Suicides Among Canadian Farm Operators

8/13/2019 Suicides Among Canadian Farm Operators

1/38

In this issue

105 Suicides Among Canadian Farm OperatorsWilliam Pickett, Will D King, Taron Faelker, Ronald EM Lees, Howard I Morrisonand Monica Bienefeld

111 Mortality Attributable to Tobacco Use in Canada and its Regions,1994 and 1996Eva M Makomaski Illing and Murray J Kaiserman

118 Mental Health of the Canadian Population: A Comprehensive AnalysisThomas Stephens, Corinne Dulberg and Natacha Joubert

127 CommentaryEthical Issues in the Use of Computerized Databases for Epidemiologicand Other Health ResearchWilfreda E Thurston, Michael M Burgess and Carol E Adair

132 Status ReportConceptual Framework for Child Maltreatment SurveillanceLil Tonmyr and Gordon Phaneuf

134 Book ReviewEpidemiologic Methods for Health PolicyReviewed by Vivek Goel

Volume 20, No 31999

Chronic Diseasesin Canada

(continued on reverse)

Health

Canada

Sant

Canada


2/38

135 New Resources

137 Calendar of Events

5 Information for Authors (on i nside back cover)

Our mission is to help the people of Canada

maintain and improve their health.

Health Canada

Published by authority of the Minister of Health

Minister of Public Works and Government Services Canada 1999

ISSN 0228-8699

Aussi disponible en franais sous le titreMaladies chroniques au Canada

(Contents continued)


3/38

Suici des A mong Canadian Farm O perat ors

William Pickett, Will D King, Taron Faelker, Ronald EM Lees, Howard I Morrisonand Monica Bienefeld

Abstract

The objective of this epidemiologic study was to describe rates of suicide among male farmoperators in Canada and to compare rates with those in the general male population. Thedata were obtained from the Canadian Farm Operator Cohort (CFOC) database. Outcomemeasures were age-specific and age-standardized rates of completed suicide (ICD-9-CM

E-codes 950S959). A total of 1,457 cases of suicide were identified from the CFOC for theyears 19711987. Age-standardized rates of suicide for those aged 3069 were 29.2 (95%CI = 27.331.1) per 100,000 person-years (PYs) in the CFOC, 24.0 (95% CI = 22.1S25.8)

per 100,000 PYs in the CFOC excluding Quebec (which had data linkage concerns) and 27.0(95% CI = 26.6S27.3) per 100,000 PYs among Canadian males in general. Age-specific ratesof suicide in the CFOC increased over time. After adjustment for age differences, provincial

suicide rates among farm operators were generally lower than or equivalent to thoseobserved in the comparison populations of Canadian males. The implications of these resultsare discussed. We speculate that high levels of social support traditionally available inCanadian farm communities may protect farm operators from abnormally high rates of

suicide.

Key words: agriculture; Canada; farm; suicide; wounds and injuries

Introduction

It is widely believed that farm operators are at highrisk for a number of stress-related disorders, including

suicide.1 Although some descriptive studies in thescientific literature support this belief,2S7 epidemiologicstudies describing farm suicide rates and comparingthem with rates in the general population are rare. Onlyone existing Canadian study7 was found in a review ofthe current biomedical literature. That study concludedthat rates of suicide among Ontario farmers were lowerthan those in the general population. In contrast, studiesfrom Kentucky2,4,6 and other states in the north-centraland mid-western areas of the United States3,5 havereported higher rates of suicide among farmers than inthe general population.

In the early 1990s, researchers at Health Canada andStatistics Canada assembled a cohort of Canadian malefarm operators. This included 326,256 people who

reported to the 1971 Canada Census of Agriculture. Wehad the opportunity of using this powerful and uniquedatabase to explore patterns of farm suicide in Canadafrom 1971 to 1987. Specific objectives were to describerates of farm suicide and identify trends in these rates bytime, age and province, and to compare age-standardizedrates of suicide among farm operators with rates reportedin the Canadian male population. It was our hope thatthis basic, epidemiologic analysis would contribute tothe development and targeting of suicide preventionefforts in rural Canada.

1999 105

Author References

William Pickett, Departments of Community Health and Epidemiology andof Emergency Medicine, Queens University; andKFLA / Queens UniversityTeaching Health Unit, Kingston, Ontario

Will D King, Department of Community Health and Epidemiology, Queens University, Kingston, Ontario

Taron Faelker, Department of Community Health and Epidemiology, Queens University; andKFLA / Queens University Teaching Health Unit, Kingston,Ontario

Ronald EM Lees, Departments of Community Health and Epidemiology andof Family Medicine, Queens University, Kingston, Ontario

Howard I Morrison, Cancer Bureau, Laboratory Centre for Disease Control, Health Canada, Ottawa, Ontario

Monica Bienefeld, Department of Public Health Sciences, University of Toronto, Toronto, Ontario

Correspondence: Dr William Pickett, Assistant Professor, Department of Emergency Medicine, Queens University, Kingston General Hospital(Angada 3), Kingston, Ontario K7M 2V7; Fax: (613) 548-1381; E-mail: [email protected]


4/38

Methods

The Canadian Farm Operator Cohort (CFOC)

Statistics Canada assembled the cohort of 326,256male Canadian farm operators by linking records fromthe 1971 Census of Agriculture,8 the 1971 Central Farm

Register

9

and the 1971 Census of Population.

10

Information on all deaths in the cohort during the period1971S1987 was then obtained through linkage to theCanadian Mortality Database.11 The distribution of thiscohort by province and age group is shown in Table 1 for1971 (time of assembly), 1987 (end of follow-up) andthe full study period (1971S1987). The table providesbasic information about the size of the cohort and itssubgroups as well as attrition within the cohort overtime.

The database contains demographic, operational(farm) and mortality data on all Canadian farmers whoowned and/or operated a farm in 1971 and responded tothe 1971 Census of Agriculture. It has been used toinvestigate associations between specific agriculturalexposures and various types of cancer, includingpossible environmental causes of non-Hodgkinslymphoma,12,13 prostate cancer14 andleukemia,15 but has not been applied widelyto the epidemiologic study of otheroutcomes, such as injury and suicidemortality.

Data Analysis

All members of the CFOC whose deathswere classified as completed suicides (ninthrevision of the International Classification

of Diseases, Clinical Modification[ICD-9-CM] E-codes 950S959

16) were

identified for study.

Age-specific rates of completed suicidewithin the cohort and associated 95%confidence intervals were calculated by10-year age group for the full study period(1971S1987) and for three specificsubperiods (1971S1975, 1976S1981 and1982S1987), chosen to give roughlyequivalent intervals. Numerators for therates came from the CFOC; denominatorswere person-years (PYs) of life within

strata during each of the time periods. Thelatter were also derived from the cohort.Confidence intervals were based on thenormal approximation to the binomial.17

Age-specific rates of completed suicidefor the general population of Canadianmales (ages 20+) were also calculated.Published counts and rates of suicide wereavailable for each year from a nationalTask Force document18 and were used toderive person-years at risk for each age

group by year as well as age-specific rates and associatedconfidence intervals for the 1971S1987 period.

For the individual provinces and the country as awhole, age-standardized rates of completed suicide werecalculated for both the farm cohort and the general malepopulation, using the direct method of standardization.19

(Although previous analyses with the CFOC12,15

haveused indirect approaches to standardization, this methodwas not followed because of our desire to present andcompare individual results for each of the 10 provinces.)

In addition, the CFOC was a closed cohort that agedover time, whereas the general population represented anopen cohort with a relatively stable age distribution.Because of this, our analyses were restricted to personsaged 30S69 in order to make the two study populationsmore comparable. There were also a priori concernsabout the quality of the data linkage with the province ofQuebec.20 National rates of suicide were thereforederived both with and without that province being

included in the calculations. The 1971 Canadianpopulation of males, 20S69 years,10 was used as thestandard age structure in all of these calculations.

106 C hronic Diseases in C anada Vol 20, No 3

TABLE 1Distribution of the Canadian Farm Operator Cohort and of

suicides in the cohort, by province and age

Person-years Suicides

Province 19711987

(n= 5,052,114)

1971

(n= 325,485)

1987

(n= 258,522)

19711987

% % % n (%)

CANADA(10 provinces)

100.0 100.0 100.0 1,457 (100.0)

Newfoundland/

Labrador2.6 0.3 0.3 1 (0.1)

Nova Scotia 1.6 1.7 1.6 11 (0.8)

Prince EdwardIsland

1.3 1.3 1.3 11 (0.8)

New Brunswick 1.5 1.5 1.5 14 (1.0)

Quebec 16.3 16.8 15.3 382 (26.2)

Ontario 25.9 25.9 25.8 344 (23.6)

Manitoba 9.8 9.7 9.9 114 (23.6)

Saskatchewan 21.4 21.1 21.7 259 (7.8)

Alberta 17.4 17.1 17.7 259 (17.8)British Columbia 4.9 4.9 5.0 62 (4.3)

Age group(years)

2029 2.1 7.3 0.0 18 (1.2)

3039 10.4 18.1 1.4 156 (10.7)

4049 21.5 27.6 13.5 310 (21.3)

5059 28.1 26.9 25.2 473 (32.5)

6069 23.7 15.5 30.7 316 (21.7)

70+ 14.1 4.8 29.1 184 (12.6)


5/38

Calculation of the age-standardizedrates for the CFOC was done in thefollowing manner. Age-specific ratesper 100,000 PYs (five-year age groups,ages 30S69) were calculated for the totalstudy period (1971S1987) and the three

subperiods. These age-specific rateswere then applied to the demographicstructure of the standard population.Age-standardized rates were calculatedfor Canada (with and without Quebec)and for each province individually. Forthe comparison population (Canadianmales, ages 30S69), age-specific rates ofsuicide for 1971S1987 were derivedfrom suicide counts and rates publishedin the Canadian Task Force document.

