Results

DEMOGRAPHICS: BASELINE AND FOLLOW UP

At year 1 of the study, the cohort of 340 firefighters had a mean (SD, range) age of 40 (6.4, 28–58.) years. At year 3, the follow up cohort of 316 firefighters and the total population of those who were examined in year 3 (n=332) both had a mean (SD, range) age of 41 (6.9, 22–59) years. The population included the same four women (1%) at both time points.

Table 2 summarises the characteristics of the technicians and support controls in year 1. The mean (SD) age of the technicians was significantly greater (p<0.001) than the mean age of the support workers ((n=253) 41 (6.4) versus (n=70) 35 (6.9)). Although both groups were predominantly men, there was a higher proportion of female (4.2%) support controls than female technicians (0.4%, p<0.05). Higher proportions of technicians than support workers had increased blood pressures (systolic>140 or diastolic>90), reported taking antihypertensive medications; had a diagnosis of diabetes mellitus; were on lipid lowering agents, and had increased cholesterol measurements (cholesterol >6.22 mmol/l or >240 mg/dl). These differences were significant for cholesterol >6.22 mmol/l and for increased blood pressure and use of antihypertensives (both p<0.05).

χ² Tests showed no differences in the proportions of technicians (78%–81%) and support workers (20%–22%) examined at each of the three hospitals in year 1 (p=0.971). Likewise, these proportions of technicians (77%–84%) and support workers (16%–23%) did not differ significantly among the three hospitals in year 3 (p=0.852).

We found no difference at year 1 between the mean (SD) age of firefighters who remained on the teams (n=303) 39 (6.9) and the mean (SD) age of those who left the teams or were inactive in year 3 (n=20) 40 (8.2). Among the 24 subjects who either left the teams or were inactive in 1998 (year 3), the proportions of technicians and support workers were the same as the rest of the initial cohort: 79% and 21% (χ², p=1.000), respectively.

MEDICAL RECORDS REPOSITORY

We had no record requests from outside treating hospitals for assistance in the treatment of any team member. During the study period, no significant injuries or exposures due to work with hazardous materials were reported to the Commonwealth of Massachusetts Office of Hazardous Materials Response, which administers the teams.

LIVER FUNCTION TESTS

We found no significant differences between the liver function tests of technicians and support controls at year 1 (baseline) nor year 3 (follow up, table 3). We also found no differences in liver function tests at year 1 between subjects who remained active to the year 3 examination and those who did not remain active (table4).

In paired t tests, of year 1 versus year 3 liver function tests no significant changes were found for either hazmat technicians or support controls (table 5). We also made paired comparisons of liver function tests between year 1 and year 3 for the 12 support workers who subsequently became technicians (table 6). Again, we found no significant differences. Finally, we found no significant variation in mean liver function test values as a function of time for the team 1 subcohort over the period of 1993–8 (table7).

RENAL FUNCTION TESTS

We detected no significant differences between hazmat technicians and support controls when comparing either the mean blood urea nitrogen or creatinine values in independent sample ttests at years 1 and 3 (table 3). We also found no differences in either mean blood urea nitrogen or creatinine at year 1 between subjects who remained active to the year 3 examination and those who did not remain active (table 4).

For blood urea nitrogen, no significant differences were found in paired testing (tables 5 and 6). For technicians and for support workers who became technicians, the means for year 3 blood urea nitrogen showed a trend to be lower. Over the period 1993–8, we found no significant variation in mean blood urea nitrogen as a function of time for the team 1 subcohort (table 7).

We did find a significant decrease in mean creatinine concentrations from year 1 to year 3 for both technicians and support cohorts (table5), and for the 12 support workers who subsequently became technicians (table 6). Moreover, we found a significant longitudinal decrease in creatinine concentrations for the team 1 subgroup from 1993 to 1998 (table 7). The 1998 mean creatinine concentration was lower than the 1993 concentration (114(8.8) v 97 (8.8), p< 0.001).

Because no differences were found across exposures, subsequent analyses were done for each examining hospital. For two of the hospitals (including one that had changed laboratories) the year 3 mean creatinine was significantly lower (both p<0.001), but for the third hospital the year 3 mean creatinine was non-significantly higher (p=0.177) (data not shown).

HAEMATOLOGICAL TESTS

For mean white blood count and packed cell volume, we found no significant differences between groups (tables 3 and 4) nor within groups over time (tables 5-7).

Independent t tests between the mean platelet count of technicians and support controls at year 1, showed a trend towards a lower mean platelet count for the technicians in both years (table 3). There was no significant difference in mean platelet count between subjects who remained active to the year 3 examination and those who did not remain active (table 4).

Paired t tests of platelet count at years 1 and 3 for both technician and support worker subgroups showed significant changes for technicians (n=247): 240 (52) to 232 (54) from baseline to follow up examination (p< 0.001; table 5), whereas within support workers the decrease was non-significant. We also found a significant decrease in mean platelet count for the support workers who subsequently became technicians (table 6).

