Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
The Russian mortality experience: education, alcohol or smoking?
- Nita G Forouhi (6 June 2003)
-
Re: The Russian mortality experience: education, alcohol or smoking?
- Sviatoslav L. Plavinski, Svetlana I. Plavinskaya (7 June 2003)
-
This is testosterone, again...
- James M. Howard (9 June 2003)
-
Time of risk factor assessment is of special importance
- Ulrich Ronellenfitsch (9 June 2003)
-
Re: Time of risk factor assessment is of special importance
- Sviatoslav L. Plavinski, Svetlana I. Plavinskaya (7 July 2003)
-
Re: Time of risk factor assessment is of special importance
- Ulrich Ronellenfitsch (10 July 2003)
|
|
|||
|
Nita G Forouhi, SpR Public Health Brent PCT, Wembley Centre for Health & Care, 116 Chaplin Road, Wembley, Middlesex, HA0 4UZ
Send response to journal:
|
Sir, Plavinski et al report that in Russian men, the sharp increase in mortality in the 1990s was only partly explained by alcohol consumption, and that social factors mediated through low education level were the more important factor (1). While their study is carefully designed and they have collected information on smoking, they do not report analyses adjusted for smoking. This is of particular importance as the vast majority of the educational gradient in excess mortality risk was for cancer and cardiovascular disease, where smoking is a powerful underlying cause. Evidence also shows a strong association between smoking and education level: in a recent study of more than 53000 persons in Germany, the less educated had a significantly higher prevalence of smoking than those with higher education (2). Furthermore, a case-control study in four Russian cities reported smoking to be the main cause of high mortality (3). While it is plausible that, as in other populations, so in Russia, social factors such as educational attainment inversely affect mortality, it is both useful and important to note how much of the social gradient is due to the lifestyle behaviours (such as smoking) that it also affects. 1. Plavinski SL et al. Social factors and increase in mortality in Russia in the 1990s: prospective cohort study. BMJ 2003; 326: 1240-42 2. Rohrmann S et al. Trends in cigarette smoking in the German centers of the European Prospective Investigation into Cancer and Nutrition (EPIC): the influence of educational level. Prev.Med. 2003; 36: 448-54 3. Zaridze DG et al. Smoking: the main cause of high mortality rate among Russian population. Vestn.Ross.Akad.Nauk. 2002; (9): 40-45 Competing interests: None declared |
|||
|
|
|||
|
Sviatoslav L. Plavinski, Dean, Colledge of Public health, MAPS 191015, St.Petersburg, Russia, Svetlana I. Plavinskaya
Send response to journal:
|
Dear Sir, Nita G Forouhi (1) in responce to our article (2) proposed hypothesis that mortality gradient in Russia could be explained by different smoking prevalence in different educational groups. Though smoking is important risk factor for both total and cardiovasular mortality it is not very probable that it played significant role in results reported by us. The smoking prevalence was almost the same in both groups of lowest education persons (current smokers 67% in mid-70th and 63% in mid-80th; ever smoked 86% in the first and 84% in the second group). It is quite improbable that in responce to social changes smoking has increased among middle-aged males, two-thrids of whom already were smokers to such extent that it caused almost 75% increase in mortality. We also have fitted Cox proportional regression model including cohort membership, age, smoking, lipid parameters, body mass index and blood pressure level separately for each educational groups. It clearly showed increased mortality in the second cohort (screened in mid-80th) even after correction for those risk factors. When we analyzed mortality separately in two cohorts, correction for smoking also was unable to eliminate influence of low education on mortality. Covariable-adjusted (Cox model) odds ratio for mortality in lowest education group was 1.34 (1.13-1.61) in the first and 1.40 (1.08- 1.81) in the second cohorts. Though smoking is important risk factor, it is explaining only part of educational gradient (3) and it probably played a relatively small role in Russian mortality raise among persons with lowest level of education, at least in our cohort. 1. Nita G Forouhi. The Russian mortality experience: education, alcohol or smoking? (rapid responce) 2. Plavinski SL et al. Social factors and increase in mortality in Russia in the 1990s: prospective cohort study. BMJ 2003; 326: 1240-42 3. Brunner E. Toward a New Social Biology. In: Social Epidemiology/ed. by L.Berkman and I. Kawachi. Oxford University Press, 2000: 306-331 Competing interests: None declared |
|||
|
|
|||
|
James M. Howard, independent biologist 1037 North Woolsey Avenue, Fayetteville 72701, U.S.A.
