RESEARCH MENTAL HEALTH CPD

Changes over time in means of in : an analysis of mortality data from 1981 to 2018

Li Liu MSc, Colin A. Capaldi PhD, Heather M. Orpana PhD, Mark S. Kaplan DrPH, Lil Tonmyr PhD n Cite as: CMAJ 2021 March 8;193:E331-8. doi: 10.1503/cmaj.202378

ABSTRACT

BACKGROUND: Ongoing surveillance of RESULTS: The age-standardized suicide 2.1% per year from 1981 to 1993 and 5.7% the means of suicide is necessary for mortality rate declined in earlier decades per year from 1993 to 2007, but did not sig- effective prevention. We examined how for both sexes, but did not significantly nificantly change thereafter. mortality rates owing to different means change in recent decades for either sex. of suicide changed in Canada from 1981 The age-standardized rate of suicide by suf- INTERPRETATION: For both sexes, the to 2018. focation increased from 1993 for females rate of suicide by poisoning is decreas- (2.1% per year) and from 1996 for males ing, the rate of suicide by suffocation is METHODS: We obtained data from 1981 (0.4% per year). The age-standardized increasing, and the rate of suicide by to 2018 on suicide deaths of individuals rate of suicide by poisoning decreased firearm has not significantly changed in aged 10 years and older, from the Can­ for females (2.2% per year) and males the last decade. Given the high propor- adian­ Vital Statistics Death Database. (2.1% per year) from 1981 to 2018. The tion of suicide deaths by suffocation, its We used joinpoint regression analysis to age-standardized rate of suicide by firearm increasing rate and the difficulty of examine changes over time in the sui- decreased from 1981 to 2008 (7.4% per restricting the means of suffocation, cide mortality rate for the 3 most com- year) but did not significantly change there- other approaches to mon means of suicide. after for females; for males, it decreased are needed.

he overall suicide mortality rate in Canada is lower than about 3 times higher among males than females in Canada.2,3,6 it was a few decades ago, but continued declines have Higher suicide mortality rates for males have been observed in been absent in recent years and the impact of suicide almost all countries,1 and have been attributed, in part, to more Tremains high.1,2 Suicide was the ninth leading cause of death in lethal means of suicide (e.g., firearms).5,6,12 Suicide-related 2018.3 More than three-quarters of suicide deaths in Canada are a behaviour also varies across the lifespan, with the suicide mor- result of suffocation, poisoning and firearms.4–6 Suffocation is the tality rate tending to be high in middle-aged adults and males predominant suicide method in most countries.4 However, the older than 80 years, but hospital admissions owing to self- prevalence of different means of suicide can vary over time inflicted injuries tending to occur during adolescence (particu- because of changes in the accessibility of certain means.7 Evi- larly among females).2,3,5,6,13 dence suggests that the relative frequency of different means of Canadian data from 2000 to 2009 indicated that suicide by suf- suicide has changed in Canada, but those studies have not focation was more prevalent among individuals aged 15–39 years, included data from more recent years.8,9 Ongoing surveillance of while suicide by firearm was more prevalent among individuals the means of suicide and understanding changes over time are aged 60 years and older.6 Canadian data for 2001 to 2011 suggested essential for designing and implementing prevention programs.10 that rates of suicide by firearm and poisoning declined for males It is especially important to consider sex and age in suicide aged 15 years and older, but were unchanged among females dur- surveillance, because of differences across these sociodemo- ing this period.9 Less is known about whether the means used by graphic characteristics. For instance, females are more likely to age and sex groups have fluctuated over longer periods and in self-report thoughts of suicide and to be admitted to hospital more recent years. We sought to document the means of suicide owing to self-inflicted injuries,5,11 while suicide mortality rates are deaths in Canada in 2018 (the most recent data available at the