Also, we found significant longitudinal variation in mean platelet counts among the team 1 subgroup from 1993 to 1998 (table 7) as a function of time. The 1998 mean platelet count decreased significantly from the 1993 value of 237 (42) to 210 (40) (n=35; p< 0.01).

Because hospital 2 had changed reference laboratories in year 3, we performed further paired analyses stratified by examination site (appendix 2). For the other two hospitals, we found no differences over time in mean platelet count for either technicians or support workers. For hospital 2, the new reference laboratory had a lower reference interval, and in year 3 both technicians (n=82) and support (n=14) workers had lower mean platelet counts (p<0.001 and p<0.005, respectively).

Of the 12 support workers who became technicians by 1998, 10 (83%) were examined at hospital 2. Thus, the change in laboratories would explain the relative decrease in mean platelet count from 1996 to 1998. Finally, a second ANOVA as a function of time with only the means from the original laboratory (1993–7), showed no longitudinal variation in mean platelet count for the team 1 subcohort (p=0.110).

PROPORTION OF TECHNICIANS AND CONTROLS WITH VALUES OUTSIDE THE EXPECTED RANGE FOR THE COHORT

Comparison of mean values between and within groups did not show any adverse effects associated with work with hazardous materials. Therefore, we also explored the hypothesis that a few hazardous materials technicians might develop significant changes in laboratory variables associated with exposure—that is, depression of haematological test values, or increases in renal or hepatic test values. Thus, we compared the proportion of hazardous materials technicians and support workers with values outside the expected range for the entire cohort. In these analyses, we identified the number of firefighters in both groups with haematological test values of more than two SDs below the mean (technicians and support workers combined), or with renal or hepatic test values of more than two SDs above the overall mean. As each hospital used a different laboratory, we used the mean (SD) for each test at each facility separately to identify the number of firefighters outside the expected ranges. Then, these numbers were summed for technicians and support controls for year 1 and year 3, respectively.

No differences in the proportions of technicians or support controls with test values outside the expected ranges were found with the exception that significantly more controls (7%, (5/71)) had increased creatinine values in year 1 than technicians (1%, (3/268), p<0.012). A detailed breakdown of technicians and support controls with values outside the expected ranges for each hospital and year is given in appendix 3.

Discussion

This study was a prospective and cross sectional evaluation of possible health effects of firefighting work with hazardous materials. No clinical health effects of work with hazardous materials were reported during the study period. This is consistent with our previous studies of incidents with hazardous materials responded to by the same six Massachusetts regional teams from 1990 to 1996.^{15 16}In analyses of reports of incidents for the first 6 years of work with hazardous materials, hazardous materials team members experienced no notable chemical exposures. Minor musculoskeletal injuries were reported in a single incident. This is also consistent with findings among hazardous waste workers that few notable exposures occur because when the potential for exposure is greatest the workers are usually equipped with the highest levels of personal protective equipment.⁷

To study the possible subclinical health effects of hazardous materials firefighting, we studied liver and renal function tests and haematological indices as markers of end organ effect. Such testing has been recommended as standard or optional components in the medical examinations of hazardous materials workers^{1 8 9} and firefighters.^10–12

We found no significant differences in means for any of the effect markers studied across exposures (hazardous materials technicians versus support controls) at both year 1 and year 3. Also, no differences in the proportions of technicians or support controls with test values outside the expected ranges were found with the exception of creatinine in year 1. In this case, the result was not consistent with an adverse effect of work with hazardous materials, as more controls had increased creatinine values than technicians. Also, we found no differences in any of the variables studied between members of the teams who either left or became inactive and those who remained active to the 3rd year of the study.

Likewise, we found no clinically notable changes within the means for various subject groups over time. We found significant longitudinal changes for only two indices: creatinine and platelets. For all groups studied at two of the hospitals, creatinine decreased significantly over time, at the third hospital, mean creatinine increased non-significantly. Because mean creatinine decreased (improved) in support workers as well as technicians, and we found no differences between groups at either time point, we think that these changes are not related to exposure. The change in reference laboratory in 1998 at one of the hospitals may account for some of the variation.

In initial longitudinal analyses, we did find consistent and significant decreases in platelet values within technician groups and support workers who became technicians. In further analyses, however, the use by one of the hospitals of a new laboratory in year 3 (1998) with a lower reference interval for platelet counts seemed to be the cause. When we stratified the data by hospital, the decrease in platelets was isolated to those tested at a different laboratory in 1998, and was also found in support workers examined at the same site. Also, when platelet data for the team 1 subcohort were examined for 1992–3 to 1997 (excluding 1998 when a different laboratory was used) there was no significant change in platelets over time.

