Retirement age and type as predictors of frailty : a retrospective cohort study of older businessmen

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC 4.0 https://creativecommons.org/licenses/by-nc/4.0/ Retirement age and type as predictors of frailty : a retrospective cohort study of older businessmen © Authors, 2020 Published version Haapanen, Markus J.; von Bonsdorff, Mikaela B.; Perttilä, Niko M.; Törmäkangas, Timo; von Bonsdorff, Monika E.; Strandberg, Arto Y.; Strandberg, Timo E. Haapanen, M. J., von Bonsdorff, M. B., Perttilä, N. M., Törmäkangas, T., von Bonsdorff, M. E., Strandberg, A. Y., & Strandberg, T. E. (2020). Retirement age and type as predictors of frailty : a retrospective cohort study of older businessmen. BMJ Open, 10(12), Article e037722. https://doi.org/10.1136/bmjopen-2020-037722 2020 1 HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access Retirement age and type as predictors of frailty: a retrospective cohort study of older businessmen Markus J Haapanen ,1,2 Mikaela B von Bonsdorff,2,3 Niko M Perttilä,1 Timo Törmäkangas,3 Monika E von Bonsdorff,4,5 Arto Y Strandberg ,6 Timo E Strandberg6,7 To cite: HaapanenMJ, von BonsdorffMB, PerttiläNM, etal. Retirement age and type as predictors of frailty: a retrospective cohort study of older businessmen. BMJ Open 2020;10:e037722. doi:10.1136/ bmjopen-2020-037722 ►Prepublication history and supplemental material for this paper is available online. To view these files, please visit the journal online (http:// dx. doi. org/ 10. 1136/ bmjopen- 2020- 037722). Received 13 February 2020 Revised 16 November 2020 Accepted 27 November 2020 For numbered affiliations see end of article. Correspondence to Dr Markus J Haapanen; markus. haapanen@ helsinki. fi Original research © Author(s) (or their employer(s)) 2020. Reuse permitted under CC BYNC. No commercial reuse. See rights and permissions. Published by BMJ. ABSTRACT Objectives To study the association between retirement characteristics and frailty in a homogenous population of former business executives. Design Crosssectional cohort study using data from the Helsinki Businessmen Study. Setting Helsinki, Finland. Participants 1324 Caucasian men, born in 1919–1934, who had worked as business executives and managers and of whom 95.9% had retired by the year 2000. Questions on age at and type of retirement, lifestyle and chronic conditions were embedded in questionnaires. Primary and secondary outcome measures Frailty assessed according to a modified phenotype definition at mean age 73.3 years. Results Mean age at retirement was 61.3 years (SD 4.3) and 37.1% had retired due to old age. The prevalence of frailty was lowest among men retiring at ages 66–67 years but increased among those who worked up to age 70 years or older. Compared with men who retired before age 55 years, those retiring at ages 58–69 years were at decreased risk of frailty in old age relative to nonfrailty (adjusted ORs 0.07–0.29, p<0.05). Compared with men who transitioned into old age retirement, those who retired due to disability were at increased risk of prefrailty (adjusted OR 1.53, 95% CI 1.01 to 2.32) and frailty (adjusted OR 3.52, 95% CI 1.97 to 6.29), relative to nonfrailty. Conclusion Exiting working life early and continuing to be occupationally active until age 70 years and older were both associated with increased risk of frailty among the men. Promotion of longer work careers could, however, promote healthier ageing, as the lowest prevalence of frailty was observed in former business executives who retired at ages 66–67 years. INTRODUCTION The geriatric syndrome frailty, which refers to a loss of biological reserves and increased vulnerability to stressors, is associated with adverse health outcomes including falls, hospitalisation and premature mortality.1 2 Frailty is a highly significant public health concern in populations aged 65 years and older, who also comprise the globally fastest growing demographic group.3–5 In this population, retirement from paid work represents an important life transition, with potential changes to health behaviour including reduced cigarette smoking and increased physical activity in the short term.6 With an emphasis on promoting longer work careers in current retirement planning,7 and given the importance of preventive measures in reducing the prevalence of frailty in ageing populations,8 it becomes increasingly important to understand how occupational activity is associated with frailty in old age. Previous evidence suggests a twodirectional association between retirement and frailty. On one hand, being frail during the work career has been associated with an increased risk of premature exit from the workforce, comparing against nonfrail individuals.9 Accordingly, in another study, those who exited the workforce prematurely had a higher mean frailty index score than those who continued working full time until age 65 years.10 Considering the recent trend of increasing statutory retirement age in retirement planning, however, little is known about whether being occupationally active in older age, beyond age 65 years and older, would Strengths and limitations of this study ►This study is among the first to report associations between the age at retirement, type of pension and frailty in old age among a homogenous population characterised with high socioeconomic position. ►Generalisation of the results to other socioeconomic or occupational groups should be made with caution. The study included 1324 men who had worked as business executives and managers during their work careers. ►Frailty was defined according to a modified phenotype definition, assessed using questionnaire data, and information on retirement characteristics was selfreported in this crosssectional study. Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 2HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access also associate with the prevalence of frailty. To study these associations is of importance, given the employment rate of Finns aged 65 years and older has almost doubled from 7.2% in 2001 to 13.8% in the year 2016.11 Being occupationally active in older age is particularly common among individuals of higher occupational classes.12 The Helsinki Businessmen Study (HBS)13 comprises a homogenous population of business executives and managers, who provided information on their age at and type of retirement as well as an assessment of the stressfulness of their work career, and who were assessed for frailty in old age. The aim of the present study was to find whether occupational activity in older age would also associate with frailty. We hypothesised that business executives who continued to be occupationally active would also have less of frailty in old age. METHODS Study population The HBS has been described in detail previously.13–15 In brief, 3490 men born between 1919 and 1934, who mostly were business executives and managers during their work careers, participated in voluntary health checkups in midlife at the Finnish Institute of Occupational Health. Altogether, 3310 men could be traced using a personal identification number assigned to all Finnish citizens in the 1970s and were thus included in the study. In the year 2000, all 2287 surviving participants (1023 participants (29.3%) had died between 1964 and 2000) of the original cohort received a mailed questionnaire (including the Rand 36- Item Health Survey (RAND-36)/36- Item Short Form Survey healthrelated quality of life instrument, questions about lifestyle, current weight and morbidity), and of these, 1864 (81.5%) responded. After screening, 1769 participants had full information on age at and type of retirement. Frailty could be assessed among 1366 men using a modified phenotype definition16 at mean age 73.3 years (SD 4.1 years). Information on age at and type of retirement was selfreported and was obtained using questionnaires. The participants had been retired for a mean duration of 12.0 years (SD 5.2 years) prior to frailty assessment. The analytical sample of the present study consisted of 1324 men who had complete data on both frailty and age at and type of retirement. Age at and type of retirement The participants were asked to provide information on their working status in the mailed questionnaire. In the year 2000, practically all (95.9%; n=1787) of the men reported that they had retired, and they provided the exact age at which they had retired together with information on the type of retirement. Participants who had not reported that they had retired (n=95) were excluded prior and after screening, 1769 participants had full information on both age at and type of retirement. Information on occupation or the age which was considered ‘statutory retirement age’ in these sectors/occupations was not known. Due to the nonlinear association between age at retirement and frailty (figure 1), as well as given the fact that no universal retirement age could be defined in this group consisting of managers and executives, we categorised retirement age into nine groups: ≤55 years (n=95, 7.3%); 56–57 years (n=87, 6.7%); 58–59 years (n=129, 10.0%); 60–61 years (n=324, 25.0%); 62–63 years (n=295, 22.8%); 64–65 years (n=275, 21.2%); 66–67 years (n=52, 4.0%); 68–69 years (n=15, 1.2%); ≥70 years (n=23, 1.8%). The types of retirement identified were as follows: old age retirement (n=507; 38.3%), early old age retirement (n=269; 20.3%), pension benefit packages (ie, early retirement incentives or redundancy packages) (n=251; 19.0%), parttime retirement (n=19; 1.4%), unemployment retirement (n=103; 7.8%) and disability retirement (n=175; 13.2%). The category ‘parttime retirement’ was merged with the category ‘early old age retirement’ due to few observations in the former and as the groups were otherwise similar regarding background information. In Finland, the lower age limit for old age retirement was 65 and 63 years, respectively, for private and public sector workers. However, employers had individual old age retirement schemes that varied between ages 60 and 65 years. Early old age pensions could be granted if the applicant was older than 60 years and working in the private sector, or over 58 years if they worked in the public sector. Similar age criteria were applied for parttime pensions for workers transitioning from fulltime to parttime work. Unemployment retirement could be granted to individuals aged 55–60 years who had received the maximum of 500 days of daily unemployment allowance. Disability pensions can be granted to individuals aged 16 years and older who have a significant disability that reduces work capacity continually for over 300 workdays. Relaxed disability pensions require the minimum age of 55 years, a long work career and a chronic disease that causes disability.17 Frailty The modified phenotype frailty definition used in HBS has been described previously and shown to have Figure 1 Distribution of frailty classification assessed at an average age of 73.3 years according to categories of age at retirement. Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 3 HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access predictive validity.16 In brief, as walking speed was not measured in 2000, the following four criteria were implemented: (1) shrinking defined as weight loss of ≥5% from midlife, or having a current body mass index of <21 kg/ m2; (2) weakness defined as selfreported difficulty (not at all=0) in carrying or lifting a grocery bag in the physical functioning domain of the RAND-36 questionnaire18; (3) exhaustion defined as selfreported low energy most of the time during the previous 4 weeks in the vitality domain of the RAND-36 questionnaire18; (4) low physical activity defined as answering ‘No’ to the question ‘Do you exercise regularly weekly?’. The participants were classified as frail (3 or 4), prefrail (1 or 2) and nonfrail (none), according to the number of criteria met. We excluded participants with any missing questionnaire information on any of the subdomains of frailty (n=498). Covariates The postal questionnaire provided selfreported information on health behaviours and chronic physiciandiagnosed conditions in the year 2000. Smoking was coded into ‘never smokers’, ‘exsmokers’ and ‘current smokers’. Alcohol consumption was reported as the weekly number of units of alcoholic beverages and categorised into ‘zero consumption’, ‘light consumption’ (1–98 g/week, reported as maximum 7 drinks/week), ‘moderate consumption’ (99–196 g/week, maximum 14 drinks/week) and ‘high consumption’ (>196 g/week, more than 14 drinks/week) as in earlier publications.19 Participants provided information on the presence or history of the following conditions: hypertension, coronary heart disease, claudication, heart failure, diabetes, chronic pulmonary disease, memory disturbances, stroke, cancer, musculoskeletal disorders, psychiatric conditions, or any other longterm conditions, including trauma. In the year 2003, the participants were asked to rate the stressfulness of their entire work career using a visual analogue scale, translated into 0 (very stressful) to 10 (not stressful at all). This information was available for 72% (n=957) of the analytical sample in the present study. Statistical methods The data are presented as means and SDs for continuous variables, and as percentages for categorical variables. They were tested for differences using analyses of variance for normally distributed and Kruskal- Wallis test for nonnormally distributed continuous variables and crosstabulation for categorical variables. Group differences were tested across and between all frailty groups. Multinomial regression analysis was used to estimate ORs and 95% CIs of associations between age at and type of retirement and frailty in old age. Two adjustment schemes were used: model 1 was adjusted for age; model 2 was additionally adjusted for smoking, alcohol consumption, cardiovascular disease (CVD) (0/1) and diabetes (0/1). Likelihood ratio test was used to test for difference in the point estimates of frailty according to groups of retirement age. We also performed our main analyses dividing the participants into three equal groups by birth year (cutoffs in 1924 and 1929) and into two groups between historical events including the Great Depression (1929–1939), to differentiate between possible birth cohort and historical cohort effects. The analysis between age at retirement and frailty was also run separately for each retirement type. The analyses were two tailed, and significance was set at 0.05. The analyses were carried out using statistical software SPSS (IBM SPSS Statistics V.25.0 released 2017; IBM) and Mplus V.7.0 for the likelihood ratio test. Patient and public involvement This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patientrelevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy. RESULTS The prevalence of frailty was 10.7% at the mean age of 73.3 years (SD 4.1 years; range 66–81 years). Across the three frailty groups, frail men were on average older (p for trend between frailty groups <0.001), more frequently current smokers (p<0.001) and ‘high’ consumers of alcohol (p=0.014), and had more chronic conditions, including diabetes (p=0.019, online supplemental table 1). Mean age at retirement was 61.3 years (SD 4.3; range 27–76 years) and 98.5% retired at age 50 years or older. More than a third of the participants (37.1%) had retired due to old age, and 1 in 8 (12.8%) retired because of disability (table 1). Table 1 shows that the participants differed in terms of age, chronic conditions, retirement age and frailty, according to the type of retirement. Frailty could be assessed for 1366 out of 1864 responders. Participants with incomplete questionnaire data to determine frailty were older, consumed less alcohol and had less of CVD and diabetes in the year 2000. No statistical differences in retirement characteristics or smoking were observed between the groups (data not shown). Age at and type of retirement In table 1, a declining trend of mean retirement age was observed across those retiring because of old age (63.5 years), early old age (61.0 years), those receiving pension benefit packages (60.6 years), unemployment (59.1 years) and disability (57.7 years), p<0.001. The percentage of men who reported retirement transitions beyond age 65 years was 13.4% among old age retirees, 5.7% among those receiving pension benefit packages, 3.6% among early old age retirees, and minimal among unemployment (0%) and disability (0.6%) retirees (data not shown). Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 4HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access Table 1 Characteristics of the men according to retirement type (proportion and per cent unless stated otherwise) All Old age retirement Early old age retirement Pension benefit package Unemployment retirement Disability retirement P value* n (%) 1324 507 (37.1) 288 (21.1) 251 (18.4) 103 (7.5) 175 (12.8) Old age characteristics assessed in the year 2000 (n=1324) Age in years, mean (SD) 73.3 (4.1) 74.6 (4.0) 72.9 (4.0) 72.5 (3.9) 71.4 (3.1) 72.6 (4.1) <0.001 Cardiovascular disease, n (%) 812 (59.4) 307 (60.6) 159 (55.2) 155 (61.8) 47 (45.6) 124 (70.9) <0.001 Diabetes, n (%) 113 (8.6) 35 (6.9) 22 (7.6) 26 (10.4) 7 (6.8) 23 (13.1) 0.018 Smoking status, n (%) (n=1322) 0.371 Current smoker 106 (8.0) 35 (6.9) 19 (6.6) 27 (10.8) 10 (9.7) 15 (8.6) Quit smoking 719 (54.8) 269 (53.3) 160 (55.6) 129 (51.4) 59 (57.3) 102 (58.3) Never smoked 497 (37.2) 201 (39.8) 109 (37.8) 95 (37.8) 34 (33.0) 58 (33.1) Alcohol consumption, n (%) (n=1040) <0.001 None 153 (14.7) 68 (16.9) 30 (12.9) 23 (11.3) 11 (13.1) 21 (17.8) Light 448 (43.1) 176 (43.8) 114 (48.9) 79 (38.9) 36 (42.9) 43 (36.4) Moderate 234 (22.5) 103 (25.6) 39 (16.7) 47 (23.2) 18 (21.4) 27 (22.9) High 205 (19.7) 55 (13.7) 50 (21.5) 54 (26.6) 19 (22.6) 27 (22.9) Age at retirement (n=1284) In years, mean (SD) 61.3 (4.3) 63.5 (3.6) 61.0 (3.7) 60.6 (3.7) 59.1 (3.7) 57.7 (4.4) <0.001 Categorised age at retirement, n (%) (n=1284) <0.001 ≥70 years 23 (1.8) 14 (2.9) 4 (1.4) 5 (2.0) 0 0 68–69 years 15 (1.2) 11 (2.3) 1 (0.4) 2 (0.8) 0 1 (0.6) 66–67 years 52 (4.1) 40 (8.2) 5 (1.8) 7 (2.9) 0 0 64–65 years 273 (21.3) 204 (41.8) 35 (12.5) 21 (8.6) 6 (6.0) 7 (4.1) 62–63 years 292 (22.8) 126 (25.8) 82 (29.2) 45 (18.4) 15 (15.0) 24 (14.0) 60–61 years 321 (24.9) 69 (14.1) 85 (30.2) 90 (36.9) 38 (38.0) 39 (22.8) 58–59 years 127 (9.9) 11 (2.3) 41 (14.6) 37 (15.2) 13 (13.0) 25 (14.6) 56–57 years 87 (6.8) 3 (0.6) 17 (6.0) 20 (8.2) 17 (17.0) 30 (17.5) ≤55 years 94 (7.3) 10 (2.0) 11 (3.9) 17 (7.0) 11 (11.0) 45 (26.3) Frailty assessed in the year 2000, n (%) (n=1324) <0.001 Nonfrail 496 (37.5) 177 (34.9) 108 (37.5) 123 (49.0) 44 (42.7) 44 (25.1) Prefrail 686 (51.8) 281 (55.4) 152 (52.8) 110 (43.8) 49 (47.6) 94 (53.7) Frail 142 (10.7) 49 (9.7) 28 (9.7) 18 (7.2) 10 (9.7) 37 (21.1) Stressfulness of work career, assessed in the year 2003 (n=972) <0.001 0 (high stress) to 10 (no stress) 5.3 (2.8) 5.7 (2.9) 5.2 (2.8) 4.9 (2.7) 5.7 (3.0) 4.6 (2.7) *P for difference in means/distributions between groups. Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 5 HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access Age at retirement and frailty in old age Retirement age was similar for nonfrail (mean 61.4 years, SD 4.3) and prefrail (mean 61.4 years, SD 4.2) men, and younger among frail men (mean 60.4 years, SD 4.5; p for trend=0.059). The proportions of nonfrail, prefrail and frail men in the retirement age groups are shown in figure 1. The prevalence of frailty in old age was lowest (3.8%) among men who had retired at ages 66–67 years and lower than the cohort average (8.7%) among men who had retired at ages 60–63 years. Conversely, the prevalence of frailty in old age was well above the cohort average (10.7%) among men retiring at age 57 years or younger (17.6%) and at age 70 years and older (13.0%). The proportions of nonfrail, prefrail and frail participants varied according to categories of retirement age (p for trend=0.015). In table 2, compared with those who retired at age 55 years or younger, those retiring between the ages of 64 and 65 years were at decreased risk of prefrailty (adjusted OR 0.40, 95% CI 0.21 to 0.64) and frailty (adjusted OR 0.21, 95% CI 0.10 to 0.46), relative to nonfrailty. The risks of frailty in old age relative to nonfrailty were generally lower for men retiring between the ages of 58 and 69 years compared with men retiring at age 55 years or younger (adjusted ORs between 0.07 and 0.29). The point estimates showed a decreasing trend for men who retired at ages 56–69 years, and then increased in men retiring at the age of 70 years and older, compared with men retiring at age 55 years or younger. However, the differences among ORs were not statistically significant (likelihood ratio test for model 1: p=0.103; model 2: p=0.173). The findings showed a similar trend for prefrailty. Online supplemental table 2 presents associations between age at retirement and frailty separately for each retirement type and shows that the direction of the point estimates for the associations between age at retirement and frailty was generally like those observed for the whole cohort. Type of retirement and frailty in old age In table 1, the prevalence of frailty in old age was lowest for men receiving pension benefit packages (7.2%), and slightly higher for men retiring due to old age (9.7%), early old age (9.7%) and unemployment (9.7%). The prevalence of frailty was almost twice the cohort average (21.1%) among disability retirees (p≤0.001). Figure 2 shows the proportions of nonfrail, prefrail and frail participants according to the type of retirement. In table 2, compared with men who had retired due to old age, those who retired due to disability were at increased risk of prefrailty (ageadjusted OR 1.62, 95% CI 1.07 to 2.45) and frailty (ageadjusted OR 4.34, 95% CI 2.47 to 7.63), relative to nonfrailty. The association Table 2 ORs and 