Age at retirement and long term survival of an industrial population:
http://www.100md.com
《英国医生杂志》
1 Shell Health Services, Shell Oil Company, 910 Louisiana, Houston, TX 77002, USA, 2 Shell International, Hague, Netherlands
Correspondence to: S P Tsai shan.tsai@shell.com
Objective To assess whether early retirement is associated with better survival.
Design Long term prospective cohort study.
Setting Petroleum and petrochemical industry, United States.
Subjects Past employees of Shell Oil who retired at ages 55, 60, and 65 between 1 January 1973 and 31 December 2003.
Main outcome measure Hazard ratio of death adjusted for sex, year of entry to study, and socioeconomic status.
Results Subjects who retired early at 55 and who were still alive at 65 had a significantly higher mortality than those who retired at 65 (hazard ratio 1.37, 95% confidence interval 1.09 to 1.73). Mortality was also significantly higher for subjects in the first 10 years after retirement at 55 compared with those who continued working (1.89, 1.58 to 2.27). After adjustment, mortality was similar between those who retired at 60 and those who retired at 65 (1.06, 0.92 to 1.22). Mortality did not differ for the first five years after retirement at 60 compared with continuing work at 60 (1.04, 0.82 to 1.31).
Conclusions Retiring early at 55 or 60 was not associated with better survival than retiring at 65 in a cohort of past employees of the petrochemical industry. Mortality was higher in employees who retired at 55 than in those who continued working.
Few studies have evaluated the effect of early retirement on survival.1-4 A longitudinal evaluation of survival patterns requires a relatively long length of follow-up after retirement. In addition, data on age and health status at retirement are not readily available for the US general population. To date no consensus has been reached on the survival or mortality of people who retire early compared with those who retire later.4 Some researchers concluded that early retirement harms health, attributing this to illness before retirement or the change of life events associated with retirement.1 3 On the other hand, there is a widespread perception that early retirement is associated with longer life expectancy and that retiring later leads to early death.5 6 The possible health benefits of retirement, such as reduced role demand and a more relaxed lifestyle, have been postulated to improve longevity among people who retire early.2
We carried out a long term prospective cohort study of employees of the petrochemical industry in the United States who retired at 55, 60, and 65 to assess whether there is any survival advantage of early retirement.
Subjects and methods
Our study population consisted of all past employees of Shell Oil in the United States who retired at 55, 60, or 65, and employees who were actively working at 55 or 60, during a period of 31 years between 1 January 1973 and 31 December 2003. In the main analysis we compared the survival of employees who retired at 55 (n = 839) and 60 (n = 1929) and were still alive at 65 with those who retired at 65 (n = 900). The average ages at the end of the study were 72, 76, and 80, with, respectively, 10%, 20%, and 52% over 80. We followed up employees who retired at 55 or 60 from the time they reached 65. We excluded the first 10 years of survival for those retiring at 55, and we determined time to death from age 65, or the end of the study, whichever was earlier, yielding 21 years of follow-up and 173 deaths. We excluded the first five years of survival for those retiring at 60, yielding 26 years of follow-up and 581 deaths.
To assess the health status of those who retired before the normal retirement age of 65, we compared the mortality of those who retired at 55 (n = 1439) or 60 (n = 2116) with those who were still working at 55 (n = 15 543) or 60 (n = 6783). If early retirement before 65 was taken by some workers because of failing health, mortality would be expected to be higher among early retirees than among those who did not retire at these ages. Overall, 137 employees who retired at 55 and 994 employees who continued working at 55 died by age 65, whereas 98 employees who retired at 60 and 317 employees who continued working at 60 died by age 65. We calculated the hazard ratio for the first 10 years after retirement (until 65) of those who retired at 55, and for the first five years (until 65) for those who retired at 60.
