Culture, risk factors and mortality: can Switzerland add missing pieces to the European puzzle?

Transcription

1 Additional tables are published online only at jeh.bmj.om/ontent/vol63/ issue8 1 Institute of Soial and Preventive Mediine (ISPM), University of Zurih, Zurih, Switzerland; 2 Institute of Soial and Preventive Mediine, University of Bern, Bern, Switzerland Correspondene to: Dr M Bopp, University of Zurih, Institute of Soial and Preventive Mediine, Hirshengraben 84, 8001 Zürih, Switzerland; bopp@ifspm.uzh.h Aepted 23 February 2009 Culture, risk fators and mortality: an Switzerland add missing piees to the European puzzle? D Faeh, 1 C Minder, 2 F Gutzwiller, 1 M Bopp, 1 for the Swiss National Cohort Study Group ABSTRACT Bakground: The aim was to ompare ause-speifi mortality, self-rated health (SRH) and risk fators in the Frenh and German part of Switzerland and to disuss to what extent variations between these regions reflet differenes between Frane and Germany. Methods: Data were used from the general population of German and Frenh Switzerland with 2.8 million individuals aged years, ontributing deaths between 1990 and Adjusted mortality risks were alulated from the Swiss National Cohort, a longitudinal ensus-based reord linkage study. Results were ontrasted with rosssetional analyses of SRH and risk fators (Swiss Health Survey 1992/3) and with ross-setional national and international mortality rates for 1980, 1990 and Results: Despite similar all-ause mortality, there were substantial differenes in ause-speifi mortality between Swiss regions. Deaths from irulatory disease were more ommon in German Switzerland, while auses related to alohol onsumption were more prevalent in Frenh Switzerland. Many but not all of the mortality differenes between the two regions ould be explained by variations in risk fators. Similar patterns were found between Germany and Frane. Conlusion: Charateristi mortality and behavioural differentials between the German- and the Frenh-speaking parts of Switzerland ould also be found between Germany and Frane. However, some of the international variations in mortality were not in line with the Swiss regional omparison nor with differenes in risk fators. These ould relate to peuliarities in assignment of ause of death. With its ultural diversity, Switzerland offers the opportunity to examine ultural determinants of mortality without bias due to different statistial systems or national health poliies. Substantial differenes in mortality an be found between European ountries. 1 Comparing these differenes ould help to evaluate risk fators and the preditive power of self-rated health (SRH) in order to identify unexploited potentials for redution in mortality. Unfortunately, there is a serious lak of internationally omparable data. 2 The sample sizes in many European health surveys are small and longitudinal data on mortality are sare and often not omparable. Many ountries have only ross-setional mortality data, whih may be affeted by numerator/denominator bias. Another major problem of omparisons between ountries is that it is diffiult to determine whether differenes in ause-speifi mortality are real. It remains unknown how muh of the international variation is not aused by orresponding differenes in risk fators but is rather a result of variations in data olletion and proessing, of assignment of auses of death or of different definitions of risk fators or SRH. 3 5 Moreover, little is known about the ontribution of national health poliies and healthare systems to mortality differenes. Some of these limitations an be overome by omparing variation in national mortality patterns with those found between ulturally (with respet to language, lifestyle and attitudes) diverse populations within the same nation. This may offer a hane to trade off the influene of ultural against national fators and thus help to disentangle real prevention potentials from statistial artefats. Switzerland offers a unique setting for the investigation of the relationship between mortality and risk fators beause it ombines ultural diversity with a ommon national health poliy and a uniform statistial system. Thus, differenes between populations representing the German and Frenh type of ultures (eg, high and low adherene to a Mediterranean lifestyle) an be studied without bias due to differenes in national health poliies or statistial systems. Therefore, we hypothesise that Switzerland realistially mirrors variation patterns existing between the two largest European ountries. To assess this hypothesis, we analysed variations in mortality between the Frenh and German parts of Switzerland and defined ultural affiliation by language. We inluded