95% CIs of prefrailty and frailty compared with nonfrailty according to age at and type of retirement Prefrailty Frailty OR (95% CI) OR (95% CI) Model 1† Model 2‡ Model 1† Model 2‡ Categories of retirement age (years) ≥70 0.78 (0.25 to 2.46) 0.83 (0.26 to 2.62) 0.28 (0.06 to 1.40) 0.38 (0.08 to 1.97) 68–69 0.51 (0.15 to 1.73) 0.55 (0.16 to 1.89) 0.11 (0.01 to 1.03) 0.10 (0.01 to 1.02) 66–67 0.49 (0.23 to 1.04) 0.52 (0.24 to 1.12) 0.07 (0.01 to 0.34)*** 0.07 (0.01 to 0.37)** 64–65 0.37 (0.21 to 0.64)*** 0.40 (0.22 to 0.70)*** 0.19 (0.09 to 0.40)*** 0.21 (0.10 to 0.46)*** 62–63 0.61 (0.35 to 1.06) 0.67 (0.38 to 1.17) 0.20 (0.09 to 0.45)*** 0.24 (0.11 to 0.54)*** 60–61 0.49 (0.29 to 0.85)** 0.48 (0.28 to 0.83)** 0.22 (0.10 to 0.47)*** 0.22 (0.10 to 0.47)*** 58–59 0.45 (0.24 to 0.83)** 0.46 (0.25 to 0.87)* 0.28 (0.12 to 0.66)** 0.29 (0.12 to 0.70)** 56–57 0.50 (0.26 to 0.98)* 0.49 (0.25 to 0.98)* 0.51 (0.21 to 1.24) 0.45 (0.18 to 1.14) ≤55 (Ref) (Ref) (Ref) (Ref) Type of retirement Old age (Ref) (Ref) (Ref) (Ref) Early old age 1.02 (0.74 to 1.40) 0.99 (0.72 to 1.37) 1.22 (0.71 to 2.09) 1.24 (0.72 to 2.16) Pension benefit package 0.66 (0.47 to 0.91)* 0.61 (0.44 to 0.85)** 0.71 (0.39 to 1.30) 0.62 (0.33 to 1.15) Unemployment 0.89 (0.56 to 1.41) 0.86 (0.54 to 1.37) 1.36 (0.62 to 2.98) 1.44 (0.64 to 3.21) Disability 1.62 (1.07 to 2.45)* 1.46 (0.96 to 2.23) 4.34 (2.47 to 7.63)*** 3.52 (1.97 to 6.29)*** Stressfulness of work career Per 1- unit increase 0.94 (0.89 to 0.99)* 0.94 (0.90 to 0.99)* 0.94 (0.86 to 1.03) 0.95 (0.87 to 1.04) *P<0.001; **P<0.01; ***P<0.05. †Model 1 adjusted for age in the year 2000 (age at measuring frailty). ‡Model 2 adjusted additionally for smoking, alcohol consumption, cardiovascular disease and diabetes. Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 6HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access for frailty persisted after adjustment for smoking, alcohol use, CVD and OR 3.52 (95% CI 1.97 to 6.29). Compared with men who retired due to old age, men who received pension benefit packages were at decreased risk of prefrailty (adjusted OR 0.61, 95% CI 0.44 to 0.85), but not with frailty, relative to nonfrailty. Stressfulness of work career In additional analyses among the men (n=957) who had rated the stressfulness of their entire work career, each 1- unit increase in the stressfulness scale, indicating lower overall stress during the work career, was associated with a decreased risk of prefrailty (model 2- adjusted OR 0.94, 95% CI 0.90 to 0.99), relative to nonfrailty. The finding for frailty was parallel but did not reach statistical significance. DISCUSSION The HBS provided a unique opportunity to study the association between age at retirement as an indicator of the length of work career and frailty in old age in a homogenous population of men with a high socioeconomic standing. Our findings suggest that transitioning into retirement at an older age was associated with a decreased risk of subsequent frailty in old age among the men, which persisted even after allowing for lifestyle factors and main chronic conditions. However, retiring at age 70 years or older yielded no further health benefit as the proportion of frail men started to increase in this group. Men who retired due to disability were at increased risk of prefrailty and frailty when they were compared against men retiring due to old age. Additionally, we found evidence suggesting an association between higher stressfulness during the entire work career and increased risk of prefrailty. The present study corroborates and extends the scarce findings reported on the association between younger retirement age and increased risk of frailty. A recent English study found that men who exited the labour market prematurely at age 49 or 60 years were at increased risk of frailty compared with men who worked full time until the age of 65 years.10 The study found the frailty trajectory of men who entered working life late and exited at age 60 years to rise steeper than that of other groups. These men were often educated professionals and, while few in the study, highlight the need to understand these phenomena among men in higher socioeconomic groups. Men in the present study had mostly worked as business executives and managers in enterprises in the fields of industry and commerce.13 They had retired on average 2 years younger than men in other Finnish cohorts with similar occupational backgrounds,12 20 but at an older age than the national average for all men in Finland in the year 2000, which was 58.9 years.21 22 Some of the men in the present study were entrepreneurs who do not automatically have a statutory retirement scheme and thus transitioning into retirement is more flexible.23 In addition, the men had reported to be in good physical health and represented the highest socioeconomic class, both of which have been found to be protective of frailty24 25 and to predict longer work careers.12 26 27 Despite the positive influence of socioeconomic factors in the present study, we found the prevalence of frailty to be 10.8%, which is comparable to or even higher than that of communitydwelling cohorts of similar age.28 While information on the reason to continue to be occupationally active in old age was not available, we believe that as business executives the participants continued working also for reasons other than financial. Previously, continuing to be occupationally active in old age by choice and because of enjoyable work associated