© 2021 Joule Inc. or its licensors CMAJ | MARCH 8, 2021 | VOLUME 193 | ISSUE 10 E331 RESEARCH 2000 to2018areavailableonline. the PublicHealthAgencyofCanadaandStatisticsCanada;datafor data for1981to1999throughadata-sharingagreementbetween or piercing (E956, X78–X79); and other (all other codes). We obtained X69); firearmsorexplosives(E955,X72–X75);falls(E957,X80);cutting strangulation orsuffocation(E953,X70);poisoning(E950–E952,X60- We groupedmeansofsuicideintothefollowingcategories: hanging, substantially affect the number of deaths classified as . E332 arms orexplosives; falls;cuttingorpiercing;andother) inCanada,2018. Figure 1:Age-standardizedsuicidemortality rates(andnumberofdeaths)bymeanssuicide(suffocation,hanging orstrangulation;poisoning;fire- time usingtheJoinpointRegressionProgramVersion4.7.0.0. stratified analyses.Weanalyzedchangesinsuicidemortalityover standard population.Wecalculatedcrudemortalityratesforage- population by the direct method, using the 2011 population as the We calculatedage-standardizedsuicidemortalityratesper100 Statistical analysis 10). tion ofDiseasesandRelatedHealthProblems,10thRevision(ICD- codes X60-X84andY87.0fromtheInternationalStatisticalClassifica- Diseases, 9thRevision(ICD-9),and2000to2018usingdiagnostic diagnostic codesE950–E959fromtheInternationalClassificationof Database. Weidentifiedsuicidedeathsfrom1981to1999using aged 10yearsandolder,fromtheCanadianVitalStatisticsDeath We analyzed data from 1981 to 2018 on suicide deaths of individuals Data sources Methods changed formalesandfemalesacrossagegroupssince1981. time ofwriting)andinvestigatehowtheusedifferentmeanshas from StatisticsCanadatocalculatemortalityrates. 14 Comparability studies suggest ICD coding changes did not ComparabilitystudiessuggestICDcodingchangesdidnot

Male Female Both sexes Cutting orpiercing Cutting orpiercing Cutting orpiercing Suocation Suocation Suocation Poisoning Poisoning Poisoning Firearm Firearm Firearm Other Other Other Total Total Total Fall Fall Fall 16 02468 We used population estimates Weusedpopulationestimates 0.1 (21) 0.1 (18) 0.3 (54) 0.4 (67) 0.4 (118) 0.6 (97) 0.6 (206) Age-standardized rateper100 000population(numberofdeaths) 0.8 (248) 0.9 (152) 1.1 (181) CMAJ 1.5 (525) 1.7 (284) 2.1 (684) 2.4 (400) 17 | MARCH 8,2021 3.0 (487) 3.0 (507) 5.6 (931) 000 000 14,15 18 6.3 (2028)

| VOLUME 193 prevalent thanothermeans,forbothmalesandfemales.For 2018. Accordingtothedata,suicidebysuffocationwasmore and the number of deathsfrom each of themeans of suicide in Figure 1presents the age-standardized suicide mortality rates Suicide deathsin2018 Results ses stratifiedbysexandage,weusedap For sex-stratifiedanalyses,weusedap ses, we applied Bonferroni corrections for multiple comparisons. as thecut-offforstatisticalsignificance.Forstratifiedanaly- older adults.Fornonstratifiedanalyses,weusedap represent adolescents,youngadults,middle-agedadultsand 20–34, 35–64,and65yrorolder).Wechosetheseagegroupsto level andstratifiedbysex(femalemale)age(10–19, Canada (i.e.,suffocation,poisoningandfirearm)atthenational changes over time for the 3 most common means of suicide in intervals (CIs)fortheAPCandAAPCestimates.Weinvestigated years) fromthejoinpointmodel.Weobtained95%confidence weights equaltothelengthofsegments(i.e.,number from 1981to2018bytheweightedaverageofAPCs,with on year.Wecomputedtheaverageannualpercentchange(AAPC) where (APC) tointerprettheratechangeperyearusingformula changes (i.e.,joinpoints).Wecalculatedtheannualpercentchange regression function,determiningthenumberandlocationofthese linear regression.Themethodidentifieschangesintheslopeof Joinpoint regressionanalysismodelstimetrends using segmented APC =[exp(β)–1]×100 β is the regression coefficient of log suicide mortality rate istheregressioncoefficientoflogsuicidemortalityrate 10 9.6 (1541) | ISSUE 10 12 11.7 (3809) 14 16 valueof0.025;foranaly- valueof0.005. 18 17.6 (2878) value of 0.05 of0.05 value 20 RESEARCH

