Mortality and life expectancy of professional fire fighters in Hamburg, Germany: a cohort study 1950 – 2000

Background The healthy worker effect may hide adverse health effects in hazardous jobs, especially those where physical fitness is required. Fire fighters may serve as a good example because they sometimes are severely exposed to hazardous substances while on the other hand their physical fitness and their strong health surveillance by far exceeds that of comparable persons from the general population. Methods To study this effect a historic cohort study was conducted to assess mortality and life expectancy of professional fire fighters of the City of Hamburg, Germany. Fire departments and trade unions questioned the validity of existing studies from outside Germany because of specific differences in the professional career. No mortality study had been conducted so far in Germany and only few in Europe. Information on all active and retired fire fighters was extracted from personnel records. To assure completeness of data the cohort was restricted to all fire fighters being active on January 1, 1950 or later. Follow up of the cohort ended on June 30th 2000. Vital status was assessed by personnel records, pension fund records and the German residence registries. Mortality of fire fighters was compared to mortality of the Hamburg and German male population by means of standardized mortality ratios. Life expectancy was calculated using life table analysis. Multivariate proportional hazard models were used to assess the effect of seniority, time from first employment, and other occupational characteristics on mortality. Results The cohort consists of 4640 fire fighters accumulating 111796 person years. Vital status could be determined for 98.2% of the cohort. By the end of follow up 1052 person were deceased. Standardized Mortality Ratio (SMR) for the total cohort was 0.79 (95% CI, 0.74–0.84) compared to Hamburg reference data and 0.78 (95% CI, 0.74–0.83) compared to National German reference data. Conditional life expectancy of a 30 year old fire fighter was 45.3 years as compared to 42.9 year of a German male in normal population. Job tasks, rank status and early retirement negatively influenced mortality. For fire fighters with comparably short duration of employment the mortality advantage diminished with longer time since first employment. SMR of persons who retired early was 1.25 (95% CI, 1.13–1.60) in reference to the general German population and the SMR of 1.71 (1.18–2.50) in the multivariate regression model. Conclusion A strong healthy worker effect was observed for the cohort, which diminished with longer time since first employment for fire fighters with shorter duration of employment, as expected. The negative effects on mortality of job tasks, rank status and in particular early retirement indicate the presence of undetermined and specific risks related to occupational hazards of fire fighters.

tasks, rank status and in particular early retirement indicate the presence of undetermined and specific risks related to occupational hazards of fire fighters.

Background
At the end of the 1990es a raise of retirement age for police and fire fighters was discussed in the German parliament in the context of reforms of social welfare system, pension fund financing. The retirement age of 60 years was supposed to be raised to 62 years. Trade unions and fire departments opposed these plans arguing that life expectancy of fire fighters was already dramatically shorter compared to general population. Arbitrarily collected and published data from fire departments indicated a low average age at death [1]. Later inquiries showed that these self-collected data were wrong.
All previously published mortality data from other countries (Table 1) indicated a lower mortality with standardized mortality ratios between 0.76 and 0.99. However, in the light of the self-collected data in Germany these studies were considered unreliable and "not applicable" by both, the fire departments and the trade unions.
In July 2000 the head of fire department of the Free and Hanseatic City of Hamburg requested the Occupational Medical Service to conduct a scientific study to answer to this question. The validity of existing studies from outside Germany was questioned by fire departments because of presumed specific differences in the professional career in Germany like frequent life long employment, work till later age, remaining a fire fighters longer because of specific German pension benefits. No mortality study had been conducted so far in Germany and very few in Europe. In collaboration with the Institute of Mathematics and Computational Sciences in Medicine of the School of Medicine at University of Hamburg a retrospective cohort study was initiated in autumn 2000.
This study reports the findings of a retrospective cohort study on the mortality of fire fighters of the Fire Department of the City of Hamburg, Germany. The objectives were to establish the life expectancy and standardized mortality ratios compared to the Hamburg and the national reference population. We put special emphasis on disentangling the suspected strong healthy worker effect in this cohort from the effects of potential chemical exposures and heavy work load.