assoiated risk fators (inludingsrh)toestimatehowmuhofthevariationan be explained by ultural differenes in behaviour. In order to investigate whether differenes found between the Swiss regions reflet differenes between Germany and Frane, we ompared our results with available data from these ountries. METHODS Swiss National Cohort The Swiss National Cohort (SNC) is a national longitudinal researh platform based on anonymous reord linkage of data olleted by the Swiss Federal Statistial Offie. The ore ohort onsists of the million residents who partiipated in the 1990 ensus. For individuals with a 1990 ensus reord (6.9%) no satisfatory link ould be found. However, the majority of unlinked reords related to individuals aged years who were not inluded in our analysis (for more details see Bopp et al 6 ). The 1990 and 2000 ensuses in Switzerland were arried out with self-administered questionnaires. Non-partiipation is onsidered to be very low (for the 2000 ensus, overage was estimated at 98.6%). 7 Of all registered deaths between the 1990 and the 2000 ensus, 95.3% ould be suessfully linked to the SNC. 8 For this study individuals were followed up in the period between 4 Deember 1990 and 5 Deember 2000 (ensus dates) and between the J Epidemiol Community Health 2009;63: doi: /jeh

2 ages of 45 and 74 years. Deaths and person years were only aumulated in that age span. Thus, the youngest observed subjet just had passed his 35th birthday on 4 Deember 1990 and ontributed only 1 day of observation on 4 Deember Individuals aged 75 years and older at the 1990 ensus were exluded and those reahing their 75th birthday between the ensus dates were ensored. Overall, we inluded all Swiss and foreign nationals who ould be linked to a mortality or 2000 ensus reord and lived in the German (GS, n = ) and the Frenh part of Switzerland (FS, n = ), respetively. Beause of too small ell sizes, we exluded the Italian part of the ountry (,5% of total). Cross setional mortality data Swiss mortality rates of the periods , , and originate from the offiial mortality files of the Swiss Federal Statistial Offie. Cause of death was oded entrally at the Federal Statistial Offie. Exept for the deaths of some foreign nationals that ourred abroad, mortality registration in Switzerland an be regarded as virtually omplete. Mortality data by sex and 5-year age lass for Germany and Frane (in 1980, 1990 and 2000) were extrated from the World Health Organization (WHO) World Mortality Database ( int/whosis/mort/en/index.html). Assessment of risk fators Data on SRH and risk fators stem from the first Swiss Health Survey (SHS) from 1992/3. The SHS is a ross-setional, nationwide, population-based survey onduted every 5 years by the Swiss Federal Statistial Offie to monitor publi health trends. It onsists of questions on health status and health behaviours asked of a random sample of persons aged 15 years and older living in Switzerland. Data were olleted with a telephone interview and a self-administered questionnaire. Eligible subjets were hosen by stratified random sampling of a database of all private Swiss households with telephones. The survey was ompleted by partiipants (71% partiipation rate, 52% women, for more details see Eihholzer et al 9 ). For this study, we limited the age range to years. This led to a total of 5739 people (57% women). For SRH, possible answers for desribing own health inluded very good, good, fair, poor, very poor. The last three defined less-than-good health. For simpliity, we defined only risk (and not protetive) fators, with alohol having positive and negative effets on health. Risk fators were defined as follows: Current smoking was defined when smoking >1 igarette/day; daily alohol onsumption when drinking alohol at least one per day; infrequent fruit onsumption when not eating fruits daily; physial inativity when not sweating at least one per week by performing physial ativity in leisure time. Obesity was defined as body mass index ( = weight/(height) 2 ) >30 kg/m 2. Statistial analysis We transformed the individual data from the SNC into survivaltime data using the st ommands of STATA. Mortality rates for 5-year age lasses were obtained by using the stsplit ommand. Age standardised mortality rates (per person years) and 95% CIs of seleted auses of death by region (GS and FS) and sex were obtained with the dstdize ommand. Corresponding ross-setional figures for German and Frenh Switzerland were alulated from national mortality statistis , and (numerator) and data from the 1980, 1990 and 2000 ensus (denominator). Corresponding rates for Frane and Germany were alulated from the mortality and population figures 1980, 1990 and 2000 in the WHO Mortality Database. Data for 1980 and 1990 for Germany refer to the old Federal Republi of Germany. All measures were adjusted for age by applying diret standardisation to the WHO Standard Population Europe. 