18

Summary age-specific rates werethen calculated for each five-year agegroup, and these age-specific rates were

applied to the demographic structure ofthe standard population. Confidenceintervals for both sets of standardizedrates were calculated according toprocedures outlined by Breslow andDay.19

Results

A total of 1,457 cases of suicide were identified fromthe CFOC for the years 1971 through 1987. Thedistribution of these suicides by province is outlined inTable 1.

Table 2 provides age-specific rates of suicide in theCFOC and the general male population in Canada.Within the farm operator cohort, modest increases inrates of suicide were observed in some of the 10-yearage groups between 1971S1975 and 1976S1981. Thelargest increases in rates occurred in those aged 70 andover. This age group is unique within the cohort becauseit is open at the top end of the age distribution. Thismeans that the group became progressively older overtime and would include more farmers who were retiredfrom operating their farm on an active basis. Betweenthe second (1976S1981) and third (1982S1987)subperiods, the age-specific rates were equivalent in allage groups, with the exception of the 40S49 and 60S69

groups, which continued to show higher suicide rates.Age-specific rates for the entire 1971S1987 period are

also given for the CFOC and the general population(Table 2). Although there was some variation in theserates between the CFOC and the general population, theywere generally consistent.

Table 3 presents directly standardized rates of suicidefor the CFOC and the general Canadian male populationaged 30S69 years; the rates are provided for eachprovince and the country as a whole (with and withoutQuebec). Age-standardized rates are also provided for

the CFOC within the three subperiods. With theexception of Quebec, provincial suicide rates in the farmoperator cohort were lower than or equivalent to those inthe general population of Canadian males. When Quebecwas included in the calculation of the overall Canadianrate, the CFOC rate was slightly higher than thatobserved in the general population, and when Quebecwas excluded, the rate was slightly lower. General ratesof male suicide in Quebec were quite comparable with

the remainder of the Canadian population. In contrast,Quebec rates of suicide from the CFOC were high in1971S1975 relative to the rest of Canada, and thisdisparity increased over time (data not shown).

Discussion

This epidemiologic analysis presents rates of suicidein the Canadian population of male farmers. It alsoprovides a comparison of directly standardized farmsuicide rates with suicide rates in the general populationof Canadian males.

Two important findings are indicated by our analysis.The first is that all age-specific rates of suicide within

the farm operator cohort increased or remained steadybetween the first and last time periods studied. Similarincreases were not observed in the comparisonpopulation of Canadian males.18 The second finding isthat, after adjustment for age differences among groups,provincial rates of suicide among members of the CFOCwere lower or equivalent to those observed in the generalmale population. This was true for 9 of the 10 provinces;the only exception was the province of Quebec, whichshowed high rates of suicide among farmers comparedwith the general population.

1999 107

TABLE 2Age-specific rates of suicide (per 100,000 person-years) in the

Canadian Farm Operator Cohort (CFOC) and the generalCanadian male population

CFOC: Rate (and confidence interval [CI]) 19711987: Rate (and CI)

Agegroup(years)

19711975 19761981 19821987 CFOC CANADA

2029 15.9(7.324.6)

22.8(2.842.7)

0.0a 17.0

(9.124.8)28.4

(27.828.9)

3039 18.1(12.923.3)

42.2(33.151.3)

37.6(23.152.0)

29.7(25.034.3)

24.6(24.025.2)

4049 19.5(15.323.7)

29.4(24.034.7)

41.1(33.548.7)

28.5(25.331.7)

27.4(26.728.1)

5059 30.5(25.435.7)

33.0(28.137.9)

36.4(30.842.0)

33.3(30.336.3)

29.6(28.930.4)

6069 22.4(17.027.9)

24.4(19.829.0)

30.6(25.635.6)

26.4(23.529.3)

26.8(25.927.6)

70+ 9.5(3.615.4) 27.3(20.634.0) 29.4(23.934.9) 25.8(22.129.5) 24.3(23.425.2)

aNo suicides and small numbers of person-years observed in this stratum


6/38


7/38

Farm Operator Versus General Suicide Rates

It is clear from this study that, with the exception ofQuebec, provincial rates of suicide in the CFOC werelower than or equivalent to those observed in the generalpopulation. Also, when Quebec data were excluded fromthe overall Canadian analysis, a slightly lower rate was

observed in the CFOC.There are legitimate reasons why farm suicide rates

may be lower than those observed in the generalpopulation. Studies in the United States show that, withthe exception of cases of death by firearms, suicide ratesare greatest in the largest cities, lower in smaller citiesand lowest in rural areas.23 The vast majority of theCanadian population live in urban areas, and, if the USpattern holds for Canada, the overall Canadian rate maybe weighted by high urban suicide statistics.

As well, existing social theories suggest that rates ofsuicide vary inversely with the stability and durability ofsocial relationships within populations.

24Absent or

inadequate social support mechanisms are known riskfactors for suicide.18

Farm populations in Canada haveexperienced economic uncertainty and crises during thepast two decades, and the latter is a risk factor for suicideideation and completion.

18We would argue that, when

compared with their urban counterparts, farmers in mostparts of rural Canada belong to communities that areexceptionally supportive in times of need. Socialdisintegration is thus minimized during times of crisis,and this is reflected in the lower suicide rates.

It is also possible that the slightly lower rates of farmsuicide observed are reporting artifacts, explained bydifferential under-reporting of farm suicides by coronersand medical examiners when compared with theirreporting practices for other groups of Canadians. Again,although this theory has arisen in past studies of suicide,there is no evidence in the published biomedicalliterature that this explanation is responsible for theobserved differences.

Further Study

In this study we were able to document trends in farmsuicides over only 17 years, which is a relatively shortperiod in which to elucidate temporal trends. Additionalsurveillance is required to confirm these patterns ofsuicide in this and other farm populations (for example,among women, children, hired workers and others on

farms). The CFOC was created as a male cohort becauseof the demographic structure of the farm operatorpopulation in 1971, and our results are necessarilylimited by this restriction. Health risks to female farmoperators and other women on Canadian farms are notwell understood, and further study of farm suicide in thisimportant population is warranted. Nor does the presentstudy address societal conditions or personal and otheretiologic factors that have been shown to lead tovariations in suicide rates within other populationsforexample, social or geographic isolation, poverty andeconomic crisis.18

Initial research with the CFOC shows that singlemarital status, lower levels of education and being thesole occupant of a household are important risk factorsfor farm suicide.25 Findings from more definitiveepidemiologic studies are required to form the basis of abetter understanding of the root causes of suicide on

Canadian farms.

AcknowledgementsThis research was supported by Health Canada through the

National Health Research and Development Program, grant6606-5699-55. We thank Martha Fair, Doris Zuccarini andChristine Poliquin of the Occupational and EnvironmentalHealth Research Section, Statistics Canada, and Dr Rob Brisonof Queens University. Dr Pickett is a Career Scientist funded

by the Ontario Ministry of Health.

References1. Mendonca J (chair).Farm suicide in the counties of Elgin,

Oxford and Middlesex. Report of the Farm Suicide Task

Force of the Thames Valley District Health Council. StThomas (Ontario), 1988.

2. Piercy LR, Stallones L.Fatal accidents on Kentuckyfarms. St Joseph (Missouri): American Society ofAgricultural Engineers, 1984;Paper No 84-5508.

3. Daymond J, Gunderson P.An analysis of suicides amongthose who reside on farms in five north central states,1980S1985.Minneapolis (Minnesota): Center for HealthStatistics, Minnesota State Department of Health; 1987.

4. Stallones L. Suicide mortality among Kentucky farmers,1979S1985.Suicide Life Threat Behav1990;20:156S63.

5. Pylka KT, Gunderson PD. An epidemiologic study ofsuicide among farmers and its clinical implications.

Marshfield Clinic Bull1992;26:29S57.

6. Stallones L, Cook M. Suicide rates in Colorado from 1980

to 1989: metropolitan, nonmetropolitan, and farmcomparisons.J Rural Health1992;8:139S42.

7. Pickett W, Davidson JR, Brison RJ. Suicides on Ontariofarms.Can J Public Health 1993;84:226S30.

8. Statistics Canada.1971 Agricultural Census for Canada.Ottawa, 1972.

9. Statistics Canada.1971 Central Farm Register. Ottawa,1972.

10. Statistics Canada.1971 Census of Population for Canada.Ottawa, 1972.

11. Statistics Canada.Canada Mortality Database. Ottawa,1971S1987.

12. Wigle DT, Semenciw RM, Wilkins K, et al. Mortalitystudy of Canadian male farm operators: non-Hodgkins

lymphoma mortality and agricultural practices inSaskatchewan.J Natl Cancer Inst1990;82:575S82.