We identified subjects through Shell Oil's health surveillance system.7 This system was established in 1979 and contains data on vital status and other health related variables for all US employees of the company from 1973 onwards. We used several sources to determine the vital status of the subjects as of 31 December 2003. We supplemented the company records with results of a data linkage search with the National Death Index, which has been shown to ascertain 97% of deaths since 1979.8 9 Additionally, we carried out a data linkage search between company records and the master beneficiary record file of the Social Security Administration to identify any deaths that may have been missed by the National Death Index. We assumed subjects to be alive if they were not identified by the above sources. Details of this process have been reported elsewhere.10
The outcome variables for the survival analysis were time to death or end of the study, and the censoring variable (dead or living). Covariates in the analysis included categorical variables representing sex and employment grade, and a continuous variable representing the calendar year that the subjects entered the study. Adjustment for year of entry to the study controlled for the effect of changes in mortality trends over the study period. We used employment grade as a proxy for socioeconomic status, a high status being assigned for employees in managerial or professional positions and a low status being assigned for those in skilled, semiskilled, non-skilled, or clerical positions.
Fig 1 Kaplan-Meier survival curves for employees of Shell Oil, United States, who retired early at 55 and survived to 65 and those who retired at 65
Fig 2 Kaplan-Meier survival curves for employees of Shell Oil, United States, who retired early at 60 and survived to 65 and those who retired at 65
We used Kaplan-Meier survival curves to graphically present the unadjusted survival of subjects who retired early at 55 or 60 compared with those who retired at 65. We used the log rank test for the equality of survivor functions to determine whether there was a significant difference (P < 0.05) between the survival curves.
We used the Cox proportional hazards model to estimate the hazard ratios of death between the early and normal retirement groups,11 with adjustment for sex, calendar year of entry into the study, and socioeconomic group. The Wald 2 test was used to test the significance of the hazard ratios. All statistical analyses were carried out using SAS version 8.2 and Stata version 8.0.
Results
Women accounted for about 11% (10% among early retirees and 12% among those retiring at 65) of the study population during follow-up. More than half of early retirees who reached 65 (57% who retired at 55 and 53% who retired at 60) were in the high socioeconomic group whereas less than half (44%) of those who retired at 65 were in this group (table 1). The Kaplan-Meier survival curve shows a consistently lower probability of survival for employees who retired at 55 (173 deaths) than for those who retired at 65 (462 deaths; fig 1). This difference was not, however, statistically significant (P = 0.09, log rank test). Likewise, the risk of death was similar between these two retirement groups (unadjusted hazard ratio 1.17, 95% confidence interval 0.98 to 1.41). For most of the follow-up period the survival curves were similar for employees who retired at 60 and those who retired at 65 (581 and 541 deaths). On the basis of a small number of deaths (n = 51), however, the probability of survival for the group who retired early at 60 was higher in the last eight years (after age 83; fig 2). The log rank test did not support a significant difference between the groups (P = 0.20) and this early retirement group was not at increased risk of death (hazard ratio 0.92, 0.82 to 1.04).
Table 1 Characteristics of employees of Shell Oil, United States, from 1 January 1973 to 31 December 2003. Values are numbers (percentages) unless stated otherwise
After adjustment for sex, calendar year of entry to the study, and socioeconomic status, employees who retired at 55 and were still alive at 65 had significantly higher mortality than those who retired at 65 (hazard ratio 1.37, 1.09 to 1.73; table 2). The risk of dying was about 80% greater in men than it was in women (1.83, 1.34 to 2.48) and retirees in the low socioeconomic category had a higher mortality than retirees in the high category (1.17, 1.01 to 1.36). Mortality was lower among more recent retirees (0.98, 0.97 to 1.00). Among employees who retired at 60, mortality was similar to those who retired at 65 (hazard ratio 1.06, 0.92 to 1.22). Men had a higher mortality than women (1.48, 1.16 to 1.89), and mortality was lower among more recent retirees (0.97, 0.96 to 0.99). Mortality did not, however, differ significantly by socioeconomic status (1.09, 0.97 to 1.23; table 2).