10 For omparison of the SNC-based figures for of GS and FS with Germany and Frane respetively, we used the mean of the 1990 and 2000 rates for the ountries. We performed analyses fousing on the main groups of auses of death (diseases of the irulatory system, aner, other diseases and injuries). In addition to all-aner mortality, we analysed separately aners of the aerodigestive trat (UADT: oropharynx, larynx, oesophagus), stomah, intestine, liver, lung, prostate and female breast. Table A1 (available online) shows the ICD-8/9 and ICD-10 odes of seleted auses of death. For Switzerland ICD-8 was used until 1994 (thereafter ICD-10). In Germany and Frane ICD-9 was in operation for 1980 and 1990 and ICD-10 for For the analysis of the SHS data, we alulated perentages and 95% CIs using figures weighted to the Swiss population. All alulations and analyses were performed with STATA (version 9) SE (StataCorp, College Station, Texas), exept ross-setional mortality, whih was alulated with SAS (version 6.12 for Maintosh; SAS Institute, Cary, North Carolina). p Values,0.05 were onsidered signifiant. RESULTS Mortality in Swiss regions (from the SNC) Overall people aumulating person years were inluded: men and women from GS; men and women from FS. More details are shown in the appendix (see online table A2). Table 1 shows the number of deaths and the death rates (per person years) with 95% CIs by region and sex for seleted auses of death. In men, all-ause mortality rates were somewhat higher in FS than in GS, whereas in women the opposite applied. The lower mortality from oronary heart disease (CHD) and stroke in FS men was overompensated by other auses of death, notably aner. The all-ause mortality advantage of FS over GS women results mainly from irulatory disease and aner but also from other disease-related auses of death. In both sexes, the lower irulatory disease mortality of FS ompared with GS was mainly attributable to CHD and was opposed to the differene in other heart disease (OHD). When CHD and OHD are added up, the overall differene between FS and GS is about 10%, whih orresponds approximately to the differene in stroke mortality in men. GS had signifiantly higher mortality rates due to COPD (both genders), stomah aner (only men) and breast aner (women). In ontrast, men and women in FS had signifiantly higher mortality due to illdefined auses, UADT and lung aner, while rates for liver irrhosis and liver aner were higher in men only. Only minor differenes between regions were found for suiide, transport aidents, intestinal and prostate aner. Risk fators in Swiss regions (from the SHS) Figure 1 shows the prevalene of risk fators. Exept for obesity, where there were no signifiant differenes, all risk fators were more frequent in FS than in GS and the differenes tended to be larger in women than in men. The largest differenes between the regions were found for daily alohol onsumption. This differene was partiularly pronouned in women. The 640 J Epidemiol Community Health 2009;63: doi: /jeh

3 Table 1 Number of deaths, mortality rate per person years with 95% CIs by region, for seleted auses of deaths German Switzerland Frenh Switzerland Deaths Rate (95% CI) Deaths Rate (95% CI) Men All auses ( to ) ( to ) All irulatory (378.4 to 387.1) (330.9 to 345.7) Coronary heart disease (216.7 to 223.4) (158.0 to 168.2) Other heart disease (70.2 to 73.9) (90.7 to 99.1) Stroke (44.8 to 47.8) (39.1 to 44.3) All aner (415.2 to 424.4) (464.6 to 482.2) UADT aner (34.9 to 38.4) (56.3 to 62.0) Stomah aner (18.9 to 20.9) (14.9 to 18.2) Intestinal aner (41.2 to 44.2) (42.6 to 48.1) Liver aner (14.3 to 16.0) (24.6 to 28.8) Lung aner (122.1 to 127.1) (137.5 to 147.1) Prostate aner (34.8 to 37.5) (32.5 to 37.2) All other diseases (126.4 to 131.3) (128.6 to 137.4) COPD (42.2 to 45.1) (29.6 to 34.1) Liver irrhosis (27.2 to 29.6) (39.2 to 44.5) ill-defined (26.3 to 28.7) (53.7 to 59.8) All injuries (82.4 to 86.6) (85.0 to 92.7) Suiide (40.1 to 43.0) (39.9 to 45.2) Transport aident (13.3 to 15.0) (14.0 to 17.2) Women All auses (547.2 to 557.0) (521.0 to 538.2) All irulatory (134.8 to 139.6) (117.9 to 126.0) Coronary heart disease (57.6 to 60.7) (40.6 to 45.3) Other heart disease (27.1 to 30.4) (38.9 to 43.0) Stroke (26.7 to 28.9) (21.3 to 24.9) All aner (259.9 to 266.8) (249.0 to 261.1) UADT aner (6.0 to 7.1) (8.4 to 10.8) Stomah aner (7.4 to 8.6) (6.7 to 8.7) Intestinal aner (23.7 to 25.8) (21.3 to 24.9) Liver aner (3.8 to 4.6) (4.2 to 5.9) Lung aner (28.8 to 31.2) (34.1 to 38.7) Breast aner (65.3 to 68.9) (58.4 to 64.4) All other diseases (80.8 to 84.6) (67.6ç73.9) COPD (13.4 to 15.0) (9.2 to 11.6) Liver irrhosis (11.4 to 12.9) (11.9 to 14.8) Ill-defined (10.2 to 11.6) (22.3 to 26.0) All injuries (30.7 to 33.1) (31.7 to 36.2) Suiide (15.7 to 17.5) (15.8 to 19.0) Transport aident (3.8 to 4.7) (4.7 to 6.4) Swiss National Cohort , n = (age = years). COPD, hroni obstrutive pulmonary disease; UADT, upper aerodigestive trat. ICD odes of auses of death see table A1 (online). Rates were standardised to the WHO standard population Europe. 