Reply

We thank Kukreja and colleagues for taking the time to comment on our article on the interaction between mental disorders and social disconnectedness on mortality.^{Reference Laustsen, Ejlskov and Chen1} We are encouraged by the recognition of the strengths of our work, and we welcome the opportunity to clarify certain methodological details and address some of the raised concerns.

First, Kukreja and colleagues pointed out several methodological issues that call for clarification. Individuals with organic mental disorders and intellectual disabilities were not excluded from the study population; thus, our sample was still population representative. Similarly, anxiety and stress-related disorders are not included in the Nordic Multimorbidity Index^{Reference Kristensen, Lund and Jensen2} and their exclusion from this index is thus not a concern. Furthermore, given that the risk of suicide increases substantially immediately after psychiatric hospital in-patient discharge,^{Reference Chung, Ryan, Hadzi-Pavlovic, Singh, Stanton and Large3} we do not deem Neyman's bias, which is also called survival bias, redundant but rather an important limitation inherent to the applied study design.

Second, we acknowledge Kukreja and colleagues’ concern about applying parental values as proxies for individuals aged below 30 years. However, we made this analytical choice in our pre-registered analysis plan^{Reference Laustsen, Lasgaard and Ejlskov4} due to the inverse correlation between age and income. Individuals who complete a higher education will typically achieve a high income relatively late in life with the average age at completion of such education ranging from 26.6 to 29.3 years in Denmark.⁵ We welcome further discussion of adjustment procedures for socioeconomic position, as this is one of the main interests of our research group.

Third, Kukreja and colleagues correctly observe that we did not describe the proportion of survey responses obtained through electronic and print media in our article. However, the Danish National Health Survey is often used for research purposes, and these proportions as well as other detailed information are available in a referenced methodology article.^{Reference Christensen, Lau and Kristensen6} Likewise, the authors raise an important discussion regarding the quality and relevance of survey data in psychiatric epidemiology. Given the breadth of our study, we were not able to validate the applied survey items for individuals with specific categories of mental disorders. However, further studies on the validity of applied methodologies in survey-based psychiatric research would be of great relevance.

Fourth, Kukreja and colleagues raise a concern over the generalisability of our results due to the inclusion of regional samples rather than national data. However, the four regions covered by our study constituted 79% of the Danish population in the first quarter of 2017,⁷ and we have no reason to believe that the fifth region (Region of Southern Denmark) is substantially different. The only reason for not including data from the fifth region was that information on social connections in the Danish National Health Survey was not available. Nevertheless, we agree that generalisability is an important issue for survey-based studies, given potential selection in survey participation. Additionally, our assessment of mental disorders did only include hospital-based diagnoses within the specified categories, and the generalisability of our findings should thus be considered in this light. Survey-based research has the potential to curb this limitation via diagnostic questionnaires and thus provide many future opportunities for psychiatric epidemiology, despite inevitable concerns regarding selection given voluntary participation.^{Reference Plana-Ripoll, Lasgaard, Mneimneh and McGrath8} Thus, continuing discussion and methodological developments on the possibilities and limitations of applying survey data for psychiatric research will be an essential step for advancing the quality of future research in this field.