Background

Depression is associated with an increased dementia risk, but the nature of the association in the long-term remains unresolved, and the role of sociodemographic factors mainly unexplored.

Aims

To assess whether a history of clinical depression is associated with dementia in later life, controlling for observed sociodemographic factors and unobserved factors shared by siblings, and to test whether gender, educational level and marital status modify the association.

Method

We conducted a national cohort study of 1 616 321 individuals aged 65 years or older between 2001 and 2018 using administrative healthcare data. A history of depression was ascertained from the national hospital register in the period 15–30 years prior to dementia follow-up. We used conventional and sibling fixed-effects Cox regression models to analyse the association between a history of depression, sociodemographic factors and dementia.

Results

A history of depression was related to an adjusted hazard ratio of 1.27 (95% CI 1.23–1.31) for dementia in the conventional Cox model and of 1.55 (95% CI 1.09–2.20) in the sibling fixed-effects model. Depression was related to an elevated dementia risk similarly across all levels of education (test for interaction, P = 0.84), but the association was weaker for the widowed than for the married (P = 0.003), and stronger for men than women (P = 0.006). The excess risk among men attenuated following covariate adjustment (P = 0.10).

Discussion

This study shows that a history of depression is consistently associated with later-life dementia risk. The results support the hypothesis that depression is an aetiological risk factor for dementia.

Background Research evidence on the association between neighbourhood characteristics and individual mental health at older ages is inconsistent, possibly due to heterogeneity in the measurement of mental-health outcomes, neighbourhood characteristics and confounders. Register-based data enabled us to avoid these problems in this longitudinal study on the associations between socioeconomic and physical neighbourhood characteristics and individual antidepressant use in three national contexts.

Methods We used register-based longitudinal data on the population aged 50+ from Turin (Italy), Stockholm (Sweden), and the nine largest cities in Finland linked to satellite-based land-cover data. This included individual-level information on sociodemographic factors and antidepressant use, and on neighbourhood socioeconomic characteristics, levels of urbanicity, green space and land-use mix (LUM). We assessed individual-level antidepressant use over 6 years in 2001-2017 using mixed-effects logistic regression.

Results A higher neighbourhood proportion of low-educated individuals predicted lower odds for antidepressant use in Turin and Stockholm when individual-level sociodemographic factors were controlled for. Urbanicity predicted increased antidepressant use in Stockholm (OR=1.02; 95% CI 1.01 to 1.03) together with more LUM (OR=1.03; 1.01-1.05) and population density (OR=1.08; 1.05-1.10). The two latter characteristics also predicted increased antidepressant use in the Finnish cities (OR=1.05; 1.02-1.08 and OR=1.14; 1.02-1.28, respectively). After accounting for all studied neighbourhood and individual characteristics of the residents, the neighbourhoods still varied by odds of antidepressant use.

Conclusions Overall, the associations of neighbourhood socioeconomic and physical characteristics with older people's antidepressant use were small and inconsistent. However, we found modest evidence that dense physical urban environments predicted higher antidepressant use among older people in Stockholm and the Finnish cities.

Background: The reasons for the shorter life expectancy of people with depression may vary by age. We quantified the contributions of specific causes of death by age to the life-expectancy gap in four European countries.

Methods: Using register-based cohort data, we calculated annual mortality rates in between 1993 and 2007 for psychiatric inpatients with depression identified from hospital-care registers in Denmark, Finland and Sweden, and between 2000 and 2007 for antidepressant-treated outpatients identified from medication registers in Finland and Turin, Italy. We decomposed the life-expectancy gap at age 15 years by age and cause of death.

Results: The life-expectancy gap was especially large for psychiatric inpatients (12.1 to 21.0 years) but substantial also for antidepressant-treated outpatients (6.3 to 14.2 years). Among psychiatric inpatients, the gap was largely attributable to unnatural deaths below age 55 years. The overall contribution was largest for suicide in Sweden (43 to 45%) and Finland (37 to 40%). In Denmark, 'other diseases' (25 to 34%) and alcohol-attributable causes (10 to 18%) had especially large contributions. Among antidepressant-treated outpatients, largest contributions were observed for suicide (18% for men) and circulatory deaths (23% for women) in Finland, and cancer deaths in Turin (29 to 36%). Natural deaths were concentrated at ages above 65 years.

Limitations: The indication of antidepressant prescription could not be ascertained from the medication registers.

Conclusions: Interventions should be directed to self-harm and substance use problems among younger psychi-atric inpatients and antidepressant-treated young men. Rigorous monitoring and treatment of comorbid somatic conditions and disease risk factors may increase life expectancy for antidepressant-treated outpatients, especially women.

Background: Most first-generation migrants have lower mortality compared to the native population. Finnish-born migrants in Sweden instead have higher mortality; possibly because of health behaviours established before migration. To increase our understanding of this excess mortality, we compared the cause-specific mortality of Finnish migrants in Sweden to both the native population of Sweden and the native Finnish population residing in Finland.