Table 2 Adjusted hazard ratios (95% confidence intervals) by explanatory variables for employees of Shell Oil, United States, who retired early at 55 or 60 compared with those who retired at 65
Regardless of socioeconomic status, employees who retired at 55 had a higher risk of mortality than those who retired at 60 (table 3). Employees who retired at 55 in the high socioeconomic group had a 20% increase in risk of death (hazard ratio 1.21, 0.88 to 1.67), whereas employees who retired at 55 in the low socioeconomic group had nearly a 60% increase in risk (1.58, 1.15 to 2.18). We found no difference in mortality by socioeconomic status in the group that retired at 60.
Table 3 Adjusted hazard ratios (95% confidence intervals) for early retirement by socioeconomic status
We could not assess directly the issue of whether employees who retired at a younger age were in poorer health than those who retired later as data were not available to identify the type of retirement for each employee (for example, retirement due to disability compared with normal retirement). If early retirement before 65 is taken by some workers because of failing health, however, mortality would be expected to be higher early in retirement. Table 4 shows the mortality for the first 10 years after retirement at 55 and for the first five years after retirement at 60 compared with employees who continued working at these ages. Regardless of socioeconomic status, employees who retired at 55 had almost a twofold higher mortality than those who continued working (1.89, 1.58 to 2.27). The mortality of employees who retired at 60 was similar to those who continued working at 60 (1.04, 0.82 to 1.31). Employees in the high socioeconomic group who retired at 60, however, had a 20% lower mortality (0.81, 0.54 to 1.20), whereas employees in the low socioeconomic group had a 20% increased mortality (1.21, 0.91 to 1.61).
Table 4 Adjusted hazard ratios (95% confidence intervals) for first 10 years after retirement at 55 and for first five years after retirement at 60
Discussion
Haynes SG, McMichael AJ, Tyroler HA. Survival after early and normal retirement. J Gerontol 1978;33: 269-78.
Ekerdt DJ, Baden R, Bosse R, Dibbs E. The effect of retirement on physical health. Am J Public Health 1983;73: 779-83.
Morris JK, Cook DG, Shaper AG. Loss of employment and mortality. BMJ 1994;308: 1135-9.
Trichopoulos D. Any questions? BMJ 1996;312: 632.
Padfield A. Myths in medicine. BMJ 1996;312: 1611.
Lin S. Optimum strategies for creativity and longevity. 2002. http://www.geocities.com/dtmcbride/health/retirement_age.html (accessed 16 Feb 2005).
Joyner RE, Pack PH. The Shell Oil Company's computerized health surveillance system. J Occup Med 1982;24: 812-4.
Cowper DC, Kubal JD, Maynard C, Hynes DM. A primer and comparative review of major US mortality databases. Ann Epidemiol 2002;12: 462-8.
Acquavella JF, Donaleski D, Hanis NM. An analysis of mortality follow-up through the national death index for a cohort of refinery and petrochemical workers. Am J Ind Med 1986;9: 181-7.
Tsai SP, Wendt JK, Cardarelli KM, Fraser AE. A mortality and morbidity study of refinery and petrochemical employees in Louisiana. Occup Environ Med 2003;60: 627-33.
Cox DR. Regression models and life tables (with discussion). J R Stat Soc 1972;34: 187-220.
Mein G, Martikainen P, Hemingway H, Stansfeld S, Marmot M. Is retirement good or bad for mental and physical health functioning? Whitehall II longitudinal study of civil servants. J Epidemiol Community Health 2003;57: 46-9.
Mein G, Martikainen P, Stansfeld SA, Brunner EJ, Fuhrer R, Marmot MG. Predictors of early retirement in British civil servants. Age Ageing 2000;29: 529-36.
Lantz PM, House JS, Lepkowski JM, Williams DR, Mero RP, Chen J. Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA 1998;279: 1703-8.
Chandola T. Social class differences in mortality using the new UK national statistics socio-economic classification. Soc Sci Med 2000;50: 641-9.
Mackenbach JP, Bos V, Andersen O, Cardano M, Costa G, Harding S, et al. Widening socioeconomic inequalities in mortality in six Western European countries. Int J Epidemiol 2003;32: 830-7.