10 Data soure: Swiss Federal Statistial Offie/Swiss National Cohort. differene in less-than-good health between FS and GS was not statistially signifiant in men. Differenes in mortality between in Swiss regions and between Frane and Germany between 1980 and 2000 In men, between 1980 and 2000, all-ause mortality dereased between by 230% (Frane) and by 238% (Frenh Switzerland). For most auses, the derease was a bit stronger in the Swiss regions than in Germany and Frane. The most pronouned derease ould be observed for irulatory diseases in FS (259% in 20 years), although aompanied by the fastest inrease in ill-defined auses. In women, the derease in all-ause mortality was more similar in regions and ountries (between 230% and 233%). While the derease in aner mortality was idential in GS and Germany (218%), it was stronger in FS (222%) than in Frane (210%). Figure 2 shows the proportions of seleted auses of death between 1980 and 2000 in Frane, FS, GS and Germany respetively within four broad groups (irulatory diseases, aner, other diseases, external auses). Although most absolute mortality rates showed a learly dereasing trend, generally the relative ranking of the single auses of death persisted over time. For details see table A3 (available online). Figure 3 ontrasts on a logarithmi sale the ratio between mortality rates from FS and GS with that from Frane and Germany. Values above 1 mean that rates were larger in FS and Frane than in GS and Germany respetively. Values below 1 mean the opposite. Generally the ratios between FS and GS were smaller (exept for liver irrhosis in men and UADT and liver aner in women) than those between Frane and Germany. Ratios between regions and ountries run in the opposite diretion in women for lung aner, in men for J Epidemiol Community Health 2009;63: doi: /jeh

4 Figure 1 Prevalene of risk fators (weighted proportion with 95% CI), Swiss Health Survey 1992/3 (n = 5739; age = years). Data soure: Swiss Federal Statistial Offie, Swiss Health Survey 1992/3. intestinal aner and in both sexes for liver irrhosis and for OHD. The largest differene in rate ratios between regions and ountries was found for OHD (men and women), liver irrhosis in men and lung and stomah aner in women. From figs 2 and 3 three broad groups of auses of death applying to both genders an be differentiated: Consistent gradient between Frenh- and German-speaking areas (Swiss regions and ountries): CHD, aners of UADT and liver (women only adumbrated); COPD. Gradient between FS and GS but not between ountries: OHD, breast aner, liver irrhosis (preponderantly males); lung aner shares this pattern only in men whereas in women there is also a gradient between Frane and Germany, but in the opposite diretion. No (or small) gradient between FS and GS: stroke and transport aidents (both with a tendeny towards group 1); suiide, stomah, intestinal and prostate aner. DISCUSSION In this study we examined mortality rates in the Frenh and German part of Switzerland and ompared them with assoiated risk fators and SRH and with orresponding figures from Frane and Germany. In several respets, Switzerland mirrored international patterns. Deaths from CHD, stroke and COPD were more ommon in GS and Germany, while auses related to alohol onsumption were more prevalent in FS and Frane. Mortality from injuries was higher in Frane than in Germany while there was only a tendeny for higher rates in FS than in GS. For most auses, mortality differenes were larger between ountries than between Swiss regions. Not in line with differenes between Swiss regions were OHD and liver irrhosis (both sexes) and lung aner (women). Mortality differenes between FS and GS were only poorly explained by physial inativity, obesity and infrequent fruit onsumption. Little agreement ould be observed between the proportion of those with less-than-good SRH and mortality patterns: FS women more often rated their health as less than good, despite lower allause mortality. This may be explained by a omparably more pessimisti pereption of own health status, whih is a ommon feature of the Frenh and other romane language areas. Mortality from CHD and