13. Morrison HI, Semenciw RM, Wilkins K, Mao Y, WigleDT. Non-Hodgkins lymphoma and agricultural practicesin the prairie provinces of Canada. Scand J Environ

Health1994;20:42S7.

14. Morrison H, Savitz D, Semenciw R, Hulka B, Mao Y,Morrison D, et al. Farming and prostate cancer mortality.

Am J Epidemiol1993;137:270S80.

15. Semenciw RM, Morrison HI, Morison D, Mao Y.Leukemia mortality and farming in the prairie provinces ofCanada.Can J Public Health1994;85:208S11.

1999 109


8/38

16. United States Department of Health and Human Services.The international classification of diseases, 9th revision,clinical modification.1989:930S1053; DHHS Pub No(PHS) 89-1260,

17. Daly LE, Bourke GJ, McGilvray J.Interpretation and usesof medical statistics. 4th ed. Oxford: Blackwell ScientificPublications, 1991:63S6.

18. Health Canada.Suicide in Canada: update of the report ofthe Task Force on Suicide in Canada. Ottawa: HealthPrograms and Services Branch, 1994.

19. Breslow NE, Day NE.Statistical methods in cancerresearch. Volume II: The design and analysis of cohort

studies.Lyon (France): International Agency for Researchon Cancer, 1987:52S61; IARC Scientific Publications No82.

20. Statistics Canada. Canadian Farm Operators MortalityStudy general work plan [internal report of the

Occupational and Environmental Health ResearchSection]. Ottawa: Statistics Canada, 1991.

21. Syer DS, Wyndowe JP. How coroners attitudes towardssuicide affect certification procedures. In: Soubrier JP,Vedrinne J, eds.Depression and suicide: proceedings ofthe 11th Congress of the International Association forSuicide Prevention. New York: Pergamon, 1983:255S61.

22. Rothman KJ.Modern epidemiology.Boston: Little, Brownand Company, 1986:84.

23. Baker SP, ONeill B, Ginsburg MJ, Li G. Suicide. In:Theinjury fact book. New York: Oxford University Press,1992.

24. Gibbs JP, Martin WT.Status integration and suicide: asociological study. Eugene (Oregon): University ofOregon, 1964.

25. Pickett W, King WD, Lees REM, et al. Suicide mortalityand pesticide use among Canadian farmers. Am J Indust

Med1998;34:364S72. O



9/38

MortalityAttributable to TobaccoUse inCanadaand its

Regions, 1994 and 1996Eva M Makomaski Illing and Murray J Kaiserman

Abstract

Using data from the National Population Health Survey and the Canadian MortalityDatabase, we applied the Smoking-Attributable Mortality, Morbidity and Economic Costmethod to estimate national and regional smoking-attributable mortality for 1994 and 1996.The results indicate that 29,229 men and 15,986 women died in 1996 as a result of smoking,including 105 children under the age of 1. This total of approximately 45,200 deathsrepresents an increase of 3,807 deaths since 1991, of which 2,445 occurred in women. The

increase in female mortality is almost entirely due to adult diseases, divided between cancers(1,026), cardiovascular diseases (743) and respiratory diseases (870).

Key words: Canada; mortality; smoking; tobacco

Introduction

In the past, Health Canada has calculatedsmoking-attributable mortality (SAM) for the surveyyears of 1985,1 19892,3 and 19914 using a modification ofthe Smoking-Attributable Mortality, Morbidity andEconomic Cost (SAMMEC) method.5 The SAMMECmethod, with its potential for estimating SAM for 26categories known to be attributable to cigarette smoking,

is considered to be one of the most reliable methodsavailable. Since 1991, new smoking behaviour data for19946 and 19967 have become available. The purpose ofthe present report is to update this earlier work and todetermine whether SAM estimates have changednationally and regionally.

Data Sources

For both 19948 and 1996,9 mortality data on 22 adultsmoking-related diseases and four pediatric diseaseslinked with maternal smoking were drawn from theCanadian Mortality Database, maintained at the HealthStatistics Division, Statistics Canada. Deaths werecategorized by diagnosis, region, sex and five-year agegroup for persons aged 3564 and 65 or over, and infantsaged less than 1 year. Diagnoses were designated bycode from the ninth revision of the InternationalClassification of Diseases.

Data on fire deaths caused by smokers material,which includes cigarettes, cigars, pipes, matches andlighters, were obtained from the 1994 Annual Report ofFire Losses in Canada,10 classified by region and sex(unpublished provincial tabulations were providedthrough the courtesy of Human Resources DevelopmentCanada). For 1996, information on fire deaths due tosmoking was not available (at the time of writing thisreport) and was not included in our 1996 SAMcalculations.

Calculation of SAM required smoking prevalencerates for adults 35 years of age or older and for womenof childbearing age (1544 years). Current, former andnever smoker rates for 1994 and 1996 by region(province), sex and age group were obtained from theNational Population Health Survey (NPHS) 1994/956

and the NPHS 1996/97.7

The NPHS is a longitudinal household-based surveyconducted every two years by Statistics Canada. Thesurvey is designed to collect information about healthstatus and health determinants, including health

behaviour, use of health services and sociodemographicinformation. The target population of the NPHS ishousehold residents in all provinces and territories,except persons living on Indian reserves, on Canadian

1999 111

Author References

Eva M Makomaski Illing,Bureau of Tobacco Control, Health Promotion and Programs Branch, Health Canada, Ottawa, Ontario

Correspondence: Dr Murray J Kaiserman, Chief, Product Safety Laboratory, Environmental Health Directorate, Health Protection Branch,Health Canada, 1800 Walkley Road, Address Locator: 6402A1, Ottawa, Ontario K1A 2A1


10/38

Forces bases or in some remote areasof Ontario and Quebec.

Prevalence rates of current smokersaged 35 and older married tonon-smokers were not available ineither NPHS. Rates were obtained

from the Survey on Smoking inCanada 1994/95, cycle 3,11 and wereused to calculate passive smokingdeaths for both 1994 and 1996.

Diagnosis-specific relative risks forsmoking-related diseases

12were

determined from the CancerPrevention Study II (CPS-II) of theAmerican Cancer Society and werepreviously presented in Collishaw andLeahys 1989 report.2 These relativerisk estimates were based on afour-year follow-up study, from 1982

to 1986, of 1.2 million entrants in theCPS-II and were used to calculateregional SAM. The relative risksobtained were those for current andformer smokers aged 3564 and 65 orover by diagnosis and sex.

Methods

The Canadian mortality data8,9 andsmoking prevalence rates6,7 for 1994and 1996, combined with the CPS-IIrelative risks,12 provide the basis forthe smoking-related mortalityestimates presented in this report.These estimates were calculatedaccording to methods presented inMakomaski Illing and Kaisermans1991 report.4

Results

The 1996 results will be discussedin detail throughout this paper, but the1994 estimates will be presented onlyin tables and figures. Total SAMestimates, by disease and region, areprovided in Tables 1A (males) and 1B(females) for 1994 and in Tables 2A

(males) and 2B (females) for 1996.

Overall, 29,229 males and 15,986 females died ofsmoking-attributable causes in 1996, including 63 boysand 42 girls under the age of 1. Cancers accounted for17,703 of the total number of such deaths, cardiovasculardiseases accounted for another 17,762 and respiratorydiseases for the remaining 9,498 deaths.

Figure 1 shows the proportions of smoking-relateddeaths in 1996, by disease category, among 29,166 men,15,944 women and 105 children (of both sexes). Of thesedeaths, lung cancer caused 8,973 (31%) male deaths and

4,519 (28%) female deaths, while ischemic heart diseasecaused 6,441 (22%) male deaths and 3,137 (22%) femaledeaths.

In 1996, 111,405 males and 101,476 females died ofall causes in Canada; the top three leading causes ofdeath in both men and women were cancer, heart diseaseand cerebrovascular disease.

13Overall, 21% of these

deaths were attributable to smoking in 1996.

Cigarette smoking, the primary risk factor for the topthree causes of death,4 was estimated to be responsiblefor 26% of all male deaths and 16% of all female deaths


TABLE 1ASmoking-attributable mortality (SAM) estimates for current and

former smokers, by disease category and region, MALES,Canada, 1994

MALES: SAM BY REGION

Disease category ICD-9code

Canada Atlantic Quebec Ontario Prairies BC

ADULT DISEASES (35+ years) 29,657 3,024 8,467 10,251 4,528 3,386

Cancers

Lip, oral cavity, pharynxEsophagusPancreasLarynxTrachea, lung, bronchusCervix uteriUrinary bladderKidney, other urinary

140149150157161162180188189

11,895

646673405352

9,004N/A450365

1,168

54633328

912N/A

3840

3,722

204138130133

2,891N/A123102

4,105

231271139110

3,064N/A173116

1,57576

1055740

1,177N/A

6257

1,32681964541

960N/A

5350

Cardiovascular diseases

Rheumatic heart diseaseHypertensionIschemic heart disease

Ages 3564Ages 65+

Pulmonary heart diseaseOther heart diseaseCerebrovascular disease

Ages 3564Ages 65+

AtherosclerosisAortic aneurysmOther arterial disease

390398401405410414

415417420429430438

440441442448

12,005

44148

2,6094,300

1031,423

4171,389

432810331

1,235

313

246434

10180

48138

348644

3,174

1431

8231,093

29378

115324

89194

84

4,166

1358

9321,565

27381

134479179283114

2,022

828

375691

23291

66248

89148

54

1,408

716

231517

14193

54201

399935

Respiratory diseases 5,578 622 1,571 1,981 931 653

Respiratory tuberculosisPneumonia/influenzaBronchitis/emphysemaAsthmaChronic airways obstruction

010012480487491492493496

201,193

79961

3,685

2123

708

418

5220292

101,043

8443220

161,294

4221129

16562

1185

8811

367

PEDIATRIC DISEASES(


11/38

in 1996. It is interesting to note that, among men,smoking accounted for 10% of all deaths fromcardiovascular disease (CVD), 11% of all deaths from

smoking-related cancers and 5% of all deaths fromrespiratory diseases; among women, these proportionswere 6%, 6% and 4% respectively.

