REPLY to (1): We have now stated the hypothesis as follows: “exercise frequency affects the susceptibility of an adult population (aged 30+) and its majority older subgroup (aged 65+, accounting for 79% of the total) to the activity of influenza viruses in terms of influenza-associated mortality” Thus we presented both 30+ and 65+ age groups.. (See Introduction 2nd paragraph)


REPLY to (2): We have now included the details about the original study i.e. LIMOR study. (See Introduction 2nd paragraph, Method 1st and 5th paragraph). The reviewer is right to point out that the design of the original study (i.e. the LIMOR study) was not for assessing influenza-associated mortality. However, this is a population database with comprehensive data on personal lifestyles, and socio-economic and health status of the individuals. It is a unique and ideal database for assessment of effects of personal and environmental factors interactions. We have several publications on interaction between personal factors including exercise, smoking and socio-economic factor by air pollution interaction. We are now moving towards assessment for interaction between personal factors and influenza as a research area for individual susceptibility to environmental risk factors. We have the same concern on whether our results are just chance findings or not. We therefore performed the analysis with two measures of influenza (epidemic and intensity), different disease outcomes (all-cause mortality, and cardiovascular and respiratory mortality), and age groups (30+ and 65+). We also performed a series of sensitivity analyses with control for confounding factors. Our results are consistent throughout all of our analyses. We believe the results are unlikely to be due to chance alone.


REPLY to (3): We agree with the reviewer that linking deaths to influenza just because they occur during an epidemic period with the lack of a direct diagnosis to confirm influenza as a cause of death would be subject to other confounding variables. During the influenza epidemic period there may be other factors which affect one’s susceptibility to death. However the method we used for this study assesses the mortality risk of the subjects after exposure to influenza viruses for short period of time. The results are therefore robust to the confounding factors of subjects as they do not change over short period of time. Nevertheless in this study we also controlled for all identified personal confounding factors.

In our study, influenza epidemic is one of two measures for influenza activity. We also used influenza intensity as the other measure. In analysis for influenza intensity we controlled for all possible time varying confounding factors including temperature and humidity. These would minimize the chance of obtaining confounded results. With regard to the situation the reviewer raised that a subject may be more exposed during exercise, then this would probably increase than reduce the influenza-associated mortality risk. As we found that moderate exercise is related to reduced risk of influenza-associated mortality, such situation would not have occurred.

However it is necessary to conduct further study to confirm our results by means of alternative study design such as a cohort study. On the other hand our results obtained in this study would provide the evidence to generate hypotheses for further studies. (See Discussion last paragraph)


REPLY to (4): We were assessing the decedents’ exercise habits ten years before their death (around the year in 1988) by asking a close relative of the decedents. We chose this to avoid any change in the habits during development of the disease that eventually caused death, to minimize reverse causation. (See Method 1st and 5th paragraph). We also confirmed that the recall error was within an acceptable level by means of a reliability check with repeated interviews which were conducted by telephone on a random sample of 235 cases about 3 weeks on average after the initial interview. The percentage agreement for exercise frequency categorization was satisfactory (73%) (See Discussion last paragraph).


REPLY to (5): We have now revised the statements so as to be cautious, and have stated clearly and explicitly that our results may only apply particularly for older people during influenza epidemics. We clarify that we agree with a policy which encourages doing more exercise under normal circumstances. We have also recommended increasing vaccination coverage as well as doing moderate exercise during influenza epidemic periods. On the other hand doing exercise during influenza epidemics for more than three times per week, each lasts longer than 30 minutes may not be advisable for the older population.


REVIEWER:
”Overall, I think this manuscript includes many assumptions and suppositions that taken together limit the value of the conclusions that have been drawn from the data.”

REPLY: We have extensively modified the manuscript after taking into account the reviewer’s comments. In addition we have revised our discussion, pointing out that the U-shaped relationships shown in this study is consistent with those found in our recently published paper on air pollution associated mortality and levels of exercise (Wong CM et al 2007;44:386-392). In the discussion we also point out that for relationships between mortality and exercise levels in Lam et al (2004;14:391-398) which did not put influenza and air pollution in the assessment, such U-shaped patterns were not found, supporting specificity for such U-shaped relationships. We are confident that this manuscript would provide an interesting and useful addition to the literature.

Thanks for reviewed our paper and provided valuable comments.

Best wishes,
Wong CM et al