Introduction
The detrimental effect of declining social status on health is well known.1,2 Despite numerous difficulties in obtaining exact estimates of socioeconomic status, social status is believed to be fairly constant in Mediterranean countries.3 Several studies from Croatia suggest that isolated island communities may pertain higher levels of social homogeneity, reflected in both a generalized lack of health inequalities,4,5 extending even on to mortality.6 Sadly, these beneficial effects seem to be declining, as new generations most frequently achieve poorer health, diet and mortality profiles.7,8
One of the mechanisms of how social inequalities may be exerting their detrimental effects is stress. The association of psychological distress and increased mortality risk has been a subject of many studies over the years.9 Most studies reported a significant association between mortality and psychological distress,9–14 especially correlated with cardiovascular diseases and cancer.12,15,16 Psychological distress measured by different scales and questionnaires, like GHQ, was often previously found to predict mortality.17,18 GHQ is one of the most commonly used instruments for mental stress estimation.19,20 It was not designed to diagnose specific disorders but to distinguish healthy individuals from ones with self-perceived psychological disorders, especially depressive and anxiety disorders.21 The questionnaire contains questions about the general level of happiness, the experience of depressive and anxiety symptoms and sleep disturbance.10,20 While the original version was made out of 60 questions, several shorter versions were developed, with 30, 28, 20 and 12 questions. In these shorter versions, questions that could be positively answered by people suffering from physical illness were omitted.22 The 30-question version of the questionnaire (GHQ-30) is particularly well designed to diagnose minor psychiatric disorders and uses the best discriminators for establishing psychiatric cases derived from the full version. It contains an equal number of positive and negative questions, indicating health and illness, respectfully.22
Therefore, the aim of this 18-year follow-up study was to investigate how social inequalities alongside psychological distress affect mortality (all-cause, cardiovascular and cancer mortality) in the population of the island of Vis.
Methods
This study was performed within the larger project “10,001 Dalmatians,” which investigates environmental determinants of health and disease in Dalmatia, Croatia.5,23 For the purposes of this study, we included subjects who were recruited from the island of Vis. Sampling was completed in 2003 and 2004. All inhabitants listed in the voting register and general practitioner registries were invited to join the study.
The estimates of the socioeconomic status included years of schooling, subjective material status (classified in three groups; worse than others, same as others or better than others) and a composite material household status (measured as the sum of 16 items that could have been present in the household, including wooden floors, second TV, art collections, over 100 books, etc).4
Psychological distress was measured with the GHQ-30 using the original binary rating model with a four-point response scale 0-0-1-1 (presence of symptom: not at all=0, same as usual=0, more than usual=1, much more than usual=1).12,16 Points were summed for each individual, and the final GHQ score ranged from 0-30. We applied a GHQ-30 cut-off score of ≥5 where all those scoring 0-4 were considered non-distressed, and all those scoring 5-30 were considered psychologically distressed.
Causes of death for deceased subjects were obtained from the register of the Croatian National Institute of Public Health, from 2003 to 2021, thus yielding a total of 18 years of the follow-up period. This source was used to extract the date, and ICD-10 encoded cause of death, assigning subjects to cardiovascular or cancer mortality groups.
The initial comparison was made with the chi-square test and t-test, while Cox’s proportional hazards regression model with 95% confidence intervals was utilized to calculate hazard ratios. Models were adjusted for age, sex, body mass index (BMI), physical activity, subjective material status, household material status and education (years of schooling). The significance threshold was set at P<0.05. All analyses were performed with SPSS ver 18 (SPSS Inc, Chicago, IL).
Results
A total of 1,025 subjects were included (426 men and 599 women). The follow-up period lasted for 18 years, amounting to a total of 5,775,140 days. During this period, there were 279 deaths (27.2%), with 122 (43.7%) caused by cardiovascular disease, 103 (36.9%) by cancer and 54 by other causes (19.4%).