Muntaner C, Hadden WC, Kravets N. Social class, race/ethnicity and all-cause mortality in the US: longitudinal results from the 1986-1994 national health interview survey. Eur J Epidemiol 2004;19: 777-84.(Shan P Tsai, manager, epidemiology1, Jud)
Correspondence to: S P Tsai shan.tsai@shell.com
Objective To assess whether early retirement is associated with better survival.
Design Long term prospective cohort study.
Setting Petroleum and petrochemical industry, United States.
Subjects Past employees of Shell Oil who retired at ages 55, 60, and 65 between 1 January 1973 and 31 December 2003.
Main outcome measure Hazard ratio of death adjusted for sex, year of entry to study, and socioeconomic status.
Results Subjects who retired early at 55 and who were still alive at 65 had a significantly higher mortality than those who retired at 65 (hazard ratio 1.37, 95% confidence interval 1.09 to 1.73). Mortality was also significantly higher for subjects in the first 10 years after retirement at 55 compared with those who continued working (1.89, 1.58 to 2.27). After adjustment, mortality was similar between those who retired at 60 and those who retired at 65 (1.06, 0.92 to 1.22). Mortality did not differ for the first five years after retirement at 60 compared with continuing work at 60 (1.04, 0.82 to 1.31).
Conclusions Retiring early at 55 or 60 was not associated with better survival than retiring at 65 in a cohort of past employees of the petrochemical industry. Mortality was higher in employees who retired at 55 than in those who continued working.
Few studies have evaluated the effect of early retirement on survival.1-4 A longitudinal evaluation of survival patterns requires a relatively long length of follow-up after retirement. In addition, data on age and health status at retirement are not readily available for the US general population. To date no consensus has been reached on the survival or mortality of people who retire early compared with those who retire later.4 Some researchers concluded that early retirement harms health, attributing this to illness before retirement or the change of life events associated with retirement.1 3 On the other hand, there is a widespread perception that early retirement is associated with longer life expectancy and that retiring later leads to early death.5 6 The possible health benefits of retirement, such as reduced role demand and a more relaxed lifestyle, have been postulated to improve longevity among people who retire early.2
We carried out a long term prospective cohort study of employees of the petrochemical industry in the United States who retired at 55, 60, and 65 to assess whether there is any survival advantage of early retirement.
Subjects and methods
Our study population consisted of all past employees of Shell Oil in the United States who retired at 55, 60, or 65, and employees who were actively working at 55 or 60, during a period of 31 years between 1 January 1973 and 31 December 2003. In the main analysis we compared the survival of employees who retired at 55 (n = 839) and 60 (n = 1929) and were still alive at 65 with those who retired at 65 (n = 900). The average ages at the end of the study were 72, 76, and 80, with, respectively, 10%, 20%, and 52% over 80. We followed up employees who retired at 55 or 60 from the time they reached 65. We excluded the first 10 years of survival for those retiring at 55, and we determined time to death from age 65, or the end of the study, whichever was earlier, yielding 21 years of follow-up and 173 deaths. We excluded the first five years of survival for those retiring at 60, yielding 26 years of follow-up and 581 deaths.
To assess the health status of those who retired before the normal retirement age of 65, we compared the mortality of those who retired at 55 (n = 1439) or 60 (n = 2116) with those who were still working at 55 (n = 15 543) or 60 (n = 6783). If early retirement before 65 was taken by some workers because of failing health, mortality would be expected to be higher among early retirees than among those who did not retire at these ages. Overall, 137 employees who retired at 55 and 994 employees who continued working at 55 died by age 65, whereas 98 employees who retired at 60 and 317 employees who continued working at 60 died by age 65. We calculated the hazard ratio for the first 10 years after retirement (until 65) of those who retired at 55, and for the first five years (until 65) for those who retired at 60.