with better quality of life, as opposed to those continuing to work due to financial reasons, which associated with worse quality of life.29 In the present study, those who were occupationally active at and beyond age 65 years had rated the stressfulness of their work career to be less stressful than those who had retired by age 65 years. Ultimately, not all studies report beneficial health effects of working beyond the statutory pension age. In a British cohort, occupational activity in old age was not associated with better or worse health in fully adjusted analyses.30 We observed the proportion of individuals who were frail in old age to increase towards both extremes of age at retirement; the prevalence of frailty was above the cohort average among men exiting the workforce before age 55 years, as well as among those who continued being occupationally active at and beyond the age of 70 years. The risk of frailty was lowest for those who retired between the ages of 60 and 69 years. Although the number of men who were occupationally active at and beyond age 70 years was small, the finding of increasing frailty in old age in this group is relevant given recent increases in retirement age. Additional analyses showed that men who were occupationally active at age 70 years and older did not differ from men retiring at age 69 years or younger in terms of smoking and drinking habits, as well as the prevalence of chronic conditions including CVD and diabetes. Working life can impose stress on employees particularly in managerial positions31 which could facilitate and accelerate agerelated decline Figure 2 Distribution of frailty classification assessed at an average age of 73.3 years according to the type of retirement. Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 7 HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access across physiological systems. Workrelated psychosocial stress has been associated with disability,32 use of hospital care,33 as well as premature mortality.34 Moreover, there is evidence of an association between psychosocial work characteristics, such as low reward, high effort, effort to reward ratio, and effort to control ratio, and frailty in old age.35 We found evidence that those who exited working life prematurely perceived their work career to have been more stressful. Business executives who had transitioned into disability retirement were at increased risk of prefrailty and frailty in old age, when the men were compared against men who had retired due to old age. Frail individuals are at risk of negative employment outcomes including unemployment and leaving work for health reasons,9 even in higher managerial positions.36 While transitioning into disability retirement may allude poor health and highlight susceptibility to various health problems, associations between disability retirement and frailty were independent of age, smoking, alcohol use, CVD and diabetes in the present study. Business executives who received pension benefit packages, for example, early retirement incentives or redundancy packages, were at decreased risk of prefrailty, but not frailty, compared with men retiring due to old age. These packages may have included, alongside pension benefits, access to healthcare during retirement, which may also have contributed to better health in old age. We were able to investigate associations between age at and type of retirement and subsequent frailty in old age in a homogenous population comprising retired business executives belonging to the higher socioeconomic groups, of whom more than 140 could be classified as frail. The cohort included businessmen only working in nonmanual occupations at public or private institutions, or as entrepreneurs, minimising potential confounding related to sex, socioeconomic status, as well as work characteristics. We expect the occurrence of career breaks low in this group, as well as a high recall rate of retirement characteristics, given that data on retirement were selfreported. However, we were not able to characterise the participants further with information related to education, income or wealth, as previous studies have.37 We did sensitivity analyses excluding participants (n=5) with any inconsistent information on retirement (eg, a few men reported that they transitioned to early old age pension but that their retirement age was 68 years or more). These results were parallel to those observed in the present study. Cohort effects may be present as the criteria for retirement benefits may have varied with time. Though the lower age limits for old age and early old age pensions were extremely stable in Finland from the 1950s to the 1990s, that of unemployment pensions varied between 55 and 60 years during this time. Furthermore, we found no evidence of the existence of birth cohort effects in analyses stratified by birth cohort. We also found no evidence that the association between age at retirement and frailty would contradict our main hypothesis, in that retiring older would associate with less frailty, in analyses stratified by retirement type. An extraordinary recession hit the Finnish economy during 1991–1994, leaving many senior employees out of employment, giving rise to unemployment retirement for workers that otherwise could have qualified for old age retirement. In Ireland, worsening economic news during a banking crisis coincided with less intentions to retire at the state pension age at the time.38 While the first law on Occupational