stroke was higher in the Germanspeaking areas than in the Frenh-speaking areas but with muh larger differene between ountries than between the two Swiss regions. These larger differenes are in aordane with the higher prevalene of obesity and infrequent vegetable onsumption in Germany but at odds with the similar smoking prevalene in both ountries and with more physial ativity in Germany In our omparison, lower mortality from CHD and stroke in FS annot be explained by a lower prevalene of obesity, smoking, physial inativity or infrequent fruit onsumption sine these differenes were either marginal or went in the opposite diretion (fig 1). However, more frequent alohol onsumption is likely to ontribute to the lower CHD mortality in FS and Frane. The protetive effet of alohol appears to be partiularly strong in the framework of a healthy lifestyle, as exists in many Frenh-speaking regions Nevertheless, larger international than regional differenes in CHD annot entirely be explained by risk fators. Most likely, nationally different assignment of auses of death leads to an underestimation of CHD in Frane and one ould expet that differenes between Frane and Germany would be loser to those found between FS and GS if assignment pratie was more similar between the three ountries. In fat, a population-based autopsy study showed that death ases from irulatory diseases were overregistered in mortality statistis in Germany. 18 In ontrast, omparisons of offiial rates with figures from the MONICA (MONItoring of Trends and Determinants in CArdiovasular Disease) study as well as international omparisons suggest that mortality rates of CHD are underestimated in Frane. 519 In both sexes, higher smoking prevalene in FS was mirrored by higher lung aner rates. The opposite pattern found between ountries (higher lung aner mortality in Germany) is probably due to different stages in the smoking epidemi. 20 COPD rates were higher in both German-speaking areas despite lower smoking prevalene. 14 A part of this onfliting pattern may be due to different assignment. The diagnosti validity of COPD may be relatively poor. In fat, international differenes in COPD mortality ould not be explained by differenes in smoking. 21 In Frenh-speaking regions, COPD appears to be assigned rather relutantly. Caners of the UADT and the liver were more frequent in Frane than in Germany and in FS than in GS. Alohol onsumption, as a major risk fator was higher in the two Frenh-speaking areas than in the German-speaking areas (fig 1 and Cavelaars et al 14 ). Aordingly, the rates for liver irrhosis in men and women were lower in GS than in FS. However, on the international level rates for liver irrhosis are reversed with higher rates in Germany than in Frane. Type of alohol, obesity and nutritional aspets other than alohol (eg, pork meat, infrequent fruit and vegetable onsumption) may play a role in the development of liver irrhosis, while viral hepatitis is unlikely to have a major impat. 25 Nevertheless, known risk fators are unlikely to explain the large differene between Germany and GS. Based on the logial onstellation of mortality from liver irrhosis in Frane, FS and GS, the high rates in Germany must be regarded 642 J Epidemiol Community Health 2009;63: doi: /jeh

5 Figure 2 Proportion of auses of death as a perentage of group total based on age-standardised mortality rates (deaths per population; age = years) in 1980, 1990 and UADT, upper aerodigestive trat; COPD, hroni obstrutive pulmonary disease. Rates were standardised to the WHO standard population Europe. 10 Data soure: WHO Mortality Database ( and Swiss Federal Statistial Offie, mortality statistis. For orresponding mortality rates see table A3 (available online). Mortality rates from Germany from 1980 and 1990 refer to West Germany. *1980: ; 1990: ; 2000: as unexplained and as a hallenge for further investigation. The larger rates for stomah aner found in GS and Germany than in FS and Frane may root in different traditional diets with more proessed meat in German-speaking areas and more fruits and vegetables in Frenh speaking areas Higher onsumption of proessed meat and lower fruit and vegetable onsumption in Germany than in GS ( ) are likely to explain the larger differene between ountries. 