Methods: We used Swedish and Finnish register data, applying propensity score matching techniques to account for differences in sociodemographic characteristics between the migrants, Swedes and Finns. The index population were Finnish migrants aged 40–60, residing in Sweden in 1995. We compared patterns of all-cause, alcohol- and smoking-related, and cardiovascular disease mortality across the groups in the period 1996–2007.

Results: Finnish migrant men in Sweden had lower all-cause mortality compared to Finnish men but higher mortality compared to the Swedish men. The same patterns were observed for alcohol-related, smoking-related and cardiovascular disease mortality. Among women, all three groups had similar levels of all-cause mortality. However, Finnish migrant women had higher alcohol-related mortality than Swedish women, similar to Finnish women. Conversely, migrant women had similar levels of smoking-related mortality to Swedish women, lower than Finnish women.

Conclusions: Finnish-born migrants residing in Sweden have mortality patterns that are typically in between the mortality patterns of the native populations in their country of origin and destination. Both the country of origin and destination need to be considered in order to better understand migrant health.

Background Genetic vulnerability to coronary heart disease (CHD) is well established, but little is known whether these effects are mediated or modified by equally well-established social determinants of CHD. We estimate the joint associations of the polygenetic risk score (PRS) for CHD and education on CHD events.

Methods The data are from the 1992, 1997, 2002, 2007 and 2012 surveys of the population-based FINRISK Study including measures of social, behavioural and metabolic factors and genome-wide genotypes (N=26 203). Follow-up of fatal and non-fatal incident CHD events (N=2063) was based on nationwide registers.

Results Allowing for age, sex, study year, region of residence, study batch and principal components, those in the highest quartile of PRS for CHD had strongly increased risk of CHD events compared with the lowest quartile (HR=2.26; 95% CI: 1.97 to 2.59); associations were also observed for low education (HR=1.58; 95% CI: 1.32 to 1.89). These effects were largely independent of each other. Adjustment for baseline smoking, alcohol use, body mass index, igh-density lipoprotein (HDL) and total cholesterol, blood pressure and diabetes attenuated the PRS associations by 10% and the education associations by 50%. We do not find strong evidence of interactions between PRS and education.

Conclusions PRS and education predict CHD events, and these associations are independent of each other. Both can improve CHD prediction beyond behavioural risks. The results imply that observational studies that do not have information on genetic risk factors for CHD do not provide confounded estimates for the association between education and CHD.

Background: Previous studies have shown that the risk of death is elevated after the death of a spouse. Limited evidence is available on changes in mental health before and after bereavement among individuals whose spouse dies of dementia. Methods: We analyzed changes in the 3-month prevalence of antidepressant use for 5 years before and 3 years after widowhood for individuals whose spouses died of either dementia or other causes. The study used data of 41 855 widowed individuals and repeated-measures logistic regression analyses. Antidepressant use was based on the prescription register of Finland in 1995-2007. Results: Five years before widowhood, the 3-month prevalence for antidepressant use was 4% among widowing men and 6-7% among widowing women, regardless of whether the spouse died of dementia or other causes. Further changes in antidepressant use depended on a spouse's cause of death. Women whose spouses died of dementia experienced large increase in antidepressant use starting from 3 to 4 years prior to widowhood, whereas other widows did not experience large increase until after widowhood. The trajectories for men were similar. Antidepressant use following the death of a spouse with dementia stayed at a new heightened level after widowhood. Conclusions: The trajectories of antidepressant use indicate that the process of losing a spouse to dementia is bad for mental health, already a few years prior to widowhood. There are no clear improvements in mental health after the death of a spouse with dementia. Support services for individuals whose spouses' dementia progresses are needed.

Objectives To assess the association between multiple indicators of socioeconomic position and dementia-related death, and to estimate the contribution of dementia to socioeconomic differences in overall mortality at older ages.

Design Prospective population-based register study.

Setting Finland.

Participants 11% random sample of the population aged 70-87 years resident in Finland at the end of year 2000 (n=54 964).

Main outcome measure Incidence rates, Kaplan-Meier survival probabilities and Cox regression HRs of dementia mortality in 2001-2016 by midlife education, occupational social class and household income measured at ages 53-57 years.

Results During the 528 387 person-years at risk, 11 395 individuals died from dementia (215.7 per 10 000 person-years). Lower midlife education, occupational social class and household income were associated with higher dementia mortality, and the differences persisted to the oldest old ages. Compared with mortality from all other causes, however, the socioeconomic differences emerged later. Dementia accounted for 28% of the difference between low and high education groups in overall mortality at age 70+ years, and for 21% of the difference between lowest and highest household income quintiles. All indicators of socioeconomic position were independently associated with dementia mortality, low household income being the strongest independent predictor (HR=1.24, 95% CI 1.16 to 1.32), followed by basic education (HR=1.14, 1.06 to 1.23). Manual occupational social class was related to a 6% higher hazard (HR=1.06, 1.01 to 1.11) compared with non-manual social class. Adjustment for midlife economic activity, baseline marital status and chronic health conditions attenuated the excess hazard of low midlife household income, although significant effects remained.

