3.0 Results

3.1 Demographics

Table 1 shows demographic data for the three groups (CC, CCT, and Controls).  Although not statistically significant, CC tended to be:  younger, with fewer years of service (as of 5 October 2004), were more likely to be junior NCOs, were more likely to be currently smoking, and had a lower BMI rate of obesity.

With respect to postings, CC had served for significantly more time on submarines than Controls (p=<0.01), and these postings made up a significantly greater proportion of their total posting time (compared to Controls and CCT; p=0.002 and p=0.04, respectively).

Notes: s.d. = standard deviation. Fire Date = 5 October 2004
* Average posting time expressed in years and as a percentage of total posting time, including all subjects.
** CC vs. Controls, p< 0.01

Table 2 shows that as of 31 December 2009, over one third (35%) of the CC had been released from the CAF, which was similar to the number released in the Controls (34%).  A significantly larger number of the CC released medically, 12.5% (95% CI:  5.2, 24.1), as compared with only 5.3% (95% CI:  2.3, 10.1) of the Controls (p=0.04).  The percentage of the CCT who released medically, 4.8% (95% CI:  0.6,16.1) was not statistically significantly different than the Controls (p=0.90).

*p=0.04 CC vs Controls

3.2 Sick Leave and Medical Employment Limitations

Table 3a summarizes the frequency of SL or MELs (all types included versus UFS or UFA only), all diagnostic categories combined, by group, and time period, with baseline referring to the pre-fire period.  Roughly 45% of members in all three subject groups had no documented SL over the study periods.  Post-fire, the CC had 37.5% of its members recorded as having new SL, as compared to the 18.4% for Controls; 28.6% of the CCT had new SL during this time.  The number of subjects in each group with new post-fire MEL’s, all types included, was the same as that of SL, except the CCT who had three fewer members affected post-fire.  Post-fire, 50% of the CC had new MELs UFS or UFA only, as compared to 17.8% of the Controls; the CCT had 16.7% of their members in this category.

In Table 3b, composite baseline is defined as a member who had a condition in the baseline period, plus those who had the conditions in both the baseline and post-fire periods. Composite post-fire was defined as those who had a condition post-fire, plus those who had the condition of interest both in the baseline and post-fire periods. The number of Controls with composite post-fire SL (34.2%) was very similar to the number of them who had composite baseline SL (35.5%). The number of CCT subjects with composite SL in the post-fire period (33.4%) was slightly greater than the number of them who had composite baseline SL (26.2%). In contrast, 50.0% of CC subjects had composite post-fire SL (50.0%) as compared to composite baseline SL (17.9%) and the 32.2% difference was significant (CI: 15.7, 48.7, p< 0.001). Similar to SL, the CC had 30.4% more composite post-fire MELs, all types included, as compared to 2.0% for the Controls; the CCT had 4.7% more. Finally, the CC had 48.2% more composite post-fire MELs UFS or UFA only, as compared to 10% for the Controls; the CCT had 4.8% more members who had post-fire MELs UFS or UFA only.

*Composite Baseline = condition baseline, plus condition baseline and post-fire
** Composite Post-fire = condition post-fire, plus condition baseline and post-fire

SL and MELs for all diagnostic categories were combined in Figure 2 (all MEL types) and Figure 3 (UFS or UFS/UFA). The proportion of subjects with either SL or MELs was roughly similar in the baseline and post-fire time periods when looking at the Controls or CCT subjects. However, for CC, there is a marked difference between the proportions of subjects with SL or MELs in the post-fire time period only (black bars) as compared to the baseline time period only (white bars).

Figure 4 illustrates SL and MELs (all types) by diagnostic categories.  In the baseline and post-fire time period musculoskeletal was the most common diagnostic category for the CCT.  For Controls, there were more subjects with psychiatric MELs in the post-fire time period compared to the baseline time period (14 subjects versus three subjects).  Among CC, there were substantial differences in baseline and post-fire SL results with the incidence of gastrointestinal, musculoskeletal, neurological, psychiatric, urological, and “other” diagnostic categories being higher in the latter period.  For subjects with MELs, the most striking difference between the baseline and post-fire time periods for CC is for the psychiatric diagnostic category:  no subjects in baseline versus 30 subjects post-fire.  Musculoskeletal was the most common diagnostic category across all subject groups and time periods with the notable exception of the CC in the post-fire time period, where the most common diagnostic category was psychiatric.  Among the CCT, there was an increase in SL for gastrointestinal diagnoses post-fire that accounted for 36% of sick days versus only 10.4% for Controls.

To simplify Figure 4, one Control subject with MELs in the post-fire period due to malignancy was not included.