Methods
All male fire fighters of the Fire Department of the City of Hamburg, Germany being active between January 1, 1950 and June 30, 2000 (end of follow-up) were included in the study population. They were all full-time, professional employees. The number of female fire fighters was negligible and thus they were excluded. We included only active fire fighters and no administrative personnel of the Fire Department.
Date of birth, date of death, date of employment, date of end of active duty, reason of end of active duty, duration of employment according to the type of tasks (fire fighting & rescue service vs. administration) and rank were obtained for pensioners and active fire fighters from the Hamburg Fire Department. This included e.g. card files of pensioners and the archive of personnel records. c) High exposures (for example, from carcinogenic chemicals such as diesel fumes) occur during maintenance and routine work in the station. d) Even when major fires occur it is unclear who in the brigade has actually the highest exposure. The attack team have also the highest level of personal protection (SCBA); two attack teams (on ground and on ladder) might have completely different exposure levels; the commander who tries to get a good glimpse on the situation but doesn't use his mask in order to be able to communicate better via walkie-talkie might even have the higher exposure to burn products and fumes. It is known that the highest exposures are often during overhaul and destruction, so the position of the fire fighter during the actual fire is not predictive of his/her exposure during the entire course of the sortie.
e) Even if a position or task could have been documented the use of SCBA is unknown.

Standardized Mortality Ratios
Overall SMR using the age specific death rates of Hamburg and Germany were notably reduced (Table 4). SMR for the cohort was 0.79 with a 95% CI of 0.74-0.84 (Reference Hamburg) and 0.78 with a 95% CI of 0.74-0.83 (Reference Germany)). Because the results of the SMR computation were similar for both reference populations the subgroup analysis given below only refers to the German population.
Stratification by year of entry shows a reduction in mortality over time. More detailed analysis on SMR in sub-groups is given in Dividing the cohort into 'rank groups' according to the German state employment categories (in German: "mittlerer, gehobener und höherer Dienst") as indicators of socio-economic status showed higher mortality (SMR, 0.82; 95% CI, 0.77-0.87) for the 'middle rank' than for fire fighters in the 'high' or 'higher rank' group combined (SMR, 0.51; 95% CI, 0.41-0.63) with 4309 available data sets ( Table 5). The 'middle rank' is the lowest income group a fire fighter can join in the last decades. The terminology has historical reasons: the "low rank" category was abandoned in the middle of last century. Depending on prior qualification fire fighters can join the higher rank groups directly. For 248 persons reliable information on rank was not available.
Regarding the 'reason of retirement' persons with a regular retirement showed a reduction in mortality (SMR, 0.79; 95% CI, 0.73-0.85) compared to the German reference population (Table 5). Fire fighters who left the department because of other reasons showed an increase in mortality: early pensioners (SMR, 1.35; 95% CI, 1.13-1.60) left the department for health reasons and other fire fighters who left the fire department on their own request (SMR, 1.12; 95% CI, 0.77-1.57) for instance to move to a different city. Table 6 shows the result of the SMR calculations for duration of and time since first employment simultaneously. The SMR in the first five years after joining the fire fighters was only 0.3 (0.12-0.61). This decreased mortality seemed to disappear for the subgroups of fire fighters with a total duration of employment of less than 10 years, however, numbers in these subgroups were quite small. For the subgroups with longer duration of employment the mortality appeared to increase with longer period of follow up, but after 30 years of observation the SMR continued to be significantly below 1.

Multivariate subgroup analysis with proportional hazard models
To include time dependent covariates a subgroup was formed excluding all workers already active on the 1/1/ 1950. Table 7 presents the results of the proportional hazard model (N = 3576) using following time dependent covariates: duration of employment, year of employment, age at employment, rank group, reason for leaving simultaneously. Adding the variable 'type of task' into the model did not improve it. It was hence not included.  Table 7).

Duration of employment
Year of entry A later date of entry is associated with a lower mortality risk. In comparison to the reference group 'entry between 1950-1954' the mortality risk for the group 'entry after 1970' is reduced to 46 % (RR, 0.46; 95% CI, 0.14-1.57).

Age at entry
The age of entry had been categorised in 3 groups. In comparison to the reference group 'age 15-25 years' the mortality of the group 25-30 years reveals no difference. An increased relative risk has been observed for the workers with an age at entry higher than 30 years (RR, 2.38; 95% CI, 1. 36-4.15).

Rank group
A lower risk was found for the group of the high and highest rank group combined in comparison to the middle rank group (RR, 0.39; 95% CI, 0.24-0.62).

Early retirement
Workers who leave the fire department due to early retirement have an increased relative risk (RR, 1.71; 95% CI, 1.18-2.50) in comparison to all others.