Safety was passed in Finland in the year 1930, working conditions are subject to change over time. Although the definition of frailty used in the present study comprised four criteria and was assessed using questionnaires, it has been validated to be predictive of outcomes including slower walking speed, disability and mortality in this cohort.16 Mortality and loss to followup may have resulted in survivor effects, leaving out participants in poor health. This can potentially undermine the associations in the present study, as we expect that the potentially most frail participants, who died before the questionnaire sent in 2000, also retired earlier. As executives and managers, the participants in the present study are likely to have been in better overall health than men who were unemployed or in other occupational groups, summarised in ‘the healthy worker effect’.39 To control this and possible reverse causality, we adjusted the analyses with factors known to be associated with retirement characteristics. Hence, generalisations of the results to other occupational or socioeconomic groups, women or other ethnicities should be made with caution. In conclusion, exiting working life early and continuing to be occupationally active until older age (age 70 years and older) were both associated with an increased risk of frailty in old age among businessmen and managers. However, this needs to be further studied in larger and more diverse populations. Postponing retirement decisions in professional nonmanual occupations was not observed to be disadvantageous for frailty, provided that the retirement transition occurred before age 70 years. Promotion of longer work careers could also promote healthier ageing, which was supported in the present study by the fact that the lowest prevalence of frailty (3.6%) was observed in the men who retired at ages 66–67 years, which is clearly older than the average recorded for Finnish men. Author affiliations 1Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Helsinki, Finland 2Folkhälsan Research Centre, Helsinki, Finland 3Department of Health Sciences, Gerontology Research Centre, University of Jyväskylä Faculty of Sport and Health Sciences, Jyvaskyla, Finland 4School of Business and Kokkola University Consorium Chydenius, University of Vaasa, Vaasa, Finland 5Department of Management and Leadership, Jyväskylä University School of Business and Economics, Jyväskylä, Finland 6Department of Medicine, University of Helsinki and Helsinki University Hospital, Helsinki, Finland 7Center for Life Course Health Research, University of Oulu, Oulu, Finland Twitter Mikaela B von Bonsdorff @vonbonsdorffx2 Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from 8HaapanenMJ, etal. BMJ Open 2020;10:e037722. doi:10.1136/bmjopen-2020-037722 Open access Contributors MJH designed the present study, drafted the article, analysed the data and drafted the manuscript. MBvB designed the present study, interpreted the data and revised the article critically for important intellectual content. MEvB, NMP, TT and AYS interpreted the data and revised the article critically for important intellectual content. TES acquired funding for the cohort, data collection, interpreted the data and revised the article critically for important intellectual content. Funding The Academy of Finland supported MBvB with grant 257239. The Academy of Finland supported TT with grant 286536. The Academy of Finland supported MEvB with grants 294530, 307114 and 303920. TES is supported by the Helsinki University Hospital, Department of Medicine. Disclaimer The sponsor had no role in the design, methods, subject recruitment, data collections, analysis and preparation of the paper. Competing interests None declared. Patient consent for publication Not required. Ethics approval The Ethics Committee of the Department of Medicine, University of Helsinki, has approved the research protocol of the followup study of HBS. All participants provided a written informed consent before participating in the study. The study adheres with the principles stated in the Declaration of Helsinki. Provenance and peer review Not commissioned; externally peer reviewed. Data availability statement Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available. Inquiries regarding the data sets used and/or analysed during the current study can be directed to the principal investigator (TES) of the Helsinki Businessmen Study. Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peerreviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BYNC 4.0) license, which permits others to distribute, remix, adapt, build upon this work noncommercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is noncommercial. See:http:// creativecommons. org/ licenses/ bync/ 4. 0/. ORCID iDs Markus JHaapanen http:// orcid. org/ 0000- 0002- 8632- 7906 Arto YStrandberg http:// orcid. org/ 0000- 0002- 6080- 1729 REFERENCES 1 Fried LP, Tangen CM, Walston J, etal. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146–57. 2 Bergman H, Ferrucci L, Guralnik J, etal. Frailty: an emerging research and clinical paradigmissues and controversies. J Gerontol A Biol Sci Med Sci 2007;62:731–7. 3 United Nations, Department of Economic and Social Affairs PD. World population prospects: the 2015 revision, key findings and advance tables Working Paper No. ESA/P/WP.241. New York, 2015. 4 Morley JE, Vellas B, van Kan GA, etal. Frailty consensus: a call to action. J Am Med Dir Assoc 2013;14:392–7. 