Of the total 45,215 smoking-attributable deaths inCanada in 1996, 15,642 occurred in Ontario;12,328 inQuebec; 7,080 in the Prairies; 5,860 in British Columbia;and 4,305 in the Atlantic region (Tables 2A and 2B).

Figure 2 shows the changes in SAM for both sexesfrom 1985 to 1996. The number of smoking-attributabledeaths among women rose from 9,009 in 1985 to 15,986in 1996, an increase of 77%. The number of deaths

among men remained relatively constant throughout thisperiod (from 28,321 in 1985 to 29,229 in 1996),representing a drop in the male-to-female SAM ratio,

from 3.1 to 1.8.The SAM trends by disease category for both sexes

are presented in Figure 3. From 1989 to 1996, femaleSAM increased by 48% (5,166 deaths), whereas maleSAM increased only slightly, by 6% (1,692 deaths). Thelarge increase in female SAM is mainly due to largejumps in cancers (19%; 2,016 deaths), cardiovasculardiseases (15%; 1,670 deaths) and respiratory diseases(16%; 1,691 deaths). Among men aged 35 and over,smoking-related deaths from both cancers andrespiratory diseases rose by 3% and, from cardiovascular

1999 113

TABLE 1BSmoking-attributable mortality (SAM) estimates for current and former smokers,

by disease category and region, FEMALES, Canada, 1994

FEMALES: SAM BY REGION TOTAL

SAM

Disease category ICD-9 code Canada Atlantic Quebec Ontario Prairies BC M+FADULT DISEASES (35+ years) 15,287 1,455 3,966 5,484 2,311 2,071 44,945

Cancers 5,523 476 1,472 2,002 808 766 17,418


140149150157161162180188189

175230522

804,148

144167

58

151554

6349

1120

6

4355

13431

1,107335117

6191

18524

1,508526218

263181

9603

321710

31386810

5801716

6

820902927432

13,151144617423

Cardiovascular diseases 6,228 644 1,634 2,183 963 803 18,233


Ages 3564Ages 65+


Ages 3564Ages 65+


390398401405410414

415417420429430438

440441442448

69145

6152,588

871,012

338465443270196

517

65242

7137

2858313025

2441

183719

23265

97106

735549

2150

245882

27278

120138253

9772

1025

78387

17176

6177554532

913

45358

12156

3486314217

113293

3,2236,889

1892,435

7551,854

8751,081

527

Respiratory diseases 3,536 335 861 1,299 540 502 9,294


010012480487491492493496

81,093

41072

1,954

2116

307

180

2203130

16511

3437119

27714

1179

689

282

1158

6313

267

282,2861,209

1335,639

PEDIATRIC DISEASES (


12/38

diseases, rose minimally by 0.2%. Figure 4 comparesincreases in smoking-related deaths from lung cancer,ischemic heart disease and chronic airways obstructionbetween 1989 and 1996 among men and women.

Discussion

In 1996, smoking remained the number onepreventable cause of death and disease in Canada.Accounting for over 45,200 deaths that year, smoking farexceeded the second most important preventable causeof deathaccidents (8,603 deaths).

13Since 1991, the

number of smoking-attributable deaths in Canada has

increased by an estimated 8% (3,807 deaths); of these,almost two thirds (64%; 2,445 deaths) occurred infemales. On a regional basis, there were 1,188 moresmoking-related deaths (+20%) in the Prairies over thesame five years, followed by an increase of 1,149 deaths(+8%) in Ontario.

The number of smoking-related deaths amongfemales is rising faster than among males. In 1985, theratio of male-to-female deaths attributable to smokingwas approximately 3.1; in 1989, this ratio had fallen to2.5; in 1991, to 2.1; and in 1996, it had declined furtherto 1.8.


TABLE 2ASmoking-attributable mortality (SAM) estimates for current and former smokers,

by disease category and region, MALES, Canada, 1996

MALES: SAM BY REGION

Disease category ICD-9 code Canada Atlantic Quebec Ontario Prairies BC

ADULT DISEASES (35+ years) 28,952 2,838 8,144 9,841 4,535 3,594

Cancers 11,844 1,087 3,772 3,981 1,615 1,389


140149150157161162180188189

614702410329

8,973N/A435381

37633522

844N/A

3947

201141121136

2,935N/A131107

226278140110

2,974N/A144110

82118

6332

1,184N/A

6768

69103

5129

1,036N/A

5349

Cardiovascular diseases 11,429 1,182 2,877 3,964 1,959 1,448


Ages 3564Ages 65+

Pulmonary heart disease

Other heart diseaseCerebrovascular disease

Ages 3564Ages 65+


390398

401405410414

415417420429430438

440441442448

38148

2,3924,049

111

1,473

3781,389

326778346

116

237407

10

184

34140

179642

1330

6821,020

29

346

110307

76176

87

1352

8671,443

33

423

137460138275122

527

359677

23

302

55288

57110

57

622

248503

17

218

41194

39121

39

Respiratory diseases 5,679 569 1,495 1,897 961 757


010012480487491492493496

151,183

71063

3,708

1114

555

393

5213242

91,025

5429195

221,245

1238122

14586

2187

951463



13/38

The World Health Organization predicted that 16,000

Canadian women and 30,000 Canadian men would die in1995 from smoking-related causes.14,15 As can be seenfrom our results (Figure 2), this forecast was reasonable.

The smoking behaviour of the population two decadesearlier is reflected in the present trends insmoking-attributable mortality. Among women, smokingrates peaked in the late 1970s, and lung cancer deathrates are now more than four times as high as rates in1969.

16Among men, however, smoking rates peaked in

the mid 1960s; after decades of continuous increases,lung cancer death rates reached a peak in the late 1980sand have fallen slightly since then.

2

The number of smoking-related deaths is also

influenced by the growth and aging of the population.

17

As baby boomers age, it can be expected that largenumbers of Canadians will continue to die fromsmoking-related causes, in particular from lung cancer,heart disease and cerebrovascular disease.14

Given the levelling off of male SAM over the pastfew years coupled with the decrease in smokingprevalence among males, there is little reason to changeearlier predictions that male SAM may start to decline bythe new millennium and eventually level off, whilefemale SAM will continue to rise and may eventuallyreach, or even exceed, male levels.

1999 115

TABLE 2BSmoking-attributable mortality (SAM) estimates for current and former smokers,

by disease category and region, FEMALES, Canada, 1996

FEMALES: SAM BY REGION TOTAL

SAM

Disease category ICD-9 code Canada Atlantic Quebec Ontario Prairies BC M+FADULT DISEASES (35+ years) 15,811 1,423 4,027 5,645 2,492 2,224 44,763

Cancers 5,859 508 1,577 2,047 905 822 17,703


140149150157161162180188189

160235527

664,519

146142

63

102053

1391

1710

8

3144

14523

1,250303717

65101168

271,555

625119

31358811

687202112

243673

5636

1623

7

775937936395

13,492146577444

Cardiovascular diseases 6,133 557 1,538 2,257 982 798 17,562


Ages 3564Ages 65+


Ages 3564Ages 65+


390398401405410414

415417420429430438

440441442448

65150

5682,569

851,007

307511315270216

515

68192

6138

2544172919

2536

163644

20271

85113

616653

1854

2081,003

28319

111192144

9387

926

78372

20204

5179634931

819

50359

12145

3483293227

104298

2,9606,618

1972,550

6851,900

6411,048

562

Respiratory diseases 3,819 358 912 1,341 605 604 9,498


010012480487491492493496

111,074

41466

2,254

2105

315

216

2189132

12577

3399120

29790

2200

6810

325

3180

6311

347

262,2571,124

1305,962



14/38


Respiratory Diseases5,67919.4%

CardiovascularDiseases11,42939.1%

Pediatric Diseases

630.2%

Males Females

Cancers11,84440.5%

Passive Smoking Deaths2140.7%

Respiratory Diseases3,81923.9%

CardiovascularDiseases6,13338.4%

Pediatric Diseases

420.3%

Cancers5,85936.7%

Passive Smoking Deaths1330.8%

FIGURE 1Proportion and number of deaths due to smoking in Canada, 1996

Note: Fire deaths due to smoking are excluded because of the unavailability of 1996 data.

3.1

2.6

2.1

1.9 1.91.8

1985 1989 19911994

1995*1996**

0

5

10

15

20

25

30

35

1.5

2

2.5

3

3.5Ratio

Numberofdeaths

(thousands)

Men

Women

Ma

le-to-femaleratio

FIGURE 2Changes in SAM by sex,

Canada, selected years

* 1995 estimates were forecasted by the World Health Organization.** 1996 estimates exclude fire deaths due to smoking.

