recipient currently engaged in treatment, thus being captured by the survey? Furthermore, the reliance on the Danish National Survey of Psychiatric Patient Experiences risks substantial selection bias, a point the authors aptly note regarding caregivers of patients with severe schizophrenia. This underrepresentation is not merely a sampling nuance but a critical threat to the validity of comparisons across diagnostic groups. It offers a plausible, alternative explanation for the counterintuitive finding of lower burden associated with schizophrenia relative to affective or eating disorders. The post hoc explanation involving Denmark's specialised early intervention services, while reasonable, remains speculative without empirical support within the study's design. A more rigorous assessment would employ sensitivity analyses or formal quantitative bias analysis to estimate the potential impact of such missing data on the effect estimates (Lash et al. 2014). Additionally, the dichotomisation of ‘high caregiver burden’ using a cut-off point, noted as a source of discrepant prevalence rates across studies, represents a significant statistical simplification. Dichotomisation reduces statistical power and can distort the magnitude of associations, as the choice of cut-off is often arbitrary and influential (Altman and Royston 2006). Analysing burden as a continuous measure or employing methods like ROC-derived thresholds would yield more reliable and informative evidence. The thematic analysis, while valuable, remains curiously isolated from the quantitative results. A more integrated mixed-methods framework could have significantly strengthened the study's explanatory power. For example, quantifying the frequency or intensity of themes like ‘Encountering the health care system’ across different caregiver subgroups could have provided direct empirical support for the hypothesis that parents are burdened across all domains. In its current form, this remains an untested proposition within the dataset. Moreover, the emergence of healthcare system interactions as a core burden theme reveals a critical gap in the quantitative model: the absence of system-level variables as potential confounders or effect modifiers. Caregiver burden is undoubtedly modulated by the quality of support, communication and collaboration with professional services. The omission of covariates such as perceived clinician communication, degree of involvement in care planning or access to caregiver-specific resources may lead to an overestimation of effects attributed solely to patient diagnosis or familial relationship. In conclusion, the study serves as a vital alarm regarding the severe toll on informal caregivers within psychiatric care. To advance from identifying associations to informing precise, actionable interventions, future research must build upon this foundation with enhanced methodological rigour. Key recommendations include: Full transparency in statistical reporting, detailing all adjusted covariates and considering advanced techniques like propensity score matching to better mitigate confounding between caregiver groups (Austin 2011). Explicit acknowledgement of the limits of cross-sectional data for causal inference, advocating for longitudinal designs that track the dynamic interplay of burden, patient illness trajectory and caregiver support over time. Proactive investigation of selection biases, potentially through data linkage with national registries to characterise non-respondents. Employment of a truly integrated mixed-methods design where qualitative insights directly shape and are tested within quantitative models. Interventions, particularly for highly burdened parental caregivers, will be most effective if grounded in a nuanced understanding of whether the primary levers for reduction lie in emotional support, practical respite, systemic advocacy skills or a combination thereof. This study provides an essential, although preliminary, map of this challenging terrain. The authors have nothing to report. The authors declare no conflicts of interest. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.
Xie et al. (Wed,) studied this question.