The bivariate analysis showed that the deceased subjects were significantly older, more commonly men, with lower education and material household status, with elevated body mass index and less physically active (Table 1). Interestingly, psychological distress and subjective material status did not show a formally significant difference between these two groups (Table 1). Instead, psychological distress was more common in the elderly, women, and subjects with lower socioeconomic status (Table 2).
The multivariate analysis of the all-cause mortality only had a marginal effect on the material household status (Table 3). In addition, neither cardiovascular mortality nor cancer-related mortality showed any significance concerning the socioeconomic status or psychological distress (Table 3). Finally, we created an interaction model, where three socioeconomic variables and psychological distress were entered into a Cox regression model as the interaction term, alongside age and gender as confounders. The model yielded no significant results, with the hazard ratio for the interaction item for all-cause mortality of 1.00 [95% confidence intervals 0.99-1.01]. Similarly, neither the cardiovascular nor the cancer mortality showed significance when a similar interaction item was entered in the model.
Discussion
This study shows an exceptional level of social homogeneity in the isolated population of the island of Vis in Croatia, reflected in the absence of effect of social inequalities or psychological distress on mortality. This absent correlation disagrees with numerous previously published studies.1,9–14,24 However, some existing studies reported an association between absent mortality and anxiety and depression,25 or psychological distress.26 Although a link between psychological distress and cardiovascular mortality is well established,12,27–29 some studies suggested that psychological symptoms were only associated with cardiovascular events, but not with cardiovascular mortality.17,30 Interestingly, we found comparably lower cardiovascular mortality and greater cancer-related mortality than national data,31–34 despite a rather prevalent extent of psychological distress.
The results of this study showed significant differences between psychologically distressed and non-distressed regarding socioeconomic factors, confirming some previous studies.11,17 It has also been hypothesized that socioeconomic inequalities may partly be caused by psychological distress that stems from social deprivation.35 Our study was not able to reproduce these results as the interaction model of socioeconomic status showed no correlation with psychological distress. All these findings support previous ideas that the interplay of socioeconomic status, psychological distress, and mortality are complex.36,37
The limitations of this study include a relatively small sample size or a possibility that the results of the study might be specific for the investigated population. Another problem may arise from the mortality data, which may be biased, especially in small communities. Nevertheless, these results support the findings of previously reported high levels of social homogeneity in small and isolated populations.
Acknowledgements
We would like to thank to all subjects from the island of Vis who participated in this study. In addition, we would like to thank to numerous institutions and individuals who assisted in this study, namely the Institute for Anthropological Research, the Institute for Public Health and Medical School of the University of Zagreb and the University of Split.
Competing interests
OP is the Editor-in-Chief of the Journal of Global Health Economics and Policy. To ensure that any possible conflict of interest relevant to the journal has been addressed, this article was reviewed according to best practice guidelines of international editorial organizations. All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author).
Funding
This study was funded by the Medical Research Council UK, The Croatian Ministry of Science, Education and Sports (grant 216-1080315-0302), the European Union framework program 6 EUROSPAN project (contract no. LSHG-CT-2006-018947), the Croatian National Center of Research Excellence in Personalized Healthcare (grant number KK.01.1.1.01.0010) and the Center of Competence in Molecular Diagnostics (KK.01.2.2.03.0006).
Author contributions
OP and IK conceived the study. ML and AC performed the analysis and drafted the manuscript. All authors approved the final version and are held responsible for the entire article content.
Raw data
The entire raw dataset is available as the Supplementary material.
Ethics
This study was approved by the Ethical boards of the Medical School, University of Zagreb, the Multi-Centre Research Ethics Committee for Scotland, The University of Split School of Medicine and the Lothian NHS Board (approvals available upon request to the corresponding author). The use of mortality data was approved be the Ethical Board of the Croatian National Institute of Public Health.
Correspondence
Ozren Polašek, Department of Public Health, University of Split School of Medicine, Split, Croatia; op@mefst.hr