We identified subjects through Shell Oil's health surveillance system.7 This system was established in 1979 and contains data on vital status and other health related variables for all US employees of the company from 1973 onwards. We used several sources to determine the vital status of the subjects as of 31 December 2003. We supplemented the company records with results of a data linkage search with the National Death Index, which has been shown to ascertain 97% of deaths since 1979.8 9 Additionally, we carried out a data linkage search between company records and the master beneficiary record file of the Social Security Administration to identify any deaths that may have been missed by the National Death Index. We assumed subjects to be alive if they were not identified by the above sources. Details of this process have been reported elsewhere.10
The outcome variables for the survival analysis were time to death or end of the study, and the censoring variable (dead or living). Covariates in the analysis included categorical variables representing sex and employment grade, and a continuous variable representing the calendar year that the subjects entered the study. Adjustment for year of entry to the study controlled for the effect of changes in mortality trends over the study period. We used employment grade as a proxy for socioeconomic status, a high status being assigned for employees in managerial or professional positions and a low status being assigned for those in skilled, semiskilled, non-skilled, or clerical positions.
Fig 1 Kaplan-Meier survival curves for employees of Shell Oil, United States, who retired early at 55 and survived to 65 and those who retired at 65
Fig 2 Kaplan-Meier survival curves for employees of Shell Oil, United States, who retired early at 60 and survived to 65 and those who retired at 65
We used Kaplan-Meier survival curves to graphically present the unadjusted survival of subjects who retired early at 55 or 60 compared with those who retired at 65. We used the log rank test for the equality of survivor functions to determine whether there was a significant difference (P < 0.05) between the survival curves.
We used the Cox proportional hazards model to estimate the hazard ratios of death between the early and normal retirement groups,11 with adjustment for sex, calendar year of entry into the study, and socioeconomic group. The Wald 2 test was used to test the significance of the hazard ratios. All statistical analyses were carried out using SAS version 8.2 and Stata version 8.0.
Results
Women accounted for about 11% (10% among early retirees and 12% among those retiring at 65) of the study population during follow-up. More than half of early retirees who reached 65 (57% who retired at 55 and 53% who retired at 60) were in the high socioeconomic group whereas less than half (44%) of those who retired at 65 were in this group (table 1). The Kaplan-Meier survival curve shows a consistently lower probability of survival for employees who retired at 55 (173 deaths) than for those who retired at 65 (462 deaths; fig 1). This difference was not, however, statistically significant (P = 0.09, log rank test). Likewise, the risk of death was similar between these two retirement groups (unadjusted hazard ratio 1.17, 95% confidence interval 0.98 to 1.41). For most of the follow-up period the survival curves were similar for employees who retired at 60 and those who retired at 65 (581 and 541 deaths). On the basis of a small number of deaths (n = 51), however, the probability of survival for the group who retired early at 60 was higher in the last eight years (after age 83; fig 2). The log rank test did not support a significant difference between the groups (P = 0.20) and this early retirement group was not at increased risk of death (hazard ratio 0.92, 0.82 to 1.04).
Table 1 Characteristics of employees of Shell Oil, United States, from 1 January 1973 to 31 December 2003. Values are numbers (percentages) unless stated otherwise
After adjustment for sex, calendar year of entry to the study, and socioeconomic status, employees who retired at 55 and were still alive at 65 had significantly higher mortality than those who retired at 65 (hazard ratio 1.37, 1.09 to 1.73; table 2). The risk of dying was about 80% greater in men than it was in women (1.83, 1.34 to 2.48) and retirees in the low socioeconomic category had a higher mortality than retirees in the high category (1.17, 1.01 to 1.36). Mortality was lower among more recent retirees (0.98, 0.97 to 1.00). Among employees who retired at 60, mortality was similar to those who retired at 65 (hazard ratio 1.06, 0.92 to 1.22). Men had a higher mortality than women (1.48, 1.16 to 1.89), and mortality was lower among more recent retirees (0.97, 0.96 to 0.99). Mortality did not, however, differ significantly by socioeconomic status (1.09, 0.97 to 1.23; table 2).
Table 2 Adjusted hazard ratios (95% confidence intervals) by explanatory variables for employees of Shell Oil, United States, who retired early at 55 or 60 compared with those who retired at 65
Regardless of socioeconomic status, employees who retired at 55 had a higher risk of mortality than those who retired at 60 (table 3). Employees who retired at 55 in the high socioeconomic group had a 20% increase in risk of death (hazard ratio 1.21, 0.88 to 1.67), whereas employees who retired at 55 in the low socioeconomic group had nearly a 60% increase in risk (1.58, 1.15 to 2.18). We found no difference in mortality by socioeconomic status in the group that retired at 60.