28 However, one an speulate whether differenes between regions and ountries will derease or even disappear over time. Injuries (suiide and transport aidents) were more frequent in the Frenh-speaking populations, but with generally smaller differenes within Switzerland than between Frane and Germany. This probably reflets different national legislation onerning road traffi and possession of firearms Cultural differenes in risk pereption and behaviour (eg, patterns of alohol onsumption, adherene to speed limits, seatbelt use) ould explain the higher mortality from transport aidents in Frenh-speaking areas. Some international variation in mortality (eg, OHD and stroke) ould not be explained by the Swiss omparison and ould be rooted, among other auses, in different healthare systems. However, in this respet, there are various similarities between Germany and Frane (eg, similar per apita health expenditures, almost full insurane overage, similarly high satisfation) In fat, medial are per se has only limited impat on mortality when other health determinants are onsidered. Lifestyle fators have a muh stronger influene on disease inidene, partiularly on aner inidene Thus, poliies to promote disease prevention are likely to have a large potential for mortality redution, espeially in ountries with highly developed healthare systems. In this respet, Frane with its entralisti system may have an advantage over Germany with its federal system. For example, a smoking ban has been rapidly implemented on a nationwide level in Frane, but not in Germany. 38 Moreover, national differenes in assignment of auses of death may be more important than expeted. This may espeially be the ase for mortality from COPD, liver irrhosis, CHD versus OHD and ill-defined auses. A part of the variation between ountries and regions (eg, smoking epidemi and lung aner in FS and Frenh women) ould be due to different stages in the health transition. Trends (see online table A3) suggest that Frane and Germany will ahieve urrent Swiss mortality rates for some auses in the next one to two deades, eg, lung, UADT and stomah aner, liver irrhosis, stroke and transport aidents. Limitations Comparisons between risk fators and mortality should be interpreted autiously. Most risk fators impat with a lateny of 20 years or more on mortality. Thus, simultaneous assessments of mortality and risk fators only roughly reflet real assoiations. Also, on a population base, only ross-setional (1992/3) data on risk fators were available, preluding interpretation of trends in J Epidemiol Community Health 2009;63: doi: /jeh

6 Figure 3 Mortality rate ratios Frenh versus German Switzerland (grey, ) and Frane versus Germany (blak, average of 1990 and 2000), age = years. CHD, oronary heart disease; COPD, hroni obstrutive pulmonary disease; OHD, other heart disease; UADT, upper aerodigestive trat. Data soures: Swiss Federal Statistial Offie/Swiss National Cohort, mortality statistis (Frenh vs German Switzerland); WHO mortality database (Frane vs Germany). Mortality rates from Germany from 1990 refer to West Germany. ns: differene not statistially signifiant (p.0.05) between Frenh and German Switzerland or between Frane and Germany respetively. risk fators. However, omparisons with earlier (1977) and later (2002) surveys suggest that many major risk fators remained quite stable in the Swiss population Another limitation for omparison is overage. The SNC overs the whole population and is based on mandatory data olletion (ensus, vital statistis), whereas the SHS is a 3ø sample with a partiipation rate of 71%. Persons with an unhealthy lifestyle or belonging to lower soioeonomi groups are likely to be under-represented in health surveys. 42 Causes of death not only depend on risk fators but also on praties of assigning diagnoses. 35 This is well known, for example for ill-defined auses of death. In fat, the large variations in ill-defined auses over time, partiularly in FS, is unlikely to reflet real hanges. Probably ultural differenes also exist in reporting desirable behaviours suh as fruit onsumption or physial ativity Also our equation of Germany with GS and Frane with FS has limited validity beause these large ountries are inhomogeneous. With respet to Germany and Frane we had to rely on ross-setional data whih have inherent limitations. Finally, to preisely delineate potentials for further redution in mortality it would be essential to onsider soial inequalities in mortality and orresponding risk fators. CONCLUSION Various harateristi differenes between Frane and Germany have parallels in variations between Frenh and German Switzerland. However, this omparison also revealed magnitudes and differenes in mortality (eg, stomah and lung aner, COPD, liver irrhosis, ill-defined auses, OHD and CHD) that ould not be explained by risk fators or national healthare systems. Given the high ompleteness and quality of the data, the analysis of regional differenes in Switzerland has the potential to separate real differenes from those due to weaknesses or peuliarities in national assignment praties. When used judiiously, Swiss data an help to identify fields for publi health interventions and to antiipate future developments in Frane and Germany. Aknowledgements: We thank the Swiss Federal Statistial Offie for having enabled the onstrution of the Swiss National Cohort and for providing data from the Swiss Health Survey 1992/93 and from the mortality statistis The omparison between Frane and Germany is based on material drawn from the Mortality Database of the World Health Organization. Funding: We thank the Swiss National Siene Foundation for supporting this study (grant 33CSCO ). 644 J Epidemiol Community Health 2009;63: doi: /jeh

