Response

We read the article titled ‘Interaction between mental disorders and social disconnectedness on mortality: a population-based cohort study’, with great interest.^{Reference Laustsen, Ejlskov, Chen, Lasgaard, Gradus and Østergaard1} The study provided insights into the intricate relationship among mental illness, social disconnectedness and mortality. A comprehensive approach and appropriately conducted sophisticated statistical analysis substantially contributed to our understanding of the complex inter-relationship amongst these three variables. An important strength of the study worth mentioning is using a large database by linking the population-based Danish National Health Survey^{Reference Christensen, Lau, Kristensen, Johnsen, Wingstrand and Friis2} with the comprehensive National Register Data from Denmark. In addition, validated tools were used for assessing various variables of social-connectedness in contrast to previous studies. The authors used impressively rigorous statistical analysis to untangle the complex interactions between various variables and adjusted for a number of confounders (such as multimorbidity and socioeconomic status), and imputing for missing data, over a long period of follow-up. One of the most significant contributions of the study is the finding of gender disparities in mortality rates, particularly the increased vulnerability of men experiencing both mental disorders and diminished social connections. This observation opens up important avenues for future research and the development of targeted interventions.

While the study has many strengths, there are areas that could benefit from further elaboration. The distinction between data collected directly by the authors and data sourced from the Danish National Health Survey and Registers is not entirely clear. For example, were any questionaries sent along with the consent to request participation or analysis of the participants’ data. This makes us wonder as a reader if it was secondary analysis of already existing data or also linking of new data with already collected survey data and registers. Further, the proportion of responses obtained through electronic and print media remains unclear. The absence of discussion regarding the potential impact of different data collection methods (such as online versus paper-based surveys) is another methodological aspect that warrants attention. Variations in data collection techniques could introduce biases that may influence the study's outcomes, and addressing this could enhance the credibility of the analysis.

Another important observation includes use of parents’ socioeconomic data as proxy for participants under 30 years of age. The cut-off of 30 years remains a question and thus raises doubts on actual reflection of the participants’ own socioeconomic status. A clearer rationale for this methodological choice would help in our understanding regarding the study findings. As readers, we appreciate the authors’ concerns for excluding organic mental disorders and intellectual disabilities for improving the validity of responses, but the uncertainty still remains regarding the validity of responses from other participants, about whom the clarity on being symptomatic and having cognitive decline remains unclear. Similar challenges with respect to the validity of responses hold true for other psychiatric illnesses as well. Moreover, at what stage of data collection the above-mentioned disorders were excluded is not entirely clear from the flowchart that the authors provided within the study.

The authors provided extensive data and details in their supplementary material but the clarity of calculating the somatic multimorbidity index remains indiscrete. The factors excluded in the calculation of this index included antipsychotics, antidepressants, benzodiazepines and related drugs, along with drugs used in opioid dependence. Similarly, the list of disorders mentioned in the Supplementary material^{Reference Laustsen, Ejlskov, Chen, Lasgaard, Gradus and Østergaard1} include ICD-10³ codes F10 and F17 and prescription codes. However, this list does not seem exhaustive enough to exclude mood stabilisers, or certain illnesses that would not follow any medical prescriptions, such as cannabis dependence, certain anxiety and stress-related disorders. Details on limitations of calculating this multimorbidity index would enrich the knowledge of readers and better the interpretation of the study findings.

Another ambiguity we experience as readers is the strength of ‘no loss to follow-up’ as highlighted in the study. If the study essentially includes secondary analysis of already collected data and linking it to the already established data in national registers, the question of losing data to follow-up becomes redundant. Similarly, the limitation of ‘Neyman bias’ as highlighted by the authors as one of the limitations seems redundant as psychiatric illnesses, particularly after excluding organic mental disorders, are highly unlikely to lead to immediate death post-diagnosis, which is the situation Neyman bias is generally associated with. It could have been noteworthy if the authors highlighted limited generalisability of their findings owing to analysing regional samples instead of national data from the Danish survey in addition to limited generalisability owing to selected psychiatric disorders being excluded.

We understand that such large-scale data-based studies are pragmatically possible only with the help of extensive register-based data that has been comprehensively maintained for decades. However, one must always interpret findings of such studies in the light of certain limitations that go hand-in-hand with these registers, including validity, missing data, possibility of duplicity and errors.^{Reference Laugesen, Ludvigsson, Schmidt, Gissler, Valdimarsdottir and Lunde4–Reference Nickelsen6} While the use of register data offers several advantages, such as large sample sizes and longitudinal tracking, it also introduces limitations. Furthermore, biases related to care-seeking behaviour and the focus on more severe cases in clinical settings may have influenced the results.

In conclusion, we commend the authors for their thorough and innovative approach to this critical topic. The study makes a significant contribution to our understanding of the interplay among mental illnesses, social disconnectedness and mortality. It highlights the need for integrated mental healthcare strategies that address both clinical symptoms and social factors, potentially informing policy and clinical practice. The methodological strengths of this research set a high standard for future studies in this field. We look forward to further research that builds on these valuable findings, while addressing the complexities and potential biases inherent in register-based studies.