Table 3 Adjusted hazard ratios (95% confidence intervals) for early retirement by socioeconomic status
We could not assess directly the issue of whether employees who retired at a younger age were in poorer health than those who retired later as data were not available to identify the type of retirement for each employee (for example, retirement due to disability compared with normal retirement). If early retirement before 65 is taken by some workers because of failing health, however, mortality would be expected to be higher early in retirement. Table 4 shows the mortality for the first 10 years after retirement at 55 and for the first five years after retirement at 60 compared with employees who continued working at these ages. Regardless of socioeconomic status, employees who retired at 55 had almost a twofold higher mortality than those who continued working (1.89, 1.58 to 2.27). The mortality of employees who retired at 60 was similar to those who continued working at 60 (1.04, 0.82 to 1.31). Employees in the high socioeconomic group who retired at 60, however, had a 20% lower mortality (0.81, 0.54 to 1.20), whereas employees in the low socioeconomic group had a 20% increased mortality (1.21, 0.91 to 1.61).
Table 4 Adjusted hazard ratios (95% confidence intervals) for first 10 years after retirement at 55 and for first five years after retirement at 60
Discussion
Haynes SG, McMichael AJ, Tyroler HA. Survival after early and normal retirement. J Gerontol 1978;33: 269-78.
Ekerdt DJ, Baden R, Bosse R, Dibbs E. The effect of retirement on physical health. Am J Public Health 1983;73: 779-83.
Morris JK, Cook DG, Shaper AG. Loss of employment and mortality. BMJ 1994;308: 1135-9.
Trichopoulos D. Any questions? BMJ 1996;312: 632.
Padfield A. Myths in medicine. BMJ 1996;312: 1611.
Lin S. Optimum strategies for creativity and longevity. 2002. http://www.geocities.com/dtmcbride/health/retirement_age.html (accessed 16 Feb 2005).
Joyner RE, Pack PH. The Shell Oil Company's computerized health surveillance system. J Occup Med 1982;24: 812-4.
Cowper DC, Kubal JD, Maynard C, Hynes DM. A primer and comparative review of major US mortality databases. Ann Epidemiol 2002;12: 462-8.
Acquavella JF, Donaleski D, Hanis NM. An analysis of mortality follow-up through the national death index for a cohort of refinery and petrochemical workers. Am J Ind Med 1986;9: 181-7.
Tsai SP, Wendt JK, Cardarelli KM, Fraser AE. A mortality and morbidity study of refinery and petrochemical employees in Louisiana. Occup Environ Med 2003;60: 627-33.
Cox DR. Regression models and life tables (with discussion). J R Stat Soc 1972;34: 187-220.
Mein G, Martikainen P, Hemingway H, Stansfeld S, Marmot M. Is retirement good or bad for mental and physical health functioning? Whitehall II longitudinal study of civil servants. J Epidemiol Community Health 2003;57: 46-9.
Mein G, Martikainen P, Stansfeld SA, Brunner EJ, Fuhrer R, Marmot MG. Predictors of early retirement in British civil servants. Age Ageing 2000;29: 529-36.
Lantz PM, House JS, Lepkowski JM, Williams DR, Mero RP, Chen J. Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA 1998;279: 1703-8.
Chandola T. Social class differences in mortality using the new UK national statistics socio-economic classification. Soc Sci Med 2000;50: 641-9.
Mackenbach JP, Bos V, Andersen O, Cardano M, Costa G, Harding S, et al. Widening socioeconomic inequalities in mortality in six Western European countries. Int J Epidemiol 2003;32: 830-7.
Muntaner C, Hadden WC, Kravets N. Social class, race/ethnicity and all-cause mortality in the US: longitudinal results from the 1986-1994 national health interview survey. Eur J Epidemiol 2004;19: 777-84.(Shan P Tsai, manager, epidemiology1, Jud)