7 What is already known on this subjet Among European ountries substantial differenes in auses of death an be found. There is a lak of international data allowing omparisons of trends in risk fators and mortality. It is unlear how muh international variations are due to differenes in risk fators, national health poliies, statistial systems or peuliarities in assignment of auses of death. What this study adds Switzerland s language regions allow ultural influenes to be explored without bias due to differenes in national health poliies and statistial systems. We found harateristi differenes in mortality between Frenh and German Switzerland, many of them being similar to those between Frane and Germany. Some variation in mortality between the two ountries ould not be explained by the Swiss regional omparison nor by risk fators. Possible explanations inlude differenes in assignment praties or different health transition stages of ountries and regions. Swiss data may help to eluidate unexplained disrepanies in mortality. Competing interests: None. The members of the Swiss National Cohort Study Group are Felix Gutzwiller (Chairman of Exeutive Board), Matthias Bopp (both Zurih), Matthias Egger (Chairman of Sientifi Board), Adrian Spoerri, Malolm Sturdy (Data manager) and Marel Zwahlen (all Bern), Charlotte Braun-Fahrländer (Basel), Fred Paaud (Lausanne) and André Rougemont (Geneva). REFERENCES 1. Makenbah JP, Stirbu I, Roskam AJ, et al. Soioeonomi inequalities in health in 22 European ountries. N Engl J Med 2008;358(23): Makenbah JP. Health inequalities: Europe in profile An independent, expert report ommissioned by the UK Presideny of the EU. Rotterdam: Dept. of Publi Health, Erasmus MC, 2006: Kattainen A, Salomaa V, Harkanen T, et al. Coronary heart disease: from a disease of middle-aged men in the late 1970s to a disease of elderly women in the 2000s. Eur Heart J 2006;27: Lindeboom M, van Doorslaer E. Cut-point shift and index shift in self-reported health. J Health Eon 2004;23: Lozano R, Murray CJL, Lopez AD, et al. Misoding and mislassifiation of ishaemi heart disease mortality. Global Programme on Evidene for Health Poliy. Working Paper No. 12. WHO, Bopp M, Spoerri A, Zwahlen M, et al. Cohort Profile: The Swiss National Cohort a longitudinal study of 6.8 million people. Int J Epidemiol 2009;38: Bundesamt für Statistik. Eidgenössishe Volkszählung Abshlussberiht zur Volkszählung Neuhâtel, Bopp M, Gutzwiller F. Die Swiss National Cohort eine soziodemografishe Datenbasis für longitudinale Gesundheitsanalysen. Swiss National Cohort Report Nr. 5, Forshung und Dokumentation 30. Zürih: Institut für Sozial- und Präventivmedizin, Eihholzer M, Bisig B. Daily onsumption of (red) meat or meat produts in Switzerland: results of the 1992/93 Swiss Health Survey. Eur J Clin Nutr 2000;54(2): Day NE. Cumulative Rate and Cumulative Risk. In: Waterhouse JAH, MC, Shanmugaratnam K, Powell J, eds. Caner inidene in Five Continents (IARC Sientifi Publiations No. 42). Lyon: International Ageny for Researh on Caner, 1982: Jurges H. True health vs response styles: exploring ross-ountry differenes in selfreported health. Health Eon 2007;16(2): DeSalvo KB, Bloser N, Reynolds K, et al. Mortality predition with a single general self-rated health question. A meta-analysis. J Gen Intern Med 2006;21(3): Steptoe A, Wardle J, Fuller R, et al. Leisure-time physial exerise: prevalene, attitudinal orrelates, and behavioral orrelates among young Europeans from 21 ountries. Prev Med 1997;26: Cavelaars AE, Kunst A, Makenbah J. Soio-eonomi differenes in risk fators for morbidity and mortality in the European Community. J Health Psyhology 1997;2: Marques-Vidal P, Montaye M, Arveiler D, et al. Alohol onsumption and ardiovasular disease: differential effets in Frane and Northern Ireland. The PRIME study. Eur J Cardiovas Prev Rehabil 2004;11: Breslow RA, Graubard BI. Prospetive study of alohol onsumption in the United States: quantity, frequeny, and ause-speifi mortality. Alohol Clin Exp Res 2008;32: Mukamal KJ, Chiuve SE, Rimm EB. Alohol onsumption and risk for oronary heart disease in men with healthy