American Journal of Epidemiology Advance Access originally published online on July 4, 2008
American Journal of Epidemiology 2008 168(5):496; doi:10.1093/aje/kwn154
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ORIGINAL CONTRIBUTIONS |
Nielsen et al. Respond to "Stress and Mortality"
1 National Institute of Public Health, Copenhagen, Denmark
2 National Research Centre for the Working Environment, Copenhagen, Denmark
3 The Copenhagen City Heart Study, Epidemiological Research Unit, Bispebjerg University Hospital, Copenhagen, Denmark
Correspondence to Dr. Naja Rod Nielsen, National Institute of Public Health, Øster Farimagsgade 5A, 1399 Copenhagen K, Denmark (e-mail: nrn{at}niph.dk).
Received for publication May 1, 2008. Accepted for publication May 2, 2008.
We welcome this opportunity to answer the comments by Hotopf et al. (1) on our paper (2). They bring up a number of important issues in relation to the topic of stress and mortality, such as the definition of stress, the validity of the measurements, confounding, causal pathways, and the potential for public health interventions in this area.
With regard to the meaning of stress, it is well known that the concept has been defined in numerous ways in the literature. Stress, as applied in our paper (2), was defined as an individual state characterized by a combination of displeasure and arousal (often called distress). This measure will obviously be affected by personality traits, coping, and support. However, we do not think that these factors "cloud" the stress measure; instead, we would find it problematic to leave such factors out of the equation because they may seriously modify the impact of an external stressor on health. We even question whether it would generally be reasonable to ignore individual perception of a stressor when addressing health consequences of stress. Doing so would be similar to ignoring the fact that people metabolize alcohol differently and are therefore exposed to varying blood levels of alcohol despite reporting the same weekly number of alcoholic drinks.
It is true that stress level is likely to fluctuate for the same person over time. In fact, this factor might be the most important methodological limitation of our study. These fluctuations will most probably create bias toward the null since some of the persons with high stress at baseline will experience reduced stress levels, and vice versa. Unfortunately, we did not have measurements of stress levels in the respondents during the long period of follow-up.
In the commentary by Hotopf et al. (1), the literature is discussed as if stress and depression are two sides of the same coin. These two conditions may be correlated, but they are far from identical. Depression is characterized by low mood, lack of self-confidence, and lack of initiative. These components were not part of our measure of stress. Conversely, stress was exemplified as the sensation of tension, nervousness, impatience, anxiety, or sleeplessness. So, we question whether conclusions regarding stress and mortality can be drawn directly from the literature on depression.
We find the discussion of the different mechanisms (1)—including lifestyle changes and physiologic changes—reasonable and very much in line with what we suggest in our paper (2). Whether an unknown common cause (confounder) may explain the observed associations cannot be ruled out, and it is of course important to keep such potential bias in mind. Observational studies give us measures of associations only; whether they can be interpreted as causal relations depends on our belief in the validity of the study.
We fully agree with the final remarks by Hotopf et al. (1) Long-term stress negatively affects quality of life and should be reduced or prevented regardless of the possible consequences for mortality. As emphasized in the paper, our results on stress and mortality should be confirmed in other studies using more comprehensive measures of stress before guiding future preventive strategies. A life course perspective would be a very reasonable way to approach this issue in future studies.
 |
ACKNOWLEDGMENTS |
|---|
Conflict of interest: none declared.
|
References |
|---|
 TOP References |
|---|
- Hotopf M, Henderson M, Kuh D. Invited commentary: stress and mortality. Am J Epidemiol (2008) 168:492–5.
[Abstract/Free Full Text] - Nielsen NR, Kristensen TS, Schnohr P, et al. Perceived stress and cause-specific mortality among men and women. Am J Epidemiol (2008) 168:481–91.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
Related articles in Am. J. Epidemiol.:
- Perceived Stress and Cause-specific Mortality among Men and Women: Results from a Prospective Cohort Study
- Naja Rod Nielsen, Tage S. Kristensen, Peter Schnohr, and Morten Grønbæk
Am. J. Epidemiol. 2008 168: 481-491.[Abstract] [FREE Full Text]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||