Our results are consistent with those of the study of hazardous waste workers by Favata and Gochfeld ,¹³ which failed to find any differences in the same indices that we studied between workers with low and high potentials for exposure. The results are also consistent with our previous study of the team 1 cohort¹⁴where we found few abnormalities in these indices and none that we could link to specific exposures. Examination of the team 1 data over 5 years showed remarkable stability. This suggests that the clinically unimportant, but significant changes in some of these indices that we found from 1992–3 to 1995¹⁴ were probably due to chance, subtle changes in specimen collection, preparation, or laboratory analyses between the two time points. As expected, over many observation points, such factors not related to exposure are unlikely to affect results in a consistent way.

The major limitation of the current study was imposed by logistical constraints. In a statewide programme, all six teams could not be examined at a single hospital. Further, the investigators could not require the examining hospitals to send all specimens to a single reference laboratory for all teams and all time points. Indeed, this study highlights the potential problems that ongoing surveillance programmes may have when hospital contractual obligations mandate changes in reference laboratories. This limitation was countered by the major strength of our study, the overall study design.

Firstly, we had an excellent control group, the support workers. Both technicians and support workers perform regular fire duty with their non-state, local fire departments. Support members are presumed to have a very limited potential for exposure compared with technicians, as they do not enter the hot or contaminated zone of an accident. Therefore, they are an ideal control group for the investigation of potential health effects limited to those arising from duty within the contaminated zone of hazmat accidents. We found some differences between technicians and control firefighters on several confounding variables, but these differences would be expected to bias the study towards finding worse results for the technicians. Despite the fact that the technicians were older, more likely to have increased blood pressure, increased cholesterol, take antihypertensive or lipid lowering medications, or have diabetes mellitus, we found no significant differences in means for any of the effect markers studied across exposures.

Secondly, the proportions of control workers examined at each of the three hospitals were similar. There were no significant differences in the proportions of technicians to controls among the three hospitals at either year 1 or year 3. This fact justified pooling the data from all three hospitals for most analyses.

Thirdly, our design allowed for multiple methods of looking for possible exposure or work related effects. These included independent, cross sectional comparisons of technicians and controls at two separate time points, comparisons of those who became inactive or left the teams with those who remained active, and prospective follow up comparisons within subjects over time. These prospective comparisons included technicians, support workers who became technicians, and a 5 year time analysis for the team 1 cohort. Although the support workers who became technicians and the team 1 cohort were small, their results support the consistency of the lack of any effect related to exposure or work. Finally, we explored the hypothesis that a few technicians might be adversely affected by examining the proportion of firefighters with results outside the expected range for the cohort as a whole. Again, we found no evidence for an adverse effect of work with hazardous materials on the test values we studied. Given the study's design, it seems unlikely that we would have missed a significant effect related to exposure on the end organ effect markers studied.

Work with hazardous materials has the potential for serious exposures and injuries due to explosions, fires, and other releases of dangerous substances. Among hazardous waste workers it has been found, however, that few significant exposures occur because those with the greatest potential for exposure are usually equipped with the highest levels of personal protective equipment.⁷ This study and previous studies of surveillance^{13 14} and of a hazardous materials accident database^{15 16} also suggest that there are seldom health effects among hazardous materials technicians who wear appropriate personal protective equipment. Therefore, current protective equipment and procedures including decontamination seem to be effective.

Another important consideration in this discussion involves the most likely potential exposures during work with hazardous materials. Irritants and corrosives are the most commonly released hazardous materials, and respiratory exposures to irritants are the most commonly reported exposures.^16–18 Most irritant and corrosive substances would not affect haematological indices, or liver or renal function. On the other hand, common health problems and other non-occupational factors may produce abnormalities. Therefore, the hepatic, renal, and haematological tests used in our investigation cannot be expected to be either sensitive or specific markers of exposure to hazardous materials. In our experience, a further disadvantage of these markers is that firefighters under medical surveillance often misunderstand the limitations of these tests. They tend to overvalue normal results on these variables as ruling out the possibility of present and future health effects related to exposure.

Our results suggest that routine testing of the hepatic, renal, and haematological indices used in our investigation is not required on an annual basis, and that the use of these tests in detecting subclinical health effects is limited. Our current recommendations for haematological, hepatic, and renal indices in the medical surveillance of hazardous materials firefighters and other firefighters will be baseline measurements for comparison after notable exposures, illness, or other changes in clinical state. Because certain laboratory values may vary as a function of age, it seems desirable to retest at some interval. The results of the subcohort followed up over 5 years suggest that in the absence of a known exposure or other clinical indication, it is unnecessary to reassess the indices studied here more often than every 3–5 years. When comparing periodic testing with baseline results, there should also be consideration of variation not related to exposure or illness, but due to changes in laboratories, reference intervals, or testing methods.