Discussion
This paper reports on the first cohort study undertaken in Germany to examine the mortality of fire fighters with its typical, quasi lifelong employment of professional fire fighters. We are confident that the cohort of fire fighters is complete for the time period after 1950. It constitutes the largest cohort study of fire fighters' mortality in a European country.
The study was limited by the fact that we could not collect a detailed exposure history of fire fighters and data on causes of death.
Taking runs as a proxy parameter for exposure is not established as standard in the research on fire fighter risks. The differences in job tasks, wind direction or protective equipment on jobs at the same fire do not allow taking number of runs as proxy. The exposure matrix of fire fighters is highly complex as time at fire does not indicate that for example protective equipment was worn or not [28].
Neither the evaluation of causes of deaths nor the inclusion a control group such as officers from the Hamburg Police Department was funded by the Fire Department. We tried to account for this limitation by using the Cox

Overall mortality
Our results show that the mortality of the fire fighter cohort of Hamburg is about 20% lower than the mortality of the reference population. They confirm findings from most of the other published studies from different countries (Table 1). However, our findings are in the lower range of previously published studies. It might indicate that the selection processes and the intensive medical surveillance programs in Germany have a greater effect than the programs in other countries with lower standards (see discussion below).
The lower mortality of fire fighters in comparison to the general population is probably influenced by the healthy worker effect in several aspects. The question how much the healthy worker effect masks a potentially negative effect of occupation on mortality arises in all occupational cohort mortality studies [4,5]. The reference to 'general population' is convenient and -as it was the case hereoften the only financially feasible way. Unfortunately it is not the best comparison group to determine the occupation-induced mortality risks because of the selection of cohort members based on health status and risk factors at the beginning of work.
Using definitions according to Choi [6], several components of the healthy worker effect (HWE), e.g. the healthy hired, low-risk hired, worker healthier and the healthy worker survivor effect probably led the observed low mortality. In general the magnitude of the healthy worker effect is estimated to be around 20% advantage in mortality (see [4] for further discussion). Our result (SMR of 0.78) is very similar to that.
We observed a decline of the healthy worker effect with increasing time since first employment. This effect was more pronounced for the subgroups with a total duration of employment of less than ten years. This observation is consistent with the assumption that the "healthy hired" component disappears within this time frame from date of first employment. For fire fighters with longer duration of employment (10-29 years) the mortality advantage is also declining, but the SMR does not increase to 1. This may probably reflect a levelling off of the "low-risk-hiring" component. Finally, the SMR for the subgroup with working time durations of more than 30 years was 0.55 (95% CI, 0.43-0.69), i.e. lower than those for the other subgroups with shorter duration but the same time since first employment. This indicates a pronounced long term effect of the "work healthier" and the "healthy survivor" component of the healthy worker effect.
Physical and medical fitness for professional and voluntary fire fighters is required nationwide in Germany using common standards. This constitutes a major difference to the US American system where regular physical performance tests are suggested by National Fire Protection Association (NFPA)/USA [7] but not regularly required on national level [8]. However, on state and/or community level in the USA regular physical performance tests are sometimes mandatory.
The stringent selection process in Germany demands physical and psychological health and fitness. After joining there are regular and intensive medical examinations: until the age of 50 every 3 years, beyond 50 every year. These tests include stress-ECG to evaluate standardized physical fitness and a fitness test with heavy respiratory protection gear on the obstacle course. In the stress-ECG fire fighters below 30 years of age have to perform at an energy level of three Watt per kilogram of bodyweight. Fire fighters above 30 years of age have to show the same level of performance reduced by one percent per year of age above 30 years. The capacity to perform at an energy level of 200 Watt minimum has to be proven at all occasions. In addition to the fitness test, other medical criteria for vision test, audiometry, lung function test and acceptable blood pressure response and heart rates at stress-ECG have to be met. The rate of onduty deaths caused by coronary heart disease is reportedly higher than in other comparable occupational groups such as police or emergency services [8]. Other specific causes of deaths with higher than normal mortality in fire fighters are reported such as certain kinds of brain or colon cancer, leukemia, kidney and urethra cancer, prostate and bladder cancer [9][10][11][12].

Mortality of early retirees
Our study confirms findings that showed an elevated SMR for persons who retire early. We observed a SMR of 1.35 (95% CI, 1.13-1.60) in reference to the general German population and a SMR of 1.71 (95% CI, 1.18-2.50) in the Cox-Regression in reference to all others. We did not observe any beneficial effect of early retirement as documented in a Danish study [13] and the Whitehall II study on effects of normal retirement. [14].
In a Danish population-based study the disability benefit recipients showed were markedly elevated mortality [15]. Retirement in itself seems to be a risk factor for early death. In a British study men who were unemployed had a RR of 2.13 (95% CI, 1.71-2.65, men who retired early for reasons other than illness had still a significantly higher mortality compared with employed men (RR 1.87, 95% CI, 1.35-2.60) [16].
A study of past employees of Shell Oil, USA, showed a significantly higher mortality of employees who retired early at 55 and who were still alive at 65 (n = 839) had a significantly higher mortality than those who retired at 65 (n = 900) (hazard ratio 1.37; 95% CI, 1.09-1.73). Mortality was significantly higher for subjects in the first ten years after retirement at 55 compared with those who continued working (1.89; 95% CI, 1.58-2.27). The significant difference, however, showed only after adjusting to sex, calendar year of entry to the study, and socioeconomic status. Retired employees in the low socioeconomic category had a higher mortality than retirees in the high category (1.17, 95% CI, 1.01-1.36) [17].
Results from the British Regional Heart Study indicated that men who retired early for reasons other than illness had a significantly increased risk of mortality compared with men who remained continuously employed (relative risk 1.87 (95% CI, 1.35-2.60)). [18] Early retirement was associated with higher mortality in a construction workers cohort in Germany (RR, 1.50; 95% CI, 1.20-1.88) [19].