3.1

18.6

0.2

15.4

2.7

15.6

-0.2

-0.3

0.5

-1.2

6.1 47.7

Cancers

Cardiovascular diseases

RespIratory diseases

Pediatric diseases

Passive smokIng

TOTAL SAM

-10 0 10 20 30 40 50

Percentage change

Men

Women

FIGURE 3Changes in SAM by sex and disease category

between 1989 and 1996, Canada

1.7

14.9

-1.9

5.8

2.6

9.6

6.1 49.7

Lung cancer

Ischemic hear t disease

Chronic airways obstruction

ADULT DISEASES

-10 0 10 20 30 40 50

Percentage change

MenWomen

FIGURE 4Changes in SAM for leading diseases, by sex,

between 1989 and 1996, Canada


15/38

References

1. Collishaw NE, Tostowaryk W, Wigle DT. Mortalityattributable to tobacco use in Canada.Can J Public Health1988;79:166S9.

2. Collishaw NE, Leahy K. Mortality attributable to tobaccouse in Canada, 1989.Chronic Dis Can1991;12(4):46S9.

3. Morin M, Kaiserman MJ, Leahy K. Regional mortalityattributable to tobacco use in Canada, 1989.Chronic DisCan1992;13(4):64S7.

4. Makomaski Illing EM, Kaiserman MJ. Mortalityattributable to tobacco use in Canada and its regions, 1991.Can J Public Health 1995;86(4):257S65.

5. Schultz JM, Novotny TE, Rice DP.SAMMEC IIsmoking-attributable mortality, morbidity, and economiccosts[computer software and documentation]. Rockville(MD): US Department of Health and Human Services,Public Health Service, Centers for Disease Control; 1990.

6. Statistics Canada. National Population Health Survey,1994/95 [unpublished tabulations]. Ottawa.

7. Statistics Canada. National Population Health Survey,1996/97 [unpublished tabulations]. Ottawa.

8. Statistics Canada, Health Statistics Division. Deaths for allprovinces from each cause by sex and age, 1994. NationalMortality Database [non-catalogued tabulations]. Ottawa.

9. Statistics Canada, Health Statistics Division. Deaths for allprovinces from each cause by sex and age, 1996. NationalMortality Database [non-catalogued tabulations]. Ottawa.

10. Association of Canadian Fire Marshals and Fire

Commissioners.1994 annual report of fire losses inCanada. Ottawa, 1996.

11. Statistics Canada. Survey on Smoking in Canada, 1994/95,cycle 3 [unpublished tabulations]. Ottawa.

12. American Cancer Society. Cancer Prevention Study II,19821986 [unpublished tabulations]. Atlanta (GA).

13. Statistics Canada.The Daily1998 Apr 16.

14. Peto R, Lopez A, Boreham J, Thun M, Heath C Jr.Mortality from smoking in developed countries19502000, indirect estimates from national statistics.Oxford: Oxford University Press, 1994:61.

15. Peto R, Lopez AD, Boreham J, Heath CW Jr. Mortalityfrom tobacco in developed countries: indirect estimationfrom national vital statistics.Lancet1992;339:126878.

16. National Cancer Institute of Canada. Highlights. In:Canadian cancer statistics 1998. Toronto: NCIC, 1998.

17. Brancker A, Lim P. Causes of death 1991.Health Reports1992;5(2):2146. O

1999 117


16/38

MentalHealth of theCanadianPopulation:

AComprehensiveAnalysisThomas Stephens, Corinne Dulberg and Natacha Joubert

Abstract

This study examines eight measures of mental health and looks for associations with ninepotential demographic and psychosocial determinants. Data from the National PopulationHealth Survey (NPHS), analyzed by logistic regression, reveal consistently strong, graded,independent associations of current stress, social support, life events, education andchildhood traumas with both positive and negative indicators of mental health status. Sexdifferences exist for four of eight measures. For most indicators, mental health is relatively

poor among youth and improves with age. Physical and mental health problems areassociated. There is no independent relation between mental health and income adequacy or

province of residence. Two measures used in several previous Canadian surveys arerecommended for inclusion in the NPHS to better monitor population mental health.

Key words: Canada; cognitive impairment; depression; distress; mastery; mental health;population; self-esteem; sense of coherence

Introduction

Recent reports on the mental health of the Canadianpopulation have focused on specific measures and traits,notably depression,1S3 a general measure ofpsychological well-being called sense of coherence,4,5

cognitive status6 and work stress.7 To date, there havebeen no comprehensive studies covering a range ofpositive and negative indicators of mental health, nor hasthere been a systematic examination of factors associatedwith positive and negative mental health. However,Statistics Canadas National Population Health Surveyallows this type of examination.

Such an analysis is the purpose of this article. Weseek to answer two interrelated questions: What is thecurrent state of mental health in the Canadianpopulation? and Which psychosocial and physicalhealth factors are most closely associated with mentalhealth status? Answers to these questions will haveobvious implications for planning mental health services

and mental health promotion.

Our approach to the description of population mentalhealth and the analysis of its determinants has bothconceptual and practical rationales.

Conceptually, mental health is treated here as a set ofaffective/relational and cognitive attributes that permitthe individual to carry out valued functions withreservecapacityorresilienceand thus to cope effectively withchallenges to both mental and physical functioning.Happiness and work satisfaction are examples of such

desirable states, as are self-esteem, mastery and a senseof coherence. Although some consider these latterattributes asdeterminantsof mental health, they areconsidered here as (positive) indicators of mental healthstatus since they contribute to reserve capacity andcoping ability.

The population health framework identifies a widerange of conditions or determinants that influence healthstatus.8 Many of these determinantshealth services, thephysical environment, personal health practicesinitially gained prominence on the policy scene with thepublication ofA New Perspective on the Health ofCanadians.9 The social and economic environment is the

most notable addition from the population healthframework, and it is the focus of the determinantsexamined here for their contribution to mental healthstatus.


Author References

Thomas Stephens, Thomas Stephens & Associates, Manotick, Ontario

Corinne Dulberg, Epidemiology Consultant, Ottawa, Ontario

Natacha Joubert, Mental Health Promotion Unit, Health Canada, Ottawa, Ontario

Correspondence: Thomas Stephens, PO Box 837, Manotick, Ontario K4M 1A7; Fax: (613) 692-1027; Email: [email protected]


17/38

From a practical point of view, this analysis is limitedby the data available to describe the mental health of theCanadian population. Fortunately, the initial cycle of theongoing National Population Health Survey in 1994/95included a wide range of mental health indicators as wellas a full complement of plausible determinants. These

are described further under Methods.

Methods

Source of Data

This study involved secondary analysis of the publicuse data file of the 1994/95 National Population HealthSurvey (NPHS).10,11 As a result of its sample design andthe high response rate, the NPHS provides anauthoritative picture of the Canadian population living inthe 10 provinces. Data were collected by computer-assisted personal interview with both a householdrepresentative and selected individuals. For all of theindicators reported here on mental health status and

determinants, except for physical health and some of thedemographic information, data were obtained directlyfrom the selected household member; proxy responseswere not accepted. The maximum sample available foranalysis was 17,626 persons aged 12 and older; theactual sample for most analyses was usually closer to14,500 as a result of missing cases on any given variable.

Because stress was a major topic for the initial NPHS,the survey included a variety of indicators relevant tomental health, which was unprecedented for a majornational study. Although two of thesedepression anddistresshave since been promoted to core surveycontent and were thus repeated in 1996/97, we have

chosen to focus exclusively on the 1994/95 data in orderto compare associations across a large number ofindicators of status, most of which were notrepeated inthe second cycle of the NPHS. Most of the socialdeterminants in this analysis were not repeated either.

In 1994/95, the NPHS covered several self-reportedindicators of mental health on the positive dimension,including sense of coherence, self-esteem, mastery andhappiness/interest in life, and on the negative side,depression, level of distress, impact of distress andcognitive impairment. Thus mental health status ismeasured here with four positive and four negativeindicators, which are only modestly interrelated.

aThis

provides an unusual opportunity to compare associationswith determinants across many indicators.

Indicators of Mental Health and of Determinants

Sense of coherence(SOC), or psychologicalwell-being, refers to an outlook or enduring attitudewhereby life is seen as comprehensible, manageable andmeaningful. It has been shown in a number ofinternational studies to predict longevity and to relate to

physical health,12 a conclusion that appears to hold forthe Canadian population as well.4 The NPHS was thesecond population survey in the world, after Finlands,13

to measure SOC at the national level. SOC wasmeasured by means of 13 questions that weresummarized into a scale with potential scores rangingfrom 0 to 78. On the basis of the distribution, a highSOC was arbitrarily defined as a score of 67 or greater.This measure was completed only by persons aged 18and older.

Self-esteemrefers to a positive sense of self and wasassessed by six questions taken from the classicRosenberg scale used to measure this attribute.14 Noprevious national survey in Canada has assessedself-esteem. Since there is no accepted definition ofadequateself-esteem, the 25-point scale was arbitrarilydivided according to the distribution of scores. Highself-esteem refers to a score of 20 or greater.

