4.0 Discussion

4.1 Interpretation

The HMCS CHICOUTIMI Health Surveillance Study baseline data shows that before the fire, the CC had only 2.0% of their baseline service days on SL or MELs, as compared to 2.8% for Controls and 3.9% for the CCT.  Following the fire, however, the CC experienced substantially more SL and MEL days, compared to their baseline rates or that of the Controls.  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 and 5.5% of the days for CCT.    

The diagnostic category with the greatest impact on the CC (number affected or number of SL or MEL days) was psychiatric, accounting for almost two thirds of total SL days.  The predominant diagnosis was PTSD, which was newly diagnosed in 60% of CC after the fire.  The relative risk for this diagnosis in the CC was 45 (95% CI:  11, 190) compared to Controls.  Depression as a new diagnosis occurred in 15% of CC, a relative risk of 7.3 (95% CI:  1.9, 28) as compared to the Controls. 

The second most common post-fire newly diagnosed condition among CC was asthma/reactive airways disease, with a 21% incidence and a RR of 10 (95% CI:  2.9, 37) compared to Controls. 

Post-fire the CCT had 7% more days in the “sick leave or any type of MEL” category than Controls, and although both groups had a low baseline rate, the CCT’s rate was 42% greater.  However, this effect was not great enough to cause a statistically significant difference between the CCT’s total number of days in the “lost to sick leave or Unfit Sub/Unfit Along-side MEL” category post-fire, as compared to the Controls.

No cancers were reported in the CC or the CCT during the study period.  As detailed in a previous report, it was felt unlikely that the CC exposure would increase their incidence of cancer above that of Canadian general population background levels.Footnote 8

The HMCS CHICOUTIMI Health Surveillance Study’s completion helps to fulfill the undertaking between the CF H Svcs Gp and the Clinical Council to provide answers to the HMCS CHICOUTIMI crew about the status of their health post-fire.  It also complements an earlier peer reviewed document, HMCS CHICOUTIMI Fire Incident of 5 October 2004 Potential Chemical Exposures and Health Consequences (dated 16 June 2008).

4.2 Study Limitations

4.2.1 Bias and Blinding

The CC received enhanced medical care for the first year with additional mental health screening in early 2005, whereas the CCT and Controls received usual care.  There is potential for bias due to the additional mental health surveillance of the CC (i.e., the CC might have had higher rates of case ascertainment).  The medical practitioners providing medical care were not blinded when making diagnoses, or prescribing SL or MELs to the CC (or the other groups).  The reviewers and the OHS were not blinded during the data extraction.

4.2.2 Inter-Observer Diagnostic Variability

The reviewers and OHS relied on the accuracy of the medical service providers to document all medical diagnoses present, the standard and application of which, could vary between caregivers. For example, the criteria that practitioners used to diagnose asthma/RAD might have been made subjectively as opposed to an objective diagnosis through the use of spirometry or methacholine challenge test standards.

The validation process indicated an average file reviewer accuracy of 93.3% (range 87.1% to 98.8%).  Judgment was required by the reviewers and the OHS in abstracting the data from the medical records. 

4.2.3 Controlling for Subject Demographics and Study Design

Demographically, despite randomization, there were some differences between the CC and the other two groups.  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; p=0.002).  This has to be taken into consideration when interpreting the results of the study.

Smoking was analyzed categorically as smoker, previous smoker, and non-smoker.  The development of the data to use the metric pack-years, as a measure of smoking experience, would represent a more robust method of comparing smoking exposure.  It would be needed if consideration were given to the further development of the respiratory data. 

The loss of subjects to follow-up after their release prior to 31 December 2009 limited the health information available to the OHS for the full five-year post-fire period.

SL was not recorded unless the duration was at least three days.  Total days missed for all reasons per year could provide further insights into the health experience of the subjects.

4.2.4 Statistical Power Calculations

The fixed small size of the CC limited the statistical power of the study.  Large effect sizes for common diagnoses were therefore needed to show that the incidence of a condition was different between the Controls and the CC.  In contrast, the study was not powered to make sound statistical inferences of low incident uncommon diagnoses. Such diagnoses, which could occur by chance alone, as could happen in the general population, would require analysis by comparison to other reference populations.  Similar limitations would apply to the CCT group.

4.3 Possible Development and Application of Remaining Data

Data collected from before and after a fire in a submarine is rare in the world literature.  Although further analysis of the existing data at this time will not provide significant improvement in the medical care and surveillance of the CC, it may provide guidance for similar incidents in the future (e.g., pulmonary function data).

4.4 Conclusion

Together the data and results identify, document, and provide insights into the short to medium term health effects associated with the HMCS CHICOUTIMI fire.  The high incidence of major post-fire diagnoses that were evident clinically were confirmed statistically, despite the small size of the CC cohort.  No cancers were reported in the CC or the CCT.

As outlined in the post hoc statistical power calculations referred to in the executive summary, making sound statistical inferences of diagnoses which occurred with a low incidence was hampered by the small number of crew members. This limitation will be compounded progressively as more study subjects leave the military and transition to civilian medical care.  This would make subsequent stages of the study impractical and ineffective in providing a follow-up to the crew of HCMS CHICOUTIMI.  Therefore, future monitoring of this group should occur through normal standard medical care of individuals, the Veterans Affairs claims process, and the ongoing Canadian Forces Cancer and Mortality Study IIFootnote 9. This approach would provide a more effective follow-up mechanism to identify and address long-term health concerns or outcomes.

Page details

Date modified: