Department of Epidemiology and Health Promotion and
1 Department of Mental Health and Alcohol Research, National Public Health Institute, Mannerheimintie 166, 00300 Helsinki,
2 University of Helsinki, Research Unit of Substance Abuse Medicine, P. O. Box 63, 00014 Helsinki University, Finland,
3 FIT Biotech Oyj Plc, Lenkkeilijänkatu 8, 33520 Tampere, Finland and
4 World Health Organization, Department of Non-communicable Disease Prevention and Health Promotion, 20 avenue Appia, 1211 Geneva 27, Switzerland
Received 15 June 2001; in revised form 1 November 2001; accepted 28 November 2001
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The cardiovascular disease (CVD) mortality rates in Russia are high and presently account for over 50% of all mortality (World Health Organization, 1998). The association of alcohol consumption with CVD mortality is usually seen as a J-shaped curve (Shaper et al., 1994
; Pearson, 1996
). Light to moderate alcohol consumption has been associated with reduced risk for coronary heart disease (CHD), but heavy alcohol intake provokes several cardiovascular problems such as hypertension, elevated risk for stroke, cardiomyopathy and dysrhythmias (Wannamethee and Shaper, 1992
; Shaper et al., 1994
; Pearson, 1996
).
When self-reports of alcohol drinking are used as the indicator, the respondents may underestimate or overestimate the amount of alcohol they have consumed, and it has been noted that alcoholics frequently deny that they drink very much. Therefore, it is beneficial to use biological markers of alcohol drinking to determine the alcohol-associated risk. The underestimation of alcohol consumption by population surveys (Simpura et al., 1995) is due to two major reasons: low participation by alcoholics and heavy drinkers in the health surveys and the natural tendency of all individuals to under-report their alcohol consumption in questionnaires and interviews (Simpura, 1987
). Certain biological markers, such as carbohydrate-deficient transferrin (CDT) and gamma-glutamyltransferase (GGT), which are sensitive to high alcohol intake, could be of value in estimating the real consumption rates in the population (Sillanaukee, 1996
; Salaspuro, 1999
).
CDT, which becomes elevated after 23 weeks of heavy drinking and is largely independent of non-cirrhotic liver disease, has a high sensitivity and specificity in distinguishing heavily drinking subjects from abstainers or very light social drinkers (Allen et al., 1994; Sillanaukee, 1996
). However, it has lower accuracy in drinkers with a low level of alcohol consumption (Gronbaek et al., 1995
), younger alcoholics (La Grange et al., 1994
; Huseby et al., 1997
) and in the female population (Stibler et al., 1988
; La Grange et al., 1995
).
In 1992, a collaborative project to assess CVD rates and risk factor levels and to develop feasible prevention strategies was launched in the Republic of Karelia, Russia (RUS). The population risk factor surveys in 1992 and 1997 of this project were carried out in connection with the National FINRISK Study in Finland (Puska et al., 1993). In the risk factor survey in 1992, the self-reported alcohol consumption in the Republic of Karelia was assessed and the results were compared with the neighbouring province of North Karelia in Finland (FIN) (Laatikainen et al., 1997
). The results from the survey in 1992 did not show any major differences in alcohol consumption between these two areas (Laatikainen et al., 1997
), although the national estimates of total alcohol consumption in Russia were almost double the corresponding value in Finland (Ryan, 1995
; STAKES, 1997
; Nemtsov, 2000
). The aim of the present study was to assess the self-reported alcohol consumption and its association with the levels of CDT and GGT in a random sample of the general population in the Republic of Karelia, Russia, and in North Karelia, Finland. A further aim was to study the value of these markers in assessing the alcohol consumption in populations which have different cultural backgrounds, possibly leading to mis-reporting of alcohol use.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The survey methods followed closely the World Health Organization MONICA protocol and were as identical as possible in the two areas. The survey was carried out by a trained survey team, which consisted mainly of the same persons in both areas. The survey included a self-administered questionnaire, physical measurements and biochemical analyses. The questionnaire included questions about the socio-economic background, medical history, health behaviour and use of health services. Questionnaires were translated by qualified translators and checked by the Russian survey team. In North Karelia, Finland, the questionnaires were mailed to the subjects and filled in mainly at home. In Pitkäranta, Russia, the participants filled in the questionnaires at the survey place and received assistance from the survey team, if needed. In both survey areas, the study nurses checked the questionnaires to ensure that they were fully completed. Physical measurements included height, weight, hip, waist, blood pressure and pulse measurements. A venous blood specimen was drawn for determination of GGT and CDT.
The questionnaire included two sets of questions about alcohol intake: the quantityfrequency set of questions and the last week's alcohol recall. In both sets of questions, intakes of different beverages were assessed separately. In FIN spirits, beer, wine, cider and long drinks were included, while in RUS only spirits, wine and beer were considered, because of the different availability of alcoholic beverages. In the questionnaire, the portions of alcohol asked were adapted to local availability and habits. In analyses, the portions used in the questionnaire were transformed to general proportions used. Average alcohol intake was estimated by the amount of alcohol portions consumed per week and expressed as intake of pure alcohol in g per week. In the calculations, the portion of wine, cider or spirit was considered to contain 12 g, beer 12.5 g and long drinks 14.5 g of pure alcohol. In the analyses, the estimates from both the quantityfrequency questions (Alc2) and the last week's alcohol recall (Alc1) were used.
For CDT and GGT measurements, a venous blood specimen was taken into a 10 ml vacutainer tube. After 30 min, the serum was separated by centrifugation. Fresh serum samples were transferred to the laboratory of the National Public Health Institute in Helsinki. The levels of GGT were measured by the kinetic method (Oy Medix Biochemica AB, Kauniainen, Finland) following the recommendation of the European Committee for Clinical laboratory Standards (ECCLS) (Leino et al., 1995). The cut-off values, 80 U/l for men and 50 U/l for women, were used for the GGT in the analyses.
The serum samples for the CDT analyses were frozen immediately after centrifuging and stored before analysis at 70°C. The laboratory, where the analyses were run, is under CDTect Quality Club control and the average monthly CV of the assay was <10%. Analyses were performed with a double antibody kit (CDTectTM; Pharmacia & Upjohn Diagnostics, Uppsala, Sweden). The detection limit of this test is 1 U/l and the measuring range is 5300 U/l. The cut-off values for CDT were set at 20 U/l for men and 26 U/l for women, following the recommendations of the manufacturer.
Statistical analyses were performed using SAS (Statistical Analysis System, version 6.12). Means of biochemical markers and reported alcohol intakes were analysed by area, gender and age groups. The comparisons between areas and age groups were assessed by analysis of variance separately for men and women. Logarithmic transformation of CDT, GGT and amount of alcohol intake were used in analyses. Association between self-reported alcohol intake and biochemical markers were assessed by Pearson's partial correlation analyses. Age group, body mass index (BMI) and smoking status were used as partial variables. Also in these analyses, logarithmic transformation of CDT, GGT and level of alcohol intake were used.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
This is the first study of the association of CDT values and alcohol intake conducted in the Russian Federation. It was of great interest to assess self-reported alcohol intake and its biochemical markers in a country where the estimated alcohol consumption per capita is among the highest in the world. In 1994, the estimated consumption of pure alcohol in Russia was 13.3 l per capita (Ryan, 1995; Nemtsov, 2000
) while in Finland it was only 6.9 litres per capita (STAKES, 1997
).
The surveys were carried out as identically as possible in both areas. However, some differences were inevitable. The survey was carried out 1 month later in Pitkäranta, RUS, than in North Karelia, FIN. The data collection was not identical in both countries, due to infrastructural and cultural differences. The questionnaires were hand-delivered in RUS because of unreliable mail. In FIN, the questionnaire was filled in at home, but in RUS the majority of questionnaires were filled in at the examination place, because many of them were not fully completed at home due to lack of understanding and thus the Russian subjects needed assistance from the team. This could partly explain why the self-reported alcohol consumption in RUS was low. Also the participation rate in Pitkäranta, RUS, was slightly higher in women, but lower in men, than in North Karelia, Finland. However, the reasons for non-participation are most probably very similar in these two areas, despite the environmental differences; the slightly different participation rates are, however, unlikely to have influenced the results. In either area, no general festival time or other general celebration time occurred during the study and data collection.
The self-reported alcohol consumption in both men and women in RUS and in FIN is considerably less than the official statistics would indicate for total alcohol consumption in both Finland and Russia. The difference between self-reported consumption and the official alcohol consumption statistics is likely to be due to non-participation of alcoholics and heavy drinkers and the generally recognized underestimation of alcohol consumption in self-reporting questionnaires (Simpura, 1985). However, the difference is much more evident in RUS where the self-reported alcohol consumption, in this survey as well as in previous surveys (Laatikainen et al., 1997
), seems to be 33.5 times less than that indicated by official statistics.
The discrepancy in self-reported consumption rates and biochemical markers between the areas is an interesting and more complicated issue. It is likely to be due to higher underreporting of alcohol consumption in the questionnaires in RUS. It seems that, even when the questionnaires are adapted to local availability of alcoholic beverages and drinking habits, the ways of answering vary according to culture. Drinking habits are different in Finland and Russia. The quantity frequency questionnaire and the last week's alcohol recall used in this study are the most commonly used self-report methods for assessing alcohol consumption. In the quantity frequency questionnaire, respondents tend to give the modal rather than the arithmetic volume of consumption and to underreport events with an unusually high alcohol use. Thus, the method underestimates heavy drinking (Duffy and Alanko, 1992; Romelsjö et al., 1995
). The last week's consumption method overestimates the proportion of consumers with low levels of consumption (Duffy and Alanko, 1992
). Both methods are prone to underestimate infrequent heavy use and are better at estimating alcohol amounts in frequent drinkers. The pattern of alcohol use in RUS tends to be heavy drinking at the weekends, whereas in FIN the drinking habits have slowly changed towards more frequent but less intense drinking (Simpura et al., 1995
; Laatikainen et al., 1997
).
In accordance with self-reported alcohol use, the mean serum GGT values were higher in FIN compared to RUS. However, the CDT values in RUS were significantly higher than in FIN in both men and women. GGT and CDT are both classical markers for the detection of high alcohol consumption. The average sensitivity for GGT is 60%, whereas for CDT it is as high as 80% (Stibler, 1991; Sillanaukee, 1996
). Elevated GGT activity, in addition to being linked to high alcohol use, is also associated with a variety of other conditions, for example hepatobiliary disorders, hypertriglyceridaemia, obesity, diabetes mellitus, smoking and CHD (Whitfield et al., 1972
; Ellis et al., 1979
; Kristenson et al., 1980
; Salaspuro, 1989
; Sillanaukee, 1996
). Thus, the higher GGT values among men in North Karelia could be partly explained by a higher body mass index, higher cholesterol levels and higher prevalence of diabetes (Puska et al., 1993
; Valle et al., 1997
). False-positive CDT values are rare and arise primarily from a genetic D-variant of transferrin (Stibler et al., 1988
), an inborn error of glycoprotein metabolism (Hagberg et al., 1993
), severe liver disease, such as advanced cirrhosis with ascites (Takase et al., 1985
), primary biliary cirrhosis (Behrens et al., 1988
), chronic active hepatitis (Stauber et al., 1995
) and chronic viral hepatitis (Perret et al., 1997
). Chronic iron deficiency and pregnancy may also influence the response of CDT to heavy alcohol consumption (La Grange et al., 1995
; Stauber et al., 1996
). With respect to the high levels of CDT in RUS, we need to keep in mind the possibility of iron deficiency, hepatitis and cirrhosis (Behrens et al., 1988
; Stauber et al., 1995
; Perret et al., 1997
; De Feo et al., 1999
), which are all likely to be more prevalent in Russia than in Finland (Anonymous, 1995
; Kela et al., 2000
). However, in RUS, the high proportion of persons with elevated CDT values, rather than only a few very high values, tends to direct suspicion at the possibility of underreporting of alcohol use, rather than of rare diseases. Also the similar CDT values among men with no alcohol use in both areas compared with higher increase of CDT with self-reported alcohol use in RUS support the possibility of higher underreporting in RUS.
There were statistically significant differences in self-reported alcohol consumption and also in biochemical markers, especially among women. The mean CDT and proportion of persons with high CDT values are to some extent in accordance with self-reported alcohol consumption data. One exception was found in the youngest age groups both in RUS and FIN, where the proportion of persons with elevated values was higher than in the older age groups and even higher than among men in FIN. For the females, this is most probably due to an iron deficiency anaemia, which is usually most prevalent among fertile-aged women with menstruation and childbirth and which increases CDT levels (Looker et al., 1997; Quintas et al., 1997
; De Feo et al., 1999
). The high mean GGT and proportion of elevated values among the oldest age groups, who report very low rates of alcohol consumption, can be explained by the high prevalence of obesity in this age group (Puska et al., 1993
).
Both CDT and GGT showed a relatively high correlation with self-reported alcohol use. However, their correlation with each other was surprisingly low. This suggests that the responses of CDT and GGT to alcohol consumption may occur through different mechanisms (Randell et al., 1998). Only in men in RUS did the CDT values correlate significantly with the GGT values. One explanation for this might be that men in RUS, in addition to alcohol consumption and smoking, did not have many other factors (i.e. obesity or high coffee consumption) that could have influenced their GGT levels (Poikolainen and Vartiainen, 1997
).
In conclusion, the high levels of CDT and the high proportion of the population with elevated CDT values suggest higher alcohol consumption among both men and women in RUS, when compared to FIN. There are no previous data available on CDT levels in Russia. It would be of interest to compare these values from the Republic of Karelia to other areas of Russia. The self-reported alcohol consumption data and the GGT levels provided opposite results, compared to the CDT data. The high GGT values in FIN could be explained by other risk factors and conditions affecting GGT levels. It is very likely that people in the Republic of Karelia, Russia, are more prone to misreport their alcohol consumption, than people in North Karelia, Finland.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
FOOTNOTES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Anonymous (1995) Vital and health statistics: Russian Federation and United States, selected years 198093. Vital & Health Statistics Series 5, Comparative International Vital & Health Statistics Reports pp. 128. US Department of Health and Human Services, CDC. National Center for Statistics, Hyattsville, MD.
Behrens, U. J., Worner, T. M., Braly, L. F., Schaffner, F. and Lieber, C. S. (1988) Carbohydrate-deficient transferrin, a marker for chronic alcohol consumption in different ethnic populations. Alcoholism: Clinical and Experimental Research 12, 427432.[ISI][Medline]
De Feo, T. M., Fargion, S., Duca, L., Mattioli, M., Cappellini, M. D., Sampietro, M., Cesana, B. M. and Fiorelli, G. (1999) Carbohydrate-deficient transferrin, a sensitive marker of chronic alcohol abuse, is highly influenced by body iron. Hepatology 29, 658663.[ISI][Medline]
Duffy, J. and Alanko, T. (1992) Self-reported consumption measures in sample surveys: a simulation study of alcohol consumption. Journal of Official Statistics 8, 327350.
Ellis, G., Worthy, E. and Goldberg, D. M. (1979) Lack of value of serum gamma-glutamyl transferase in the diagnosis of hepatobiliary disease. Clinical Biochemistry 12, 142145.[ISI][Medline]
Gronbaek, M., Henriksen, J. H. and Becker, U. (1995) Carbohydrate-deficient transferrin a valid marker of alcoholism in population studies? Results from the Copenhagen City Heart Study. Alcoholism: Clinical and Experimental Research 19, 457461.[ISI][Medline]
Hagberg, B. A., Blennow, G., Kristiansson, B. and Stibler, H. (1993) Carbohydrate-deficient glycoprotein syndromes: peculiar group of new disorders. Pediatric Neurology 9, 255262.[ISI][Medline]
Huseby, N. E., Nilssen, O., Erfurth, A., Wetterling, T. and Kanitz, R. D. (1997) Carbohydrate-deficient transferrin and alcohol dependency: variation in response to alcohol intake among different groups of patients. Alcoholism: Clinical and Experimental Research 21, 201205.[ISI][Medline]
Kela, E., Holmström, P. and Nuorti, P. (2000) Tartuntataudit Suomessa 19951999. National Public Health Institute, Helsinki.
Kellera, A. (1990) MedicalGeographical Atlas of Karelian ASSR. Karelian Research Center, Petrozavodsk.
Kristenson, H., Trell, E., Fex, G. and Hood, B. (1980) Serum gamma-glutamyltransferase: statistical distribution in a middle-aged male population and evaluation of alcohol habits in individuals with elevated levels. Preventive Medicine 9, 108119.[ISI][Medline]
La Grange, L., Anton, R. F., Crow, H. and Garcia, S. (1994) A correlational study of carbohydrate-deficient transferrin values and alcohol consumption among Hispanic college students. Alcoholism: Clinical and Experimental Research 18, 653656.[ISI][Medline]
La Grange, L., Anton, R. F., Garcia, S. and Herrbold, C. (1995) Carbohydrate-deficient transferrin levels in a female population. Alcoholism: Clinical and Experimental Research 19, 100103.[ISI][Medline]
Laatikainen, T., Vartiainen, E., Puska, P., Pokusajeva, S. and Uhanov, M. (1997) Self-reported alcohol use and drinking habits in the Republic of Karelia, Russia, and in North Karelia, Finland. In Demystifying Russian Drinking, Simpura, J. and Levin, B. eds, pp. 205221. STAKES, Saarijärvi.
Leino, A., Impivaara, O., Irjala, K., Mäki, J., Peltola, O. and Järvisalo, J. (1995) Health-based reference intervals for ALAT, ASAT and GGT in serum, measured according to the recommendations of the European Committee for Clinical Laboratory Standards (ECCLS). Scandinavian Journal of Laboratory Investigation 55, 243250.
Leon, D. A., Chenet, L., Shkolnikov, V. M., Zakharov, S., Shapiro, J., Rakhmanova, G., Vassin, S. and McKee, M. (1997) Huge variation in Russian mortality rates 198494: artefact, alcohol, or what? Lancet 350, 383388.[ISI][Medline]
Looker, A. C., Dallman, P. R., Carroll, M. D., Gunter, E. W. and Johnson, C. L. (1997) Prevalence of iron deficiency in the United States. Journal of the American Medical Association 277, 973976.[Abstract]
Nemtsov, A. (2000) Estimates of total alcohol consumption in Russia, 19801994. Drug and Alcohol Dependency 58, 133142.[ISI][Medline]
Pearson, T. (1996) Alcohol and heart disease. Circulation 94, 30233025.
Perret, R., Froehlich, F., Lavanchy, D., Henry, H., Bachman, C., Pecoud, A., Bianchi, L. and Gonvers, J. J. (1997) Is carbohydrate-deficient transferrin a specific marker for alcohol abuse? A study in patients with chronic viral hepatitis. Alcoholism: Clinical and Experimental Research 21, 13371342.[ISI][Medline]
Poikolainen, K. and Vartiainen, E. (1997) Determinants of gamma-glutamyltransferase: positive interaction with alcohol and body mass index, negative association with coffee. American Journal of Epidemiology 146, 10191024.[Abstract]
Puska, P., Matilainen, T., Jousilahti, P., Korhonen, H., Vartiainen, E., Pokusajeva, S., Moisejeva, N., Uhanov, M., Kallio, I. and Artemjev, A. (1993) Cardiovascular risk factors in the Republic of Karelia, Russia, and in North Karelia, Finland. International Journal of Epidemiology 22, 10481055.[Abstract]
Quintas, M. E., Requejo, A. M., Ortega, R. M., Redondo, M. R., Lopez-Sobaler, A. M. and Gaspar, M. J. (1997) The female Spanish population: a group at risk of nutritional iron deficiency. International Journal of Food Sciences and Nutrition 48, 271279.[ISI][Medline]
Randell, E., Diamandis, E. P. and Goldberg, D. M. (1998) Changes in serum carbohydrate-deficient transferrin and gammaglutamyl transferase after moderate wine consumption in healthy males. Journal of Clinical Laboratory Analysis 12, 9297.[ISI][Medline]
Romelsjö, A., Leifman, H. and Nyström, S. (1995) A comparative study of two methods for the measurement of alcohol consumption in the general population. International Journal of Epidemiology 24, 929935.[Abstract]
Ryan, M. (1995) Alcoholism and rising mortality in the Russian Federation. British Medical Journal 310, 646648.[ISI][Medline]
Salaspuro, M. (1989) Characteristics of laboratory markers in alcohol-related organ damage. Scandinavian Journal of Gastroenterology 24, 769780.[ISI][Medline]
Salaspuro, M. (1999) Carbohydrate-deficient transferrin as compared to other markers of alcoholism: a systematic review. Alcohol 19, 261271.[ISI][Medline]
Shaper, A. G., Wannamethee, G. and Walker, M. (1994) Alcohol and coronary heart disease: a perspective from the British Regional Heart Study. International Journal of Epidemiology 23, 482494.[Abstract]
Sillanaukee, P. (1996) Laboratory markers of alcohol abuse. Alcohol and Alcoholism 31, 613616.[Abstract]
Simpura, J. (1985) Alkoholin vuosikulutus ja sen jakautuminen. In Suomalaisten juomatavat, Simpura, J. ed., pp. 5576. Alkoholitutkimussäätiö, Jyväskylä.
Simpura, J. (1987) Finnish Drinking Habits: Results from Interview Surveys Held in 1968, 1976 and 1984, Report 951-9192-36-0. Finnish Foundation for Alcohol Studies, Helsinki.
Simpura, J., Paakkanen, P. and Mustonen, H. (1995) New beverages, new drinking contexts? Signs of modernization in Finnish drinking habits from 1984 to 1992, compared with trends in the European Community. Addiction 90, 67383.[ISI][Medline]
STAKES (1997) Päihdetilastollinen vuosikirja 1997. Alkoholi ja huumeet. STAKES, Jyväskylä.
Stauber, R. E., Stepan, V., Trauner, M., Wilders-Truschnig, M., Leb, G. and Krejs, G. J. (1995) Evaluation of carbohydrate-deficient transferrin for detection of alcohol abuse in patients with liver dysfunction. Alcohol and Alcoholism 30, 171176.[Abstract]
Stauber, R. E., Vollmann, H., Pesserl, I., Jauk, B., Lipp, R., Halwachs, G. and Wilders-Truschnig, M. (1996) Carbohydrate-deficient transferrin in healthy women: relation to estrogens and iron status. Alcoholism: Clinical and Experimental Research 20, 11141117.[ISI][Medline]
Stibler, H. (1991) Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clinical Chemistry 37, 20292037.
Stibler, H., Borg, S. and Beckman, G. (1988) Transferrin phenotype and level of carbohydrate-deficient transferrin in healthy individuals. Alcoholism: Clinical and Experimental Research 12, 450453.[ISI][Medline]
Takase, S., Takada, A., Tsutsumi, M. and Matsuda, Y. (1985) Biochemical markers of chronic alcoholism. Alcohol 2, 405410.[ISI][Medline]
Tarschys, D. (1993) The success of a failure: Gorbachev's alcohol policy, 198588. EuropeAsia Studies 45, 725.
Valle, T., Tuomilehto, J. and Eriksson, J. (1997) Epidemiology of NIDDM in Europids. In International Textbook of Diabetes Mellitus, Alberti, K., Zimmet, P., DeFronzo, R. and Keen, H. eds, pp. 125142. Wiley, Chichester.
Wannamethee, G. and Shaper, A. G. (1992) Blood lipids: the relationship with alcohol intake, smoking, and body weight. Journal of Epidemiology and Community Health 46, 197202.[Abstract]
Whitfield, J. B., Pounder, R. E., Neale, G. and Moss, D. W. (1972) Serum -glutamyl transpeptidase activity in liver disease. Gut 13, 702708.[ISI][Medline]
World Health Organization (1998) World Health Statistics Annual 1996. WHO, Geneva.