Sense of masteryis the extent to which individuals

believe their life chances are under their control. It wasmeasured by means of seven questions,15 yielding scoresranging from 0 to 28. A high sense of mastery wasarbitrarily defined as a score of 23 or greater, based onthe distribution.

Happiness and interest in lifeis a single item from themulti-item Health Utility Index.16 Respondents wereasked Would you describe yourself as usually ... happyand interested in life, through somewhat happy andending with ... so unhappy that life is not worthwhile.This variable was dichotomized as happy/other.

Depressionis a mood disorder characterized bypervasive feelings of sadness, sometimes accompanied

by a sense of helplessness, hopelessness, irritability andphysical symptoms such as fatigue. It was measured inthe 1994/95 NPHS by a set of 27 questions about suchsymptoms, taken from the Composite InternationalDiagnostic Interview.17 The total score was an estimateof the probability that the individual had a majordepressive episode in the previous 12 months, stated insix levels with 90% as the definition of probabledepression. For the purposes of analyzing the relationwith the determinants, but not for descriptive analysis,the small and ambiguous category of possibledepression (>0% and


18/38

life or activities? and a response of either a lot orsome was used to define a life affected by distress.

Cognitive impairmentwas measured by twoquestions, one each about difficulties with thinking andremembering, which were part of the Health UtilityIndex.16 Impairment was defined as unable to remember

anything at all, unable to think or solve problems orsome difficulty thinking.

The NPHS gathered information on a wide range ofdemographic attributes, and on psychosocial andphysical health factors that are plausibly related tomental health. The demographic factors analyzed herewere age, sex, province of residence, education, incomeadequacy and household type. The psychosocial factorsand their method of measurement were childhoodtraumas (number checked on a list of 7); life events(number checked on a list of 10); current stressors,involving time pressures, others expectations andquality of social relationships (number checked on a list

of 18); a social involvement index (based on threequestions about regularity of participation in voluntarygroups and church attendance); a frequency of socialinvolvement index (based on reported frequency ofcontact with persons considered part of the respondentssocial network); and a social support index (number ofinformal social resources checked as available in theevent of need). The physical health questions werebroad, and were analyzed to ensure that they werelimited to physical health: chronic physical conditions(number checked on a list of 18) and activity restriction(long-term limitation attributed to a physical cause).Further details on these measures have been publishedby Statistics Canada.10

Analysis Methods

For the descriptive analysis of mental health statusreported in Table 1, population estimates were producedfrom the weighted frequencies, in the usual manner.These were reviewed for reliability using StatisticsCanadas suggested criteria.10

For the analysis of associations among determinants,multiple logistic regression was employed, usingweighted data for which the weights had been adjusted toan average value of 1. For this purpose, all mental healthstatus variables (the dependent measures) weredichotomized so that the target conditions became high

sense of coherence, high self-esteem, high sense ofmastery, happy and interested in life, depressed, highlevel of distress, life affected by distress and cognitivelyimpaired. Independent variables were maintained asordinal wherever applicable, although some collapsing ofcategories was necessary for reasons of sample size.

The independent variables (psychosocial factors andphysical health) described earlier were selected for themultivariate analysis on the basis of consistently strongbivariate associations. Variables of potential interestomitted from the present analysis because of their

generally weak relation to mental health wereleisure-time physical activity and regular heavydrinking.18

Initial analyses included all of the eight independentvariables reported later, plus four othersprovince ofresidence, income adequacy, social involvement and

frequency of social contacts. When these four variableswere dropped from the analyses because of their weakassociations with the indicators of mental health, it hadlittle impact on the associations between mental healthand the remaining independent variables ordeterminants.

One final control was effected in the analyses. Inaddition to controlling for the eight determinants in alllogistic regressions, level of distress was controlled forin the analyses of impact of distress. This answers thequestion of whether, regardless of the amount of distress,some persons are more affected by distress than others.

ResultsTable 1 summarizes the results on eight measures of

mental health status, by sex, age, education and provinceof residence. As already noted, four of these indicatorswere measured with continuous scales that had beencategorized for the present analysis according to theirrespective distributions. As a result, these total scores arenot absolutely meaningful; however, intergroupcomparisons are valid. By these arbitrarily definedindicators, almost one third (31%) of Canadian adultshad a high SOC, slightly more than half (52%) had highself-esteem, close to one quarter (23%) had a high senseof mastery and more than one quarter (29%) reportedsome distress.

The non-arbitrary measures paint a moderatelypositive picture overall: three quarters (74%) ofCanadians described themselves as happy and interestedin life, 6% were depressed, one in six (16%) reportedthat stress affects their lives and 9% had some cognitiveimpairment.

Although the results in Table 1 are unadjusted forrelations to other variables, it is instructive to note someconsistencies across mental health indicators. Forexample, there were similar sex differences on six ofeight indicators, suggesting modestly better mentalhealth for males than females. There were also consistent

associations between these self-reported measures ofmental health and age: on most indicators, youth aged12S19 or 12S29 had the lowest prevalence of positivemental health and the highest prevalence of mentalhealth problems.

Among the few consistent provincial differences wasthe good mental health in Newfoundland and PrinceEdward Island, where respondents reported among thehighest SOC, most happiness and least amount ofdepression and distress. No province consistently rankedlow in mental health, but Quebec was noteworthy for thenumber of measures on which it was at the extreme of



19/38

the distribution. Quebeckers reported among the highestlevels of self-esteem and mastery but the least happiness,lowest sense of coherence and most distress.

Although these descriptive data may be useful foridentifying population groups at risk, they raisequestions about the underlying reasons. To begin toanswer this question, we conducted a series of multiplelogistic regressions. The essence of this statisticaltechnique is that it reveals the unique contribution of apotential determinant of health status whilesimultaneously controlling for the influence of all other

determinants. Results are expressed as odds ratios.

Table 2 shows the relations of three demographicvariables (age, sex, education), four psychosocialvariables (childhood traumas, current stress, life eventsand social support) and two physical health variables(chronic problems and activity restriction) to the fourmeasures of positive mental health. Table 3 providessimilar information for the four measures of mentalhealth problems. Province has been omitted from theanalyses because the associations with mental healthwere consistent across provinces once all the other

factors had been taken into account. Similarly, incomeadequacy, social involvement and frequency of socialcontacts are not reported here, as they did not have anyindependent association with mental health in othermodels that were examined.

With eight dependent variables and nine independentvariables, what can be concluded from these results? Arethere demographic, psychosocial or physical healthindicators that have consistent relations with theseindicators of mental health? Is the relation with mentalhealth problems simply the inverse of any relation with

positive mental health?

Table 4 provides an overview of the findings toanswer these questions. Because of the large andcomplex sample of the NPHS and the number ofrelations used for the analysis, a strict standard forstatistical significance was adopted (p< 0.001) for thisoverview. Further, the order of the categories for eachvariable in Tables 2 and 3 was taken into account inTable 4 (but not tested for trends).

1999 121

TABLE 1Indicators of population mental health status, by sex, age, education completed

and province, Canada, ages 12+, 1994/95

Positive mental health Mental health problems

Populationestimate

(thousands)

Highsense of

coherence(%)

Highself-esteem

(%)

Highmastery

(%)

Happy,interested in

life(%)

Depressed(%)

High distresslevel(%)

Distressaffects life

(%)

Somecognitive

impairment(%)

TOTAL 23,949 31a 52 23 74 6 29 16 9

MalesFemales

11,78012,168

32a

30a5351

2521

7474

47

2632

1418

99

Ages 1219Ages 2029Ages 3039Ages 4049Ages 5059Ages 6069Ages 70+

3,3723,8795,2104,2352,8252,2822,145

12a

212730354347

44515456575148

18252426211918

72727672777673

7766523

40382925232122

17171516141517

1397968

14

Less than highschoolHigh schoolCollegeUniversity

7,9869,0073,8063,109

33283034

45535563

16232534

70747681

6655

33302623

17161414

13875

NewfoundlandPrince EdwardIslandNova ScotiaNew BrunswickQuebecOntarioManitobaSaskatchewanAlbertaBritish Columbia

483110

764626

6,0309,050

891792

2,1663,037

3935*

3029273234373030

3742

3944665136364749

1419

2115242414172423

7682

7375727474757873

##

8*4*568*5*56

2523*

2728352830232626

14*#

1917131715141518

116

1111

6101110

911

a Limited to ages 18+, thus the population estimates for the first four rows are, in thousands:Total 19,818 , Males 9,477, Females 10,341, Ages 1819 754. The provincial populations are also lower by approximately 17% than the figures reported.

* Moderate sampling variability to be interpreted with caution

# High sampling variability not sufficiently reliable for publication


20/38

The significant and gradedrelations between thesedeterminants and the measuresof mental health status may besummarized as follows.

Independent of all other

variables, age was clearly relatedto psychological well-being(SOC), which increasedimpressively with age. The oddsof seniors reporting a high senseof psychological well-beingwere as much as five times thoseof teens. Level of distress alsotended to decline with age, butnot as regularly as psychologicalwell-being increased.Interestingly, cognitiveimpairment among teens wassecond only to the level among

seniors aged 70 and older.Compared with the data in Table1, the associations between ageand mental health were fewerbut led to the sameconclusionthat poorer mentalhealth is more common amongyouth than older age groups, atleast on these indicators.