Conclusion Several indicators of socioeconomic position predict dementia mortality independently and socioeconomic inequalities persist into the oldest old ages. The results demonstrate that dementia is among the most important contributors to socioeconomic inequalities in overall mortality at older ages.

Background Previous studies have found depression to be negatively associated with the prognosis of both cardiovascular disease (CVD) and cancer, but this may partly reflect reverse causality. We limited the possibility of reverse causality by measuring depression before the first diagnosis of CVD or cancer.

Methods We used an 11% longitudinal random sample of the Finnish population aged 25 years or older who are residents of Finland for at least 1year between 1987 and 2007, with an 80% oversample of those who died during this period. Those who had their first incidence of coronary heart disease (CHD) (n=107966), stroke (n=68685) or cancer (n=113754) between 1998 and 2012 were followed up for cause-specific mortality from the date of diagnosis until the end of 2012. Depression was defined as having antidepressant purchases two to three calendar years before the incidence. Logistic and Cox regression models were used to examine short-term and long-term mortality by depression status.

Results Long-term mortality after diagnosis was 1.34 (95% CI 1.25 to 1.44) for CHD, 1.26 (95% CI 1.15 to 1.37) for stroke and 1.10 (95% CI 1.04 to 1.16) for cancer in those who had used antidepressants in two consecutive calendar years as compared with those with no purchases. Short-term mortality from CHD was elevated among persons with depression (OR=1.30; 95%CI 1.06 to 1.61), but no association was found for stroke.

Conclusion Pre-existing depression is associated with a worse prognosis of CHD, stroke and cancer. More attention in the healthcare system is needed for patients with chronic diseases who have a history of depression.

The effects of substance abuse on other family members are not fully established. We estimate the contribution of parental substance abuse on offspring psychiatric morbidity in late adolescence and early adulthood, with emphasis on the timing and persistency of exposure. We used a nationally representative 20% sample of Finnish families with children born in 1986-1996 (n = 136,604) followed up in 1986-2011. We identified parental substance abuse and offspring psychiatric morbidity from hospital discharge records, death records and medication registers. The effects of parental substance abuse at ages 0-4, 5-9 and 10-14 on psychiatric morbidity after age 15 were estimated using population averaged and sibling fixed effects models; the latter controlling for unobserved factors shared by siblings. Parental substance abuse at ages 0-14 was associated with almost 2-fold increase in offspring psychiatric morbidity (HR = 1.86, 95% CI 1.78-1.95). Adjustment for childhood parental education, income, social class and family type reduced these effects by about 50%, with some further attenuation after adjustment for time-varying offspring characteristics. In the sibling fixed effects models those exposed at 0-4 or 5-9 years had 20% (HR = 1.20, 95% CI 0.90-1.60) and 33% (HR = 1.33, 95% CI 1.01-1.74) excess morbidity respectively. Also in sibling models those with early exposure at ages 0-4 combined with repeated exposure in later childhood had about 80-90% higher psychiatric morbidity as compared to never exposed siblings (e.g. for those exposed throughout childhood HR = 1.81, 95% CI 1.01-3.25). Childhood exposure to parental substance abuse is strongly associated with subsequent psychiatric morbidity. Although these effects are to a large extent due to other characteristics shared within the parental home, repeated exposure to parental substance abuse is independently associated with later psychiatric morbidity.

Objectives The effects of socio-demographic and economic factors on institutional long-term care (LTC) among people with dementia remain unclear. Inconsistent findings may relate to time varying effects of these factors as dementia progresses. To clarify the question, we estimated institutional LTC trajectories by age, marital status and household income in the eight years preceding dementia-related and non-dementia-related deaths. Methods We assessed a population-representative sample of Finnish men and women for institutional LTC over an eight-year period before death. Deaths related to dementia and all other causes at the age of 70+ in 2001-2007 were identified from the Death Register. Dates in institutional LTC were obtained from national care registers. We calculated the average and time-varying marginal effects of age, marital status and household income on the estimated probability of institutional LTC use, employing repeated-measures logistic regression models with generalised estimating equations (GEE). Results The effects of age, marital status and household income on institutional LTC varied across the time before death, and the patterns differed between dementia-related and non-dementia-related deaths. Among people who died of dementia, being of older age, non-married and having a lower income predicted a higher probability of institutional LTC only until three to four years before death, after which the differences diminished or disappeared. Among women in particular, the probability of institutional LTC was nearly equal across age, marital status and income groups in the last year before dementia-related death. Among those who died from non-dementia-related causes, in contrast, the differences widened until death. Conclusions We show that individuals with dementia require intensive professional care at the end of life, regardless of their socio-demographic or economic resources. The results imply that the potential for extending community living for people with dementia is likely to be difficult through modification of their socio-demographic and economic environments.