Reasons for retirement, reasons for non-fitness and possible risk factors for higher mortality of retirees
Our study was limited by the fact that we could not include common risk factors for elevated mortality such as cardiovascular risk factors (e.g. blood pressure, lipid profile [20]), psychosocial risk factors (e.g. stress, life event impact, traumatic experiences, depressive disorders [21][22][23]) or exposure factors to toxic gases (e.g. carbon monoxide). The social medicine department which handles the retirement cases of the City of Hamburg does neither publish nor hand out detailed statistics on reasons for retirement of public employees despite multiple requests.
However, we can presume that 'reasons for early retirement' were almost identical to 'reasons for restriction of fitness' because of medical problems during active duty. From studies of the occupational health service of the Fire Department we know that the reasons for restricted fitness because of medical problems, both temporary or permanent, were cardio-vascular diseases in 39% of all cases (N = 230) and 44% in fire fighters over 50 years (N = 132), musculoskeletal diseases (25% and. 21%), respiratory disorders (5% and 6%), injuries & surgeries (9% and 5%), metabolic disorders (3% and 5%) and psychiatric disorders including addiction and abuse (6% and 5%) [24]. High blood pressure accounted only for 5% of the medical fitness restrictions but was prevalent in 20% and 23% of the unfit fire fighters.
This pattern of diseases in events of non-fitness is distinctly different from reasons for retirement because of ill health in an analysis of retirees from the National Health Service, United Kingdom, which listed musculoskeletal (49%), psychiatric (20%), and cardiovascular conditions (11%) as most common reasons [25]. The rate of high blood pressure in Hamburg fire fighters is consistent with findings from other studies which reported a prevalence of high blood pressure between 20% and 23%, the majority of the men were untreated [26].

Unspecified risk factors
Despite the pronounced healthy worker effect, our study yields several results of subgroup SMR and Cox-regression analyses which support the assumption that occupational hazards in fire fighting, which are not specified here such as stress, raised the mortality.
First, the SMR of fire fighters who worked more than 50% of their time in administrative units 0.53 (95% CI, 0.35-0.78) is lower than the mortality of persons who worked more than 50% in active fire fighting with 0.79 (95% CI, 0.74-0.84). Second, we observed a striking difference in mortality between rank groups. This confirms results of studies which show a reduced mortality in higher socioeconomic groups [27]. However, this difference could also reflect different tasks and job exposure profiles. Higher ranks are usually not part of the attack or rescue teams. Third, causes for early retirement are partly diseases often caused or triggered by the job. The elevated SMR for those persons may reflect individual susceptibility in combination with or reaction to special hazards from the job.
We are unable to forward any explanation for the rise of SMR in the group 5 to 10 years of duration of employment. As causes and circumstances of death could not be included in this study, a discussion of reasons for this finding was considered speculative by the authors.
The decreasing mortality with later date of entry is indicative of major changes in the work environment during the study period. Tactics and safety equipment for fire fighters were improved. Accident rates fell due to better techniques and safety equipment. Especially the widespread introduction of heavy respiratory protection equipment in the 80'ties lowered the exposure to fumes and gases drastically. In Hamburg respiratory protective gear was already used end of the 70'ties. [28] On the other hand, during the study period changes have occurred which may have influenced mortality negatively e.g. plastics were introduced en mass into the household environment and hence became part of structural fires resulting in toxic and carcinogenic burn products.
The multivariate analysis showed a considerably higher relative risk for fire fighters who joined the department after 30 years of age (SMR, 2.38). For this result we are also unable to provide any reasonable explanation as it is in contrast to findings of other studies [29]. In the absence of other explanations and due to the low numbers it could be a chance finding.

Conclusion
In summary, we could confirm in our German study the results of previous studies of a lower-than-normal mortality of fire fighters. As expected, a strong healthy worker effect with all its components was observed. The findings also suggest that the intensive medical surveillance is beneficial to the overall health and mortality of professional fire fighters.
However, results indicate negative effects of type of task, rank status and early retirement on mortality. It appears that fire fighting and rescue services have a distinct negative influence on mortality.