Education was stronglyrelated to six measures of mentalhealth and had a consistent,graded association with four.

Self-esteem, mastery andhappiness/interest in life allincreased with amount of formaleducation. The odds of a highsense of mastery amonguniversity graduates were 2.2times the odds amonghigh-school dropouts, even afterother factors had been accountedfor. Interestingly, with moreeducation the impact of distressbecame increasinglymorelikelyto affect ones life.

Number of childhood traumaswas strongly associated withSOC, depression and distress,and, to a lesser extent, withmastery and happiness. Suchtraumas appear not to be relatedto self-esteem, however.

Amount of current stress was one of the strongestcorrelates of mental health status, being strongly andconsistently related to all positive and negativemeasures. The odds ratios were quite high. For example,

compared with persons reporting low stress, the odds ofthose with a lot of stress being depressed were aboutthree times as high, and of being distressed, four times ashigh.

Number of life events was also important for itsapparent impact on mental health: it was negatively


TABLE 2Adjusted odds ratiosa (OR) and standard errors (SE) for four measuresof positive mental health, by demographic, psychosocial and physical

health determinants, Canada, ages 12+, 1994/95

Sense of

coherence (highvs lower)

Self-esteem(high vs lower)

Mastery(high vs lower)

Happiness(high vs lower)

Determinant OR SE

(n= 14,477b)OR SE(n= 14,665b)

OR SE(n= 14,590b)

OR SE(n= 14,703b)

Age12192029303940495059606970+


21/38

associated with three measuresof positive health and positivelywith three indicators ofproblems. Persons reporting twoor more significant life events inthe previous year had more thantwice the odds of beingdepressed as those reporting noevents.

Social support was secondonly to current stress in itsimportance for mental health: itwas strongly and positivelyassociated with SOC,self-esteem, mastery andhappiness, and negativelyrelated to depression, level ofdistress and impact of distress.Persons with high levels ofsocial support had only half the

odds of being affected bydistress, even when the amountof distress was held constant.

In contrast to most of thesedemographic and psychosocialdeterminants, the associationbetween physical and mentalhealth was limited to only someof the indicators in this study.The number of chronic physicalhealth problems was closelyassociated with depression andcognitive impairment, and

physical restriction wasassociated with all four mentalhealth problems but none of theindicators of positive mentalhealth.

After all the other variableshad been controlled for, sexdifferences remained on four ofthe eight indicators of mentalhealth: the odds were twice ashigh that women weredepressed, and they weresomewhat more prone to and

affected by distress, whereas theodds of men reporting highmastery were higher.

Discussion

These results from the NPHSprovide an unusuallycomprehensive look at the mental health of a populationand the factors that may influence it. They provideevidence of consistently strong, graded, independentrelations linking current stress, social support, life

events, education and childhood traumas to severalindicators of both positive mental health and mentalhealth problems. They also provide evidence ofdifferences related to sex, age and physical health status,although these are concentrated among the indicators of

1999 123

TABLE 3Adjusted odds ratiosa (OR) and standard errors (SE) for four measures

of mental health problems, by demographic, psychosocial andphysical health determinants, Canada, ages 12+, 1994/95

Depression

(probable vsnone)

Distress level(high vs lower)

Distress

affects lifed(some vs none)

Cognitive

Impairment(some vs none)

Determinant OR SE

(n= 14,288b)OR SE

(n= 14,674b)OR SE

(n= 11,156b)OR SE

(n= 14,708b)

Age12192029303940495059606970+


22/38

mental health problems, and there are few differences inpositive mental health associated with sex and physicalhealth. Relations between age and mental health are the

most complex but can be summarized by noting that it isusually youth who are most likely to have mental healthproblems and least likely to report positive mentalhealth.

It is important to note that terms such as influenceand determinants are not strictly correct in this context,since these results are based on cross-sectional data.Only childhood traumas and life events imply a temporalorder that is logically necessary for inferring causation,and even then longitudinal analysis would be needed toconfirm this. Indeed, in other analyses of NPHS data,SOC has been described as a determinant of physicalhealth

4rather than an outcome of it, and depression has

been described as affecting social life19

rather than beingaffected by it, as is implied here. In reality, there isprobably a negative and self-reinforcing spiral betweenmental health and many of the factors reported here. Forexample, longitudinal analysis of US population datareveals that distress leads to negative assessment ofself-reported health status, which in turn elevatessubsequent distress.20 Similarly, Icelandic data show thatchronic physical conditions affect depression, in part byundermining personal resources such as mastery andself-esteem.21

Leaving aside the question of direction of theassociations, many of the relations reported here echofindings from other population studies, which have

typically been limited to a single outcome variable.

Many of these psychosocial factors have previouslybeen reported as important for depression and otheroutcomes. For example, stressors and life events affectboth sexes in the Canadian population, whereaschildhood traumas are additionally important forwomen.

1A British cohort study links parental divorce in

childhood (an item on the childhood trauma scale) withpsychological distress at ages 23 and 33,

22and

Norwegian data reveal how economic hardship andfamily dissension (other scale items) in childhood lead topoorer mental health in adulthood.23 Longitudinal datafrom the Whitehall II study in Britain show that

emotional support predicts good mental health in menand negative social support predicts poor mental healthin both sexes.24 Among Canadian workers, psychologicaldistress is greater for women when support fromco-workers is low, and for men, when job-relatedstressors are high.25

The distribution of mental health problems amongdemographic groups reported here is consistent withother recent studies. The higher prevalence of depressionamong Canadian women replicates the results of a studyacross 10 countries, including Canada, using a different


TABLE 4Summary of relations among eight measures of mental health status and

nine demographic, psychosocial and physical health determinants,Canada, ages 12+, 1994/95

Determinant Sense of

coherenceSelf-

esteem Mastery Happy Depressed Distress

levelDistress

affects life

Some

cognitiveimpairment

Age *++ * * * * * *

Education *++ *++ *++ * *+ *

Childhood traumas * * * *++ *++ *++ *+

Current stress * * * * *++ *++ *++ *++

Life events * * * *++ *++ *++

Social support *+ *++ *++ *++ * * *

Physical healthproblems (number) * *++ *++ *++

Sex * * * *

Restricted activity * * * * *

Key

* p< 0.001 for the association of the variable with the measure of mental health statusIn addition top < 0.001 for the association,

++ a consistent, ordered positive association of all levels of the determinant+ a consistent, ordered positive association of all but one level a consistent, ordered negative association of all levels of the determinant a consistent, ordered negative association of all but one level


23/38

measure.26 Sex differences in depression have beenreported to start as early as 15 years of age in Canada. 3

However, it is noteworthy that, although this studyconsidered eight distinct measures of mental health, sexdifferences were found for only four of them. Previousstudies limited to one or two measures of mental healthproblems may have left the impression that sexdifferences are more widespread than they appear to behere.

The association we found between age and mentalhealth is important because it runs counter to bothintuition and much previous evidence. On severalindicators, mental health is shown to improve with age atleast until the middle years and, on SOC, well into thesenior years. With respect to depression and distress inCanada, this has been reported before, based on NPHSdata, and the contrast with the inverseassociationbetween age and depression in the United States hasbeen hypothesized to result from the different racialmake-up of the two populations.2 However, there may be

another explanation: in 1978/79, distress and age wereinversely related in Canada as well, according to theCanada Health Survey.26 Since that time, however, thesocial and economic lot of seniors has improvedmarkedly in Canada while that of youth has declined,relatively speaking.27 The strong link betweensocio-economic status and mental health is wellestablished, as illustrated by the results in this report andpopulation studies in Britain and the United States.28,29

Thelackof an independent relationship betweenincome adequacy and mental health status in the NPHSdata is especially interesting in this light, but consistentwith the view that education improves psychological

well-being because it provides access to non-alienatingwork29 and that, independent of income, a sense ofmastery is related to greater life satisfaction and lessdepression.30

Conclusions

This study has implications for the strategy of mentalhealth promotion, further analysis of the NPHS and otherdata, and future monitoring of mental health in Canada.

A unique feature of this study is the mix of positiveand negative outcome measures, an approach madepossible by the number and variety of relevant indicatorsin the first cycle of the NPHS. Our analysis of

associations of demographic and psychosocial factorswith all of these outcome measures leads to an importantconclusion: the psychosocial and demographic factorsassociated with mental health problems were also foundto be (inversely) associated with the indicators ofpositive mental health. This implies that strategies thatpromote resilience and other psychological resourceswill also contribute to problem reduction or evenprevention.More generally, health promotion anddisease prevention can be seen as two sides of the samecoin and entirely compatible, even mutually reinforcing.

Mental health promotion consists of establishingthose conditions that will foster resilience and support,and lead to positive states such as satisfaction andhappiness.31 It is clear from this analysis that suchconditions include, broadly stated, a reduction in currentstressors and childhood traumas and a fostering of social

support. More detailed analysis of longitudinal data fromthe NPHS and other sources is required to be morespecific about the desired conditions, although some ofthis analysis has been started with respect to job factors

7

and types of social support.24

Notwithstanding the comparative richness of themeasures in the NPHS and the fact that the measures ofdepression and distress will be repeated in every cycle,improvements are possible. In particular, it would behighly desirable to repeat the mental health statusmeasures of the 1978/79 Canada Health Survey,including the Affect Balance Scale 32 and the HealthOpinion Survey.33 This would permit more systematiccomparisons with earlier times. The Affect Balance

Scale was also used in the 1981 and 1988 Canada FitnessSurveys and the General Social Surveys of 1985 and1990; it remains in use internationally34 and isconceptually compatible with the NPHS. The HealthOpinion Survey is a long-standing measure of anxietyand depression that is conceptually similar to the currentNPHS distress scale. The possibility of direct andunambiguous comparisons of mental health status in theCanadian population over approximately 25 years arguesstrongly for its inclusion in the NPHS or comparablenational surveys of population health.