lifestyles. Arh Intern Med 2006;166: Modelmog D, Rahlenbek S, Trihopoulos D. Auray of death ertifiates: a population-based, omplete-overage, one-year autopsy study in East Germany. Caner Causes Control 1992;3: Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, et al. Myoardial infartion and oronary deaths in the World Health Organization MONICA Projet. Registration proedures, event rates, and ase-fatality rates in 38 populations from 21 ountries in four ontinents. Cirulation 1994;90: Makenbah JP, Huisman M, Andersen O, et al. Inequalities in lung aner mortality by the eduational level in 10 European populations. Eur J Caner 2004;40(1): Brown CA, Crombie IK, Tunstall-Pedoe H. Failure of igarette smoking to explain international differenes in mortality from hroni obstrutive pulmonary disease. J Epidemiol Community Health 1994;48(2): Shüler G, Bopp M. Chronishe Bronhitis, Emphysem, Asthma. Atlas der Krebsmortalität in der Shweiz , Teil B. Basel: Birkhäuser, 1997: Bode C, Bode JC, Erhardt JG, et al. Effet of the type of beverage and meat onsumed by aloholis with aloholi liver disease. Alohol Clin Exp Res 1998;22: Festi D, Colehia A, Sao T, et al. Hepati steatosis in obese patients: linial aspets and prognosti signifiane. Obes Rev 2004;5(1): Bosetti C, Levi F, Luhini F, et al. Worldwide mortality from irrhosis: an update to J Hepatol 2007;46: Gonzalez CA, Jakszyn P, Pera G, et al. Meat intake and risk of stomah and esophageal adenoarinoma within the European Prospetive Investigation Into Caner and Nutrition (EPIC). J Natl Caner Inst 2006;98: Kono S, Hirohata T. Nutrition and stomah aner. Caner Causes Control 1996;7(1): Food and Agriulture Organization of the United Nations (FAO), Food Balane Sheets [ Ajdai-Gross V, Killias M, Hepp U, et al. Changing times: a longitudinal analysis of international firearm suiide data. Am J Publi Health 2006;96: Steptoe A, Wardle J, Fuller R, et al. Seatbelt use, attitudes, and hanges in legislation: an international study. Am J Prev Med 2002;23(4): Rihter ED, Berman T, Friedman L, et al. Speed, road injury, and publi health. Annu Rev Publi Health 2006;27: Sivak M, Soler J, Trankle U, et al. Cross-ultural differenes in driver risk-pereption. Aid Anal Prev 1989;21: van Doorslaer E, Masseria C. Inome-Related Inequality in the Use of Medial Care in 21 OECD Countries OECD Health Working Paper No Heijink R, Noethen M, Renaud T, et al. Cost of illness: an international omparison. Australia, Canada, Frane, Germany and The Netherlands. Health Poliy 2008;88(1): Green DG, Irvine B. Health Care in Frane and Germany: Lessons for the UK. London, England: Civitas: Institute for the Study of Civil Soiety, Mokdad AH, Marks JS, Stroup DF, et al. Atual auses of death in the United States, Jama 2004;291: Wen CP, Tsai SP, Chung WS. A 10-year experiene with universal health insurane in Taiwan: measuring hanges in health and health disparity. Ann Intern Med 2008;148(4): Renaud T, Blum K. Ban on smoking in Frane: a new step forward; Ban on smoking in Germany: a never ending story?! [ Health Poliy Monitor: Bertelsmann Stiftung, Jougla E, Pavillon G, Rossollin F, et al. Improvement of the quality and omparability of auses-of-death statistis inside the European Community. EUROSTAT Task Fore on auses of death statistis. Rev Epidemiol Sante Publique 1998;46: Gutzwiller F, Nater B, Martin J. Community-based primary prevention of ardiovasular disease in Switzerland: methods and results of the National Researh Program (NRP 1A). Prev Med 1985;14: Eihholzer M, Bernasoni F, Jordan P, et al. Ernährung in der Shweiz 2002 Resultate der Shweizerishen Gesundheitsbefragung. Shweiz Rundsh Med Prax 2005;94: Jousilahti P, Salomaa V, Kuulasmaa K, et al. Total and ause speifi mortality among partiipants and non-partiipants of population based health surveys: a omprehensive follow up of Finnish men and women. J Epidemiol Community Health 2005;59: D Amio M, Agozzino E, Biagino A, et al. Ill-defined and multiple auses on death ertifiates a study of mislassifiation in mortality statistis. Eur J Epidemiol 1999;15(2): Shüler G, Bopp M. Ungenau vermerkte Todesursahen. Atlas der Krebsmortalität in der Shweiz , Teil B. Basel: Birkhäuser, 1997: Klesges LM, Baranowski T, Beeh B, et al. Soial desirability bias in self-reported dietary, physial ativity and weight onerns measures in 8- to 10-year-old Afrian- Amerian girls: results from the Girls Health Enrihment Multisite Studies (GEMS). Prev Med 2004;38(Suppl):S Hebert JR, Hurley TG, Peterson KE, et al. Soial desirability trait influenes on selfreported dietary measures among diverse partiipants in a multienter multiple risk fator trial. J Nutr 2008;138(1):226S 234S. J Epidemiol Community Health 2009;63: doi: /jeh