Figure 4 a):

Figure 4 b):

Figure 4 c):

Figure 4:  Number of Subjects with sick leave and medical employment limitations for a) CC, b) Controls, and c) CCT.  White bars represent subjects with SL or MELs in the baseline time period and black bars represent subjects with SL or MELs in the post-fire time period.  Individual subjects may have had SL or MELs for more than one type of diagnostic category in a given time period.

Table 4 shows SL and MELs as a percentage of total lost days in service.  For cumulative SL in days (among all subjects in the group) for baseline and post-fire periods, the Controls had 908 days and 912 days, respectively, and the CCT had 146 days and 200 days, respectively.  In contrast, the CC post-fire SL exceeded baseline SL by a factor of approximately ten (1953 days versus 175 days).  Excluding the CC post-fire time period, the diagnostic category with the highest proportion of total SL days was musculoskeletal (42.7%).  For the CC and during the post-fire period, the psychiatric diagnostic category accounted for 64.6% of total SL days.

Table 4: Total duration of sick leave and medical employment limitations by diagnostic categories, number of lost days and percentage of total lost days*

* The “n” values correspond to the number of subjects with the outcome of interest (SL or MELs) for each particular subject group and time period.

All groups had more MELs in the post-fire time period. This effect was most striking for the CC, with 25,853 MEL days post-fire compared to 1851 during baseline. The CC accounted for 29% of all subjects with post-fire MELs, and approximately 50% of post-fire MEL days. For the CC, more than 57% of CC post-fire MEL days were in the psychiatric diagnostic category. Other categories with increases in MEL’s post-fire in the CC were cardiovascular, musculoskeletal, neurological, respiratory, and urological diagnostic categories.

SL and MEL (all types) days combined, expressed as a percentage of total service time in a given time period, are shown in Table 5a.

Table 5:  Incidence of days lost to sick leave and medical employment limitations, per 100 service days (95% CI)

During the five years preceding the fire, the CC spent 2.0% of all their service days on either SL or any type of MEL.  Over the same time period, the Controls and CCT respectively spent 2.8% and 3.9% of their service days on either SL or any type of MEL (Table 5a).  The incidence rate of days lost to SL or any type of MEL in the five years preceding the fire was 38% greater in Controls (IRR:  1.38; 95% CI:  1.32, 1.45) and 97% greater in the CCT (IRR:  1.97; 95% CI:  1.86, 2.08) compared to the CC.  It should also be noted that the incidence rate of days lost to SL or any type of MEL in the five years preceding the fire was 42% greater in the CCT than in Controls (IRR:  1.42; 95% CI:  1.36, 1.48).  All three of these differences were statistically significant at the 95% confidence level.

During the five years following the fire, the CC spent 30.4% of all their service days on either SL or any type of MEL.  Over the same post-fire time period, the Controls and the CCT respectively spent 8.5% and 9.1% of their service days on either SL or any type of MEL (Table 5a).  The post-fire incidence rate of days lost to SL or any type of MEL was 258% greater in the CC than in Controls (IRR:  3.58; 95% CI:  3.52, 3.65).  It should also be noted that the post-fire incidence rate of days lost to SL or any type of MEL was 7% greater in the CCT than in the Controls (IRR:  1.07; 95% CI:  1.04, 1.11).  These two differences were statistically significant at the 95% confidence level.

Post-fire the primary diagnostic category for the CC was psychiatric with 15.8% of service days on either SL or a MEL (all types).  Post-fire the CC had increases as compared to baseline in the percent of service days spent on SL and MEL (all types) in other diagnostic categories:  cardiovascular (from 0.01 to 1.0), musculoskeletal (from 1.1 to 3.4), neurological (from 0.14 to 2.0), respiratory (from 0.19 to 4.0), and urological (from 0.01 to 0.46).  Compared to the Controls, the post-fire percent of service days affected by SL or MEL (all types) for the CC were substantially greater for neurological (2.0 vs. 0.22), psychological (15.8 vs. 1.1), respiratory (4.0 vs. 1.4), and urological (0.46 vs. 0.08) categories.

Analyses were repeated using only MELs that specified “unfit sea” or “unfit sub” (UFS), or “unfit alongside” (UFS/UFA).  These types of MELs are more significant than other types of MELs, such as activity restriction and are presented in Table 5b. 

