Settings and participants

This study was based on home visits performed in the Umeå 85+/Gerontological Regional Database (GERDA) study, a population-based, cross-sectional and longitudinal study. The GERDA study aims to increase knowledge of very old adults’ health and living conditions and provide data for planning and support of their care (www.gerdacentre.com). Participants were selected on the basis of age and place of residence. In the regions listed below, every other 85-year-old, every 90-year-old, and all those aged 95 years and older were asked to participate. The choice to select every other 85-year-old was made so that the groups would be equal in size. The names, addresses and civil registration numbers of all potential participants living in the areas were collected from the National Tax Board in Sweden and the Population Register Centre in Finland.

The study started in 2000 in Umeå, with rural communities added in 2002. Data collections were repeated every 5 years (2005–2007, 2010–2012 and 2015–2017). At each data collection, new participants were added, and previous participants who were still alive were also invited. Surviving participants from previous cohorts were included as unique cases in the subsequent cohorts. All collections used identical procedures and inclusion criteria. No exclusion criteria were used. Home visits were conducted in the city of Umeå in 2000, 2005, 2010 and 2015, as well as in five rural municipalities (Malå, Vilhelmina, Storuman, Sorsele and Dorotea) in 2002, 2007, 2012 and 2017. In Österbotten, Finland, the home visits were performed twice (2006 and 2011–2012). The city of Vaasa and the rural Korsholm municipality participated in both data collections, whereas the second data collection also included home visits in rural municipalities Malax and Korsnäs.

Eligible participants were approached by postal mail and provided with information about the study. Within 2 weeks, they were asked to provide their consent for participation by telephone, and a home visit was scheduled. During the home visits, consent was again confirmed. If the participants were cognitively impaired or could not consent, next of kin or caregivers were consulted. In addition, all participants were asked for consent for access to their medical journals. The participants could withdraw from the study without explanation or consequence.

Assessments

Trained professionals (nurses, medical students, physiotherapists and physicians) conducted structured interviews and performed physical assessments during the home visits. The interviews contained questions about sociodemographic factors such as living arrangements, marital status and years of education. Height and weight were measured with a calibrated scale and measuring stick, and body mass index was calculated. Impaired vision was defined as being unable to read 5 mm text from 30 cm with or without glasses. Impaired hearing was defined as being unable to hear a normal conversation from 1 m away with or without a hearing aid. The Mini-Mental State Examination (MMSE) was used to assess cognitive status. The MMSE consists of questions and tasks that assess various aspects of cognitive function, with scores in the range of 0–30; higher scores indicate better cognitive function, and a score below 24 may indicate cognitive impairment. ^{Reference Folstein, Folstein and McHugh18} Dependency in activities of daily living was assessed using Barthel’s index. Scores on this index range from 0 to 20, with higher scores indicating greater independence. ^{Reference Collin, Wade, Davies and Horne19}

Assessments conducted during the home visits to evaluate potential depressive disorders included the Geriatric Depression Scale (GDS-15), the Philadelphia Geriatric Center Morale Scale (PGCMS), the Life Orientation Test (LOT), the Montgomery–Åsberg Depression Rating Scale (MADRS), and the Organic Brain Syndrome Scale (OBS). The interview protocol also included a question on whether the participants felt they were happy or unhappy, which was scored on a five-point Likert scale.

The GDS-15 consists of 15 questions used to assess various aspects of depressive symptoms. ^{Reference Sheikh and Yesavage20} The questions are answered yes or no and scored either 1 or 0, and the total score is summed. Scores of 5 or above indicate degrees of depressive symptoms, with higher scores indicating a greater severity of depressive symptoms. ^{Reference Sheikh and Yesavage20} The GDS-15 has been shown to be reliable for detecting depressive disorders in people over 85 years old with high sensitivity and specificity, as well as an internal reliability of α = 0.73. ^{Reference Almeida and Almeida21,Reference Snellman, Hörnsten, Olofsson, Gustafson, Lövheim and Niklasson22} It has also been shown to be reliable for use among very old adults with MMSE scores of 10 or more. ^{Reference Conradsson, Rosendahl, Littbrand, Gustafson, Olofsson and Lövheim23}

The PGCMS is designed to assess an individual’s morale, including emotional well-being, satisfaction in life and overall well-being. ^{Reference Lawton24} The PGCMS consists of 17 items, answered yes or no. ^{Reference Lawton24} The scale has been used in research to assess morale in older adults and as a tool for identifying individuals at risk of developing depression. ^{Reference Lawton24,Reference Niklasson, Näsman, Nyqvist, Conradsson, Olofsson and Lövheim25}

The LOT is a tool designed to measure a person’s overall life satisfaction, using questions concerning feelings about the future, loneliness and will to live, including the question ‘are you depressed?’. ^{Reference Fagerström26} The Revised LOT (LOT-R) is considered to be reliable and easily understood by older adults. ^{Reference Fagerström26}

The MADRS is a clinical assessment tool designed to measure the severity of depression, on the basis of the respondent’s self-report and clinical observations. ^{Reference Snaith, Harrop, Newby and Teale27} It contains questions about the core symptoms of depression. Each item is scored from 0 to 6, with higher scores indicating more severe depressive symptoms. The MADRS is most commonly used for evaluating treatment response and effectiveness of interventions for depression. ^{Reference Müller28} Here, it was used only when the interviewer was a physician or medical student trained to use the scale.

The OBS is an observational tool used to assess cognitive function and detect organic brain dysfunction. ^{Reference Jensen, Dehlin and Gustafson29} The OBS evaluates the individual’s clinical state in the past month, including emotional and behavioural symptoms. In addition, one item concerns the degree of depressed mood based on the overall impression during the full interview, rated from 0 to 3. ^{Reference Björkelund, Larsson, Gustafson and Andersson30}

Information on medical conditions, prescribed medications and diagnoses was collected from the participants’ medical journals and during the home visits. Prescribed medications were reviewed with the participants during the home visit, and only medications that the participants confirmed that they used were registered as used medications in this study. The participants’ medical records were reviewed after consent had been obtained, and information was confirmed in person by either the participant or a relative and/or caregiver. Relatives also contributed additional relevant information if applicable. For this study, response to antidepressant treatment was defined as having been prescribed antidepressants and scoring less than 5 on the GDS-15, whereas non-response was defined as having been prescribed antidepressants and having a GDS score equal to or greater than 5.

An experienced geriatrician (Y.G.) reviewed all medical records and diagnoses and set the diagnoses of depressive and dementia disorders in all cohorts in both Sweden and Finland according to the DSM-IV-TR. All available documentation, medical records and applicable assessments (GDS-15, PGCMS, LOS, MADRS, OBS and Barthel) were used. These DSM-IV-TR-based diagnoses of depressive disorders and dementia were consistently used in this study when referring to depressive disorders or dementia. The diagnostic procedure has been described in detail elsewhere. ^{Reference Corneliusson, Gustafson and Olofsson16} Mortality records were collected from the death certificates of deceased participants, population registers or medical records. The sample for this study consisted of 2551 participants, of whom 918 participants had a diagnosis of a depressive disorder according to the DSM-IV-TR. ³¹ The STROBE guidelines were adhered to throughout this study. ^{Reference von Elm, Altman, Egger, Pocock, Gotzsche and Vandenbroucke32}

Statistical analysis

Descriptive statistics, including frequencies and means, were used to analyse the characteristics of the entire sample. Chi-squared test, Fisher’s exact test and t-test were used to assess differences between groups and means. We used logistic regression models to identify factors associated with depressive disorders and mortality in men and women, and Cox proportional hazards regressions to explore factors associated with time to death. Kaplan–Meier survival analyses were used to determine differences in survival between groups and to compare survival distributions. Statistical significance was defined as P < 0.05. IBM SPSS statistics version 29.0, for Windows (IBM Corporation, Armonk, NY, USA; https://www.ibm.com/products/spss-statistics) was used for the statistical analyses.

Ethical considerations

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2013. All procedures involving human participants and/or patients were approved by the Regional Ethical Review Board of Umeå/Swedish Ethical Review Authority, Sweden (Dnr 99-326, 05-063M, 09-178M, 2015/296-31, 2021-00023) and the Ethics Committee of Vaasa Central Hospital (reg. nos. 05-87 and 10-54). All data have been stored in compliance with data regulation laws.