5 Cesari M, Prince M, Thiyagarajan JA, etal. Frailty: an emerging public health priority. J Am Med Dir Assoc 2016;17:188–92. 6 Oshio T, Kan M. The dynamic impact of retirement on health: evidence from a nationwide tenyear panel survey in Japan. Prev Med 2017;100:287–93. 7 Rechel B, Grundy E, Robine J- M, etal. Ageing in the European Union. Lancet 2013;381:1312–22. 8 Cesari M, Prince M, Thiyagarajan JA, etal. Frailty: an emerging public health priority. J Am Med Dir Assoc 2016;17:188–92. 9 Schinkel- Ivy A, Mosca I, Mansfield A. Factors contributing to unexpected retirement and unemployment in adults over 50 years old in Ireland. Gerontol Geriatr Med 2017;3:2333721417722709:2333 72141772270. 10 Lu W, Benson R, Glaser K, etal. Relationship between employment histories and frailty trajectories in later life: evidence from the English longitudinal study of ageing. J Epidemiol Community Health 2017;71:439–45. 11 Official Statistics of Finland (OSF). Labour Force Survey [e- Publication]. Helsinki, Finland, 2016. http:// tilastokeskus. fi/ til/ tyti/ index_ en. html 12 Virtanen M, Oksanen T, Pentti J, etal. Occupational class and working beyond the retirement age: a cohort study. Scand J Work Environ Health 2017;43:426–35. 13 Strandberg TE, Salomaa V, Strandberg AY, etal. Cohort profile: the Helsinki businessmen study (HBs). Int J Epidemiol 2016;45:1074–1074h. 14 Miettinen TA, Huttunen JK, Naukkarinen V, etal. Multifactorial primary prevention of cardiovascular diseases in middleaged men. risk factor changes, incidence, and mortality. JAMA 1985;254:2097–102. 15 von Bonsdorff MB, Haapanen MJ, Törmäkangas T, etal. Midlife cardiovascular status and old age physical functioning trajectories in older businessmen. J Am Geriatr Soc 2019;67:2490–6. 16 Sirola J, Pitkala KH, Tilvis RS, etal. Definition of frailty in older men according to questionnaire data (RAND-36/SF-36): the Helsinki businessmen study. J Nutr Health Aging 2011;15:783–7. 17 Järnefelt N. Education and longer working lives: a longitudinal study on Eduvation differences in the late exit from working life of older employees in Finland. 18 RAND. The Rand 36- Item health survey the Rand 36- Item health survey. Heal 1992;2:91–7. 19 Strandberg AY, Trygg T, Pitkälä KH, etal. Alcohol consumption in midlife and old age and risk of frailty. Age Ageing 2018;47:248–54. 20 von Bonsdorff MB, von Bonsdorff ME, Haanpää M, etal. Workloss years among people diagnosed with diabetes: a reappraisal from a life course perspective. Acta Diabetol 2018;55:485–91. 21 Working Group on Separate Pension Systems. Separate pension systems report of the Working group. Minist Soc Aff Heal 2019 http:// julkaisut. valtioneuvosto. fi/ handle/ 10024/ 161385 22 Kannisto J. Eläkeiän mittaaminen [Measuring retirement age]. Finnish Cent Pensions 2006:156. https://www. julkari. fi/ bitstream/ handle/ 10024/ 129385/ Elak keel lesi irty misi anmi ttaa minen. pdf? sequence= 1& isAllowed=y 23 Salonen J. Statutory pensions for the selfemployed: working life, confirmed income and financing. Cent Pensions, Res Reports, 2015. Available: https://www. etk. fi/ en/ julkaisu/ statutorypensions- forthe- selfemployed- workinglife- confirmedincome- andfinancing/ 24 Huohvanainen E, Strandberg AY, Stenholm S, etal. Association of selfrated health in midlife with mortality and old age frailty: a 26- Year followup of initially healthy men. J Gerontol A Biol Sci Med Sci 2016;71:923–8. 25 Stolz E, Mayerl H, Waxenegger A, etal. Impact of socioeconomic position on frailty trajectories in 10 European countries: evidence from the survey of health, ageing and retirement in Europe (2004- 2013). J Epidemiol Community Health 2017;71:73–80. 26 de Wind A, van der Pas S, Blatter BM, etal. A life course perspective on working beyond retirementresults from a longitudinal study in the Netherlands. BMC Public Health 2016;16:499. 27 van der Zwaan GL, Oude Hengel KM, Sewdas R, etal. The role of personal characteristics, work environment and context in working beyond retirement: a mixedmethods study. Int Arch Occup Environ Health 2019;92:535–49. 28 Collard RM, Boter H, Schoevers RA, etal. Prevalence of frailty in communitydwelling older persons: a systematic review. J Am Geriatr Soc 2012;60:1487–92. 29 Di Gessa G, Corna L, Price D, etal. The decision to work after state pension age and how it affects quality of life: evidence from a 6- year English panel study. Age Ageing 2018;47:450–7. 30 Di Gessa G, Corna LM, Platts LG, etal. Is being in paid work beyond state pension age beneficial for health? Evidence from England using a lifecourse approach. J Epidemiol Community Health 2017;71:431–8. 31 Burke RJ. Sources of managerial and professional stress in large organizations. In: Causes, coping and consequences of stress at work. New York: John Wiley and Sons Ltd, 1988: 77–114. 32 Kulmala J, von Bonsdorff MB, Stenholm S, etal. Perceived stress symptoms in midlife predict disability in old age: a 28- year prospective cohort study. J Gerontol A Biol Sci Med Sci 2013;68:984–91. 33 von Bonsdorff MB, von Bonsdorff M, Kulmala J, etal. Job strain in the public sector and hospital inpatient care use in old age: a 28- year prospective followup. Age Ageing 2014;43:393–9. 34 von Bonsdorff MB, Seitsamo J, von Bonsdorff ME, etal. Job strain among bluecollar and whitecollar employees as a determinant of Kirjasto/Kausijulkaisut. Protected by copyright. on January 19, 2021 at Jyvaskylan Yliopostohttp://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-037722 on 17 December 2020. Downloaded from