Acknowledgements

This project was supported financially by the Mental HealthPromotion Unit of Health Canada.

References1. Beaudet MP. Depression.Health Reports

1996;7(4):11S24.

2. Wade TJ, Cairney J. Age and depression in a nationallyrepresentative sample of Canadians: a preliminary look atthe National Population Health Survey.Can J Public

Health1997;88:297S302.

3. Cairney J. Gender differences in the prevalence ofdepression among Canadian adolescents.Can J Public

Health1998;89:181S2.

4. Hood SC, Beaudet MP, Catlin G. A healthy outlook.

Health Reports 1996;7(4):25S32.5. Stephens T, Catlin G.Measuring well-being in Canadas

National Population Health Survey. Presentation to theInternational Society of Behavioral Medicine; 1994 Jul;Amsterdam.

6. Hill GB, Forbes WF, Lindsay J, McDowell I, stbye T.Mortality and cognitive status among elderly Canadiansliving in the community and in institutions: the CanadianStudy of Health and Aging.Can J Public Health1997;88:303S4.

7. Wilkins K, Beaudet MP. Work stress and health.HealthReports1998;10:47S62.

1999 125


24/38

8. Federal/Provincial/Territorial Advisory Committee onPopulation Health.Strategies for population health.Ottawa: Health Canada, 1994.

9. Lalonde M.A new perspective on the health of Canadians .Ottawa: Health and Welfare Canada, 1974.

10. Statistics Canada.National Population Health Survey1994-95. Public use microdata files. Ottawa, 1995; Cat

82F0001-XDB.11. Tambay J-L, Catlin G. Sample design of the National

Population Health Survey.Health Reports1995;7:29S38.

12. Antonovsky A. The structure and properties of the senseof coherence scale.Social Sci Med1993;36:725S33.

13. Suominen SB, Helenius HY, Koskenvuo MJ, Uutela AK.Sense of coherence (SOC) as a predictor of self-reportedhealth. Presentation to the Fourth International Congressof the International Society of Behavioral Medicine; 1996Mar; Washington (DC).

14. Rosenberg M.Society and the adolescent self-image.Princeton (NJ): Princeton University Press, 1965.

15. Pearlin LI, Lieberman M, Menaghan E, Mullan J. Thestress process.J Health Soc Behav 1981;22:337S56.

16. Torrance GW, Furlong W, Feeny D, Boyle MH.Finalreport of Project No 44400900187, provisional healthstatus index for the Ontario Health Survey.Prepared forStatistics Canada, 1992 Feb.

17. Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and12-month prevalence of DSM-III-R psychiatric disordersin the United States: results from the NationalComorbidity Survey.Arch Gen Psychiatry1994;51:8S19.

18. Stephens T.Population mental health in Canada. Reportprepared for the Mental Health Promotion Unit, HealthCanada; 1998 May.

19. Bowling A. The effects of illness on quality of life:findings from a survey of households in Great Britain. J

Epidemiol Community Health 1996;50(2):14955.

20. Farmer MM, Ferraro KF. Distress and perceived health:

mechanisms of health decline.J Health Soc Behav1997;38(3):298311.

21. Vilhjalmsson R. Direct and indirect effects of chronicphysical conditions on depression: a preliminaryinvestigation.Soc Sci Med1998;47(5):60311.

22. Rodgers B, Power C, Hope S. Parental divorce and adultpsychological distress: evidence from a national birthcohort: a research note. J Child Psychol Psychiatry1997;38(7):86772.

23. Dahl E, Birkelund GE. Health inequalities in later life in asocial democratic welfare state. Soc Sci Med1997;44(6):87181.

24. Stansfeld SA, Fuhrer R, Shipley MJ. Types of socialsupport as predictors of psychiatric morbidity in a cohortof British Civil Servants (Whitehall II Study). Psychol

Med1998;28(4):88192.

25. Weissman MM, Bland RC, Canino GJ, Faravelli C,Greewald S, Hwu HG, et al. Cross-national epidemiologyof major depression and bipolar disorder. JAMA1996;276(4):293S9.

26. Health and Welfare Canada and Statistics Canada.Thehealth of Canadians: report of the Canada Health Survey.Ottawa: Minister of Supply and Services, 1981.

27. Federal, Provincial and Territorial Advisory Committee onPopulation Health.Statistical report on the health ofCanadians.Ottawa: Health Canada, 1999 Sep.

28. Marmot M, Ryff CD, Bumpass LL, Shipley M, Marks NF.

Social inequalities in health: next questions andconverging evidence.Soc Sci Med1997;44(6):90110.

29. Ross CE, Van Willigen M. Education and the subjectivequality of life.J Health Soc Behav 1997;38(3):27597.

30. Lachman ME, Weaver SL. The sense of control as amoderator of social class differences in health andwell-being.J Pers Soc Psychol1998;74(3):76373.

31. Joubert N, Raeburn J. Mental health promotion: people,power and passion.Int J Mental Health Promotion1998;1:1522.

32. Bradburn NM.The structure of psychological well-being.Chicago: Aldine Publishing, 1969.

33. MacMillan AM. The Health Opinion Survey: techniquefor estimating psychoneurotic and related types of

disorders in communities.Psychol Rep1957;3:325S

39.34. Stansfeld SA, Head J, Marmot MG. Explaining socialclass differences in depression and well-being. Soc

Psychiatry Psychiatr Epidemiol1998;33(1):19. O



25/38

Commentary

Ethical Issues in theUseof ComputerizedDatabases forEpidemiologic andOtherHealthResearch

Wilfreda E Thurston, Michael M Burgess and Carol E Adair

Abstract

Computerization of databases has increased apprehension about loss of privacy. The intent ofthis paper is to facilitate health research that gives proper respect to ethical principles,thereby increasing public comfort and reducing demands for restrictive legislation concerningaccess to databases. We review how computerization has increased the saliency of concerns

and discuss examples of the application of ethical analysis in published database research.Extreme positions notwithstanding, there is general agreement among researchers thatresearch curiosity and the convenience of database research cannot justify the suspension ofmoral concerns about privacy and confidentiality. Public and professional concerns mayaffect policy development; therefore, the methods of ensuring privacy and protectingconfidentiality must be routinely described in research proposals and published reports alongwith the benefits of the research. An important issue requiring further attention is that themoral responsibility to respect privacy increases with the sensitivity of information.

Key words: computers; databases; epidemiology; ethics; guidelines; health care research;linkage; public policy

IntroductionIn this paper we address the principles of ethical

research as they can be applied to research usingcomputerized databases. A database, computerized ornot, is a collection of information on individuals. Ourbelief is that restrictive legislation concerning access todatabases1 should be avoided by respecting the publicsapprehension that an individuals privacy has beendiminished by technology.2 Neither personal knowledgeof misuse of data, demographic variables nor exposure tomedia accounts explains the publics concerns.3

Computerization merely increases the salience of somehistoric concerns.4 However, the nature of the responseof health researchers to these concerns may have a majorimpact on policy development around database research.

Principles of Ethical Research andComputerization

The first principle to be assessed in any research withhuman subjects is that of non-maleficence: avoidingharm to the subjects caused by the research process, the

intervention or procedures being evaluated or used in thestudy, or by the uses made of the data. Health

information that is not secured and is put to another usecan affect employment status and benefits.5

Computerization has raised new security issues,including the ability to transport large quantities of datawithout being physically close to it and to alter recordsinvisibly.6 The speed and storage capacity of computershave resulted in policies to protect privacy.7

A duty to maintain confidentiality and to respectprivacy, another ethical principle, exists as a result of thenature of the relationship within which information isdisclosed. That duty may be stronger when the nature ofthe information itselffor instance, sensitivediagnosesestablishes some reasonable expectation of

confidentiality. A gradual erosion of the concept ofconfidentiality of medical records has been attributed totechnological advances.8

Fidelity to the therapeutic or other relationshipswithin which information is disclosed is an implicit

1999 127

Author References

Wilfreda E Thurston and Carol E Adair, Department of Community Health Sciences, Faculty of Medicine, University of Calgary, Calgary, Alberta

Michael M Burgess, Centre for Applied Ethics, University of British Columbia, Vancouver, British Columbia

Correspondence: Dr WE Thurston, Associate Professor, Department of Community Health Sciences, Faculty of Medicine, University of Calgary, 3330Hospital Drive NW, Calgary, Alberta T2N 4N1;Fax: (403) 270-7307; E-mail: [email protected]


26/38

commitment that the nature of that relationship will berespected. Patients share information in the belief that itis related to their health care, for direct clinical andbilling purposes. Clinicians can be encouraged to addinformation to a file in the interest of future research,regardless of its clinical significance. It is relatively easyto add variables to a

Documents

Suicides Among Canadian Farm Operators