Send response to journal:
|
It is my hypothesis that excess testosterone adversely affects many physiological functions, i.e., increases the onset of pathology, and adversely affects final development of the brain. This adversely affects attainment of higher education and therefore, affects socioeconomic level. Hence, this accounts for the universal earlier onset of these diseases, also among the poor and less educated, in all societies. Competing interests: None declared |
|||
|
|
|||
|
Ulrich Ronellenfitsch, doctoral student University of Heidelberg, Dpt. of Tropical Hygiene and Public Health, 69120 Heidelberg, Germany
Send response to journal:
|
Sir, Plavinski et al. conclude, based on their findings in two cohorts in St. Petersburg, that mortality in Russian men has increased over the last decades for those with high-school or less than high-school education whereas it remained unchanged for those with university education(1). In a sub-analysis they assess the effects of alcohol consumption on mortality changes. They find that mortality increased substantially over time within both high- and low-quantity alcohol consumer groups while the differences between high- and low-quantity alcohol consumers were much smaller. Therefore, they conclude, alcohol can only account partly for the recent rise in Russian mortality rates. While I personally agree with their point of view that there must be more underlying factors in the dramatic levels of mortality in Russia than just alcohol I'd like to jeopardize that with their study design Plavinski et al. were really able to detect alcohol-related differences in mortality. They assessed alcohol consumption only at the time of recruitment, i.e. 1975-77 and 1986-88 respectively. This is of special importance for their later cohort since its recruitment time was coinciding with Gorbachev's anti-alcohol campaign in Russia which lasted from 1985-1988 and caused a considerable decline in alcohol consumption in the Russian population(2). This decline, however, was short-lived and alcohol consumption rose again very quickly after the end of the campaign. Therefore, many people classified into the low-consumption group at recruitment might have consumed considerably higher amounts of alcohol in the course of follow-up and therefore been misclassified. Since alcohol, in contrast to e.g. smoking or hypercholesterolemia, seems to be, at least for cardiovascular diseases, a risk factor whose consumption shows a close timely relation to the event(3), it is of special importance to assess it continously in prospective studies. An additional limitation of the study design is that only the absolute ingestion of alcohol was asked for without accounting for drinking patterns which might play a crucial role in the effects of alcohol on cardiovascular mortality(3). 1. Plavinski SL et al. Social factors and increase in mortality in Russia in the 1990s: prospective cohort study. BMJ 2003; 326: 1240-42 2. Nemtsov AV. Alcohol-related human losses in Russia in the 1980s and 1990s. Addiction 2002;97:1413-25 3. McKee M., Britton A. The positive relationship between alcohol and heart disease in eastern Europe: potential physiological mechanisms. J.R.Soc.Med. 1998;91:402-7 Competing interests: None declared |
|||
|
|
|||
|
Sviatoslav L. Plavinski, Dean, College of Public Health at MAPS 191015 St.Petersburg, Kirochnaja, 41, Svetlana I. Plavinskaya
Send response to journal:
|
Dear Sir, Taking the nature of alcohol consumption behavior and stigma associated with heavy drinking it is very difficult to disprove, or, for that matter, to prove, causative role of alcohol consumption in recent mortality rise in Russia. Due to drawbacks of questionnaires and production data, researchers should rely mainly on indirect measures. Bearing this in mind, we, in contrast to Ronellenfitsch1, think that our study design2 was sufficient to answer the question of whether alcohol alone was responsible for mortality rise. 1. Misclassification error would bias toward null hypothesis, by increasing mortality differences in low-consumption group, but decreasing them in high-consumption. If the sole reason for mortality rise is increased number of heavy drinkers we would not observe mortality difference in high-consumption group. In fact we observed a rise in mortality in high-consumption group, though it has not reached statistical signifcance due to low number of persons in this group [RR=1.35 (0.97- 1.86)]. 2. We have asked about pattern of consumption and results were the same whether we have used a 7-days recall or data on individual consumption. 3. To ensure significant rise in mortality misclassification bias should be great. Even if misclassification bias erroneusly led to misclassifying 32% of patients (this is a percentage drop in alcohol consumption between two screenings) alcohol should increase mortality 4 times to provide for observed mortality RR of 2.2. To our knowledge no study has showed that alcohol consumption could increase population mortality so much. Even if we suppose that there was non-alcohol related mortality increase 1.35 times (like observed in heavy drinkers group) it would need to propose 2.7 mortality increase in heavy compared to moderate drinkers coupled with 50% misclassification of moderate for heavy drinkers to account for observed 2.02 times rise in mortality among moderate drinkers group. 1. Ronellenfitsch U. Time of risk factor assessment is of special importance (rapid responce) 2. Plavinski SL et al. Social factors and increase in mortality in Russia in the 1990s: prospective cohort study. BMJ 2003; 326: 1240-42 Competing interests: None declared |
|||
|
|
|||
|
Ulrich Ronellenfitsch, doctoral student University of Heidelberg, Dpt. of Tropical Hygiene and Public Health, 69120 Heidelberg, Germany
Send response to journal:
|
Sir, in my rapid response (1) to Plavinski et al. (2) I did not intend to jeopoardize the overall conclusions they draw from their findings. It is also my strong belief that the rise in Russian mortality rates observed since the transition can only be caused by a broad range of interweaved factors. Moreover, for the vast majority of settings I doubt that an epidemiological study, even with a perfect design and an excellent statistical analysis, has the potential to prove for a risk factor to be the single cause in the etiology of a disease or, like in this case, the change of mortality rates. The human body is a highly complex system that nowadays lives in an even more complex society. Therefore, it is most frequently a multi-faceted interplay of factors which influences our health. Nevertheless, this does not impede us from drawing hints from epidemiological studies to detect habits and influences which promote the development of disease and which ought to be tackled to improve the health of the population. The aim of my response was to address in general a frequent shortcoming of longitudinal epidemiological studies. Behaviors which tend to vary over time should be assessed continously to ensure an optimal quantification of their effects on the observed outcome. I am very well aware of the usual constraints in the carrying out of these studies and I am convinced that Plavinski et al. have assessed alcohol consumption in the best possible way for the given setting. Nevertheless, conclusions drawn from such results need to be interpreted carefully. It is true that a general misclassification of the low alcohol- consumption group in this case would have led to a bias towards the null. Still, we have no information which substrata were affected by this hypothetical misclassification. One possible scenario, albeit merely speculative, could be that participants in the substrata with the lowest educational level increased their alcohol consumption substantially after the end of the anti-alcohol campaign whereas people with a higher education did not consume more, thus pinpointing towards alcohol as a strong mediating factor between poor socioeconomic status and high mortality. A broad range of other scenarios is conceivable. Nonetheless, without a close continous assessment of the levels of consumption these theories are condemned to remain highly speculative. 1. Ronellenfitsch U. Time of risk factor assessment is of special importance (rapid response) 2. Plavinski SL et al. Social factors and increase in mortality in Russia in the 1990s: prospective cohort study. BMJ 2003; 326: 1240-42 Competing interests: None declared |
|||