During the five years preceding the fire, the CC spent 0.78% of all their service days on either SL or UFS/UFA MELs.  Over the same time period, the Controls and CCT respectively spent 1.4% and 2.2% of their service days on either SL or UFS/UFA MEL (Table 5b).  The incidence rate of days lost to SL of UFS/UFA MEL in the five years preceding the fire was 85% greater in Controls (IRR:  1.85; 95% CI:  1.71, 1.99) and 183% greater in the CCT (IRR:  2.83; 95% CI:  2.60, 3.08) compared to the CC.  It should also be noted that the incidence rate of days lost to SL or UFS/UFA MEL in the five years preceding the fire was 53% greater in the CCT than in the Controls (IRR:  1.53; 95% CI:  1.45, 1.62).  All three of these differences were statistically significant at the 95% confidence level.

Indeed, the CC were medically unfit to go to sea on 25.9% of their post-fire service days (due to either SL or MELs), as compared to 5.4% of the days for Controls.  Over the same time period, the CCT spent 5.5% of their service days on either SL or UFS/UFA MEL.  The post-fire incidence rate of days lost to SL or UFS/UFA MEL was 379% greater in the CC than in Controls (IRR:  4.79; 95% CI:  4.69, 4.89).  It should be noted that the incidence rate of days lost to SL or UFS/UFA MEL in the five years post-fire was not significantly different between the CCT and Controls (IRR:  1.01; 95% CI:  0.97, 1.05).

3.3 Medical History Text Field Diagnoses

Table 6 provides a summary of all post-fire medical conditions.  It includes newly diagnosed and pre-existing conditions.  Table 7 summarizes new diagnoses documented in the post-fire period.

For medical conditions newly diagnosed in the post-fire time period (i.e., “new” diagnoses; Table 7) PTSD stands out among the CC, i.e., 31/52 (60%) persons received this diagnosis, compared to 2/151 (1.3%) Controls (diff:  58.3%; CI:  44.8, 71.8).  Similarly, 8/55 of the CC were diagnosed with depression^{Footnote 7} in the post-fire period, compared to 3/151 Controls (diff:  12.5%; CI:  2.9, 22.1).  Asthma or Reactive Airways Disease (RAD) was newly diagnosed in 21% (11/52) of the CC and 2.0% (3/148) of Controls (diff:  19.2%; CI:  7.9, 30.5).  There were no new cases of malignancy in the CC or the CCT in the five year post-fire period, but there were two cases in the Controls.

* Each subject group has two columns, “d” and “most common diagnoses (D)”.  The “d” column refers to the number of diagnoses identified in a corresponding diagnostic category, not to subject numbers, since subjects could have had more than one diagnosis (for example, a subject with hearing loss and tinnitus would count as two diagnoses in the ENT diagnostic category).  The “most common diagnoses (D)” column provides a short list of the most frequent diagnoses within a diagnostic category, with the number in brackets providing the number of times that the diagnosis occurred.  Specific diagnoses that only occurred once were not documented in the table.  GERD = gastro-esophageal reflux disease; PTSD = Post traumatic stress disorder; RAD = Reactive airways disease; COPD = Chronic obstructive pulmonary disease.

* Each subject group has two columns, “d” and “most common diagnoses (D)”. The “d” column refers to the number of diagnoses identified in a corresponding diagnostic category, not to subject numbers, since subjects could have had more than one diagnosis (for example, a subject with hearing loss and tinnitus would count as two diagnoses in the ENT diagnostic category). The “most common diagnoses (D)” column provides a short list of the most frequent diagnoses within a diagnostic category, with the number in brackets providing the number of times that the diagnosis occurred. Specific diagnoses that only occurred once were not documented in the table. GERD = gastro-esophageal reflux disease; PTSD = Post traumatic stress disorder; RAD = Reactive airways disease; COPD = Chronic obstructive pulmonary disease.

The Relative risk (RR) and 95% confidence intervals for the three leading CC post-fire new diagnoses (PTSD, Asthma/RAD, and Depression) as compared to Controls, is presented in Table 8.  Estimates are of a 45 times increase in the RR for PTSD, a tenfold increase in Asthma/RAD, and a 7.3 times increase in depression. 

Table 8:  Relative Risk calculations and 95 % Confidence Intervals for newly diagnosed PTSD, Asthma/Reactive Airways Disease (RAD), and Depression

RR = (31/52) / (2/151) = 45 (95% CI: 11, 190)

RR = (11/52) / (3/148) = 10 (95% CI: 2.9, 37)

RR = (8/55) / (3/151) = 7.3 (95% CI: 1.9, 28)

Slight increases in musculoskeletal diagnoses were noted in the CC group but did not reach statistical significance when compared to Controls or CCT.  Other diagnostic categories that showed increases post-fire compared to baseline for the CC were:  cardiovascular, neurological, and urological.  However, making sound statistical inferences of diagnoses which occurred with a low incidence was hampered by the small sample size of the CC.