2 19,20

21 E333 Our research suggests that 2 ISSUE 10 ISSUE | The increasing rate of suicide by suffocation among some increasing rate of suicide by suffocation among some The We found a decrease in poisoning suicide rates among almost this may reflect rising rates of suicide by suffocation among among suffocation by suicide of rates rising reflect may this females younger than 65 years. Substitution of less lethal means of suicide (e.g., poisoning) with more lethal ones (e.g., suffocation) over time among younger females might explain this increase. Interpretation in Canada over almostIn this study of patterns in suicide deaths 4 decades,suicide by suffocation is gener- the rate of we found that sui- decreasing. The rate of that for poisoning is while increasing, ally cide by firearm had been declining, but we observed no significant changes in recent years among both sexes. These trends tended to prevail across age groups, although there were some exceptions. sociodemographic groups in Canada may partially explain why the overall suicide rate has not continued to decrease in recent years. all sociodemographic groups examined. The overall decrease coin- cides with increases in unintentional poisoning deaths observed during parts of the same time period, which might indicate that Increasing rates of suicide by suffocation have also been observed in other countries; for instance, Norway and the United States. Recent Canadian research reported rising suicide rates among females aged 10–24 and 45–64 years. younger than age 65 years, than age younger by rates of suicide mortality crude 1, Figure S5). recently (Appendix declined significantly poisoning older, the aged 65 years and from 2000, for females By contrast, Among year. per 1.8% by significantly, increased rate crude poisoning recent trends for age groups, the most males in all 1, Figure S6). significantly (Appendix decreased Firearms this study, the age- data examined in Over the 38 years of - rate of suicide by firearm fell by an aver standardized mortality among females, and by an average of 3.1% age of 5.2% per year occurred pri- However, these reductions per year among males. over- the for S7 Figure 1, Appendix (see decades earlier in marily results). From 1981 to 2008, the age- all sex-stratified firearm fell rate of suicide by firearm for females standardized mortality did not change significantly thereafter. by 7.4% per year, but to males decreased significantly from 1981 Similarly, this rate for 2007 (by 5.7% per year), 1993 (by 2.1% per year) and from 1993 to However, differences but did not change significantly thereafter. mortality were in age-specific patterns of firearm suicide Figure S8). The crude observed among males (see Appendix 1, for males and significantly from 1981 rate decreased consistently aged 65 years and older (by 2.4% per year), and decreased across years (except from most of the study period for males aged 35–64 Younger males change occurred). significant when no 2008–2015, rates for suicide also had significantly decreasing crude mortality not change rate did by firearm in earlier decades, but the crude 10–19 years, and it significantly from 2004 for males aged aged 20–34 yearsincreased significantly from 2008 for males (by suicide deaths by fire- 2.4% per year). Given the small number of population after arm among females (i.e., around 0.4 per 100 000 2012), we did not conduct joinpoint analyses of age groups. VOLUME 193 VOLUME | Figure 3, Figure MARCH 8, 2021 | CMAJ , we also observed a pla- a observed also we 2, Figure

Poisoning From 1981, the age-standardized mortality rate of suicide by poi- and females for year per 2.2% of average an by decreased soning 2.1% per year for males (Appendix 1, Figure S4). Among females Suffocation suffocation by suicide of rate mortality age-standardized The recent in females and males both for significantly increased - (Appen differs increase this of magnitude the although decades, dix 1, the increase has been consistent andFigure S1). For females, it males for but year, per 2.1% at 2018 to 1993 from significant 4.1% per year) and byincreased significantly from 1988 to 1996 (by 1, Figure S2, a rising0.4% per year afterward. As shown in Appendix crude mortality rate of suicide by suffocation in more recent per­ iods was apparent across almost all age groups of females, except for those aged 65 years and older, among whom this suicide mor- tality rate declined significantly (1.9% per year) across the 38-year range of this study. The crude rate of suicide by suffocation has been increasing faster for females aged 10–19 years (2.8% per year since 1995) and 20–34 years (3.7% per year since 1981) than for females aged 35–64 years (1.5% per year since 1994). In age-stratified analyses, we observed a significantly increasing crude mortality rate of suicide by suffocation in men aged 35–64 years (0.6% per year but it declined significantly since 1997 (by 1.2% per year)since 1999), for year) per 0.8% (by 1981 since and years 10–19 aged males for males aged 65 years and older (Appendix 1, Figure S3). Change in means of suicide Changes in the suicide mortality rate Changes in the suicide from the joinpoint analyses.Table 1 presents all of the estimates deaths of individuals aged 10 years andA total of 142 343 suicide observedwe Overall, 2018. to 1981 from Canada in occurred older - in Canada’s age-standardized suicide mor a significant decrease no significant but year), per (1.2% 2008 to 1981 from rate tality in plotted As that. after changes teauing when we examined males and females separately. For separately. For teauing when we examined males and females suicide mortality rate females, the overall age-standardized (by 3.3% per year), butdecreased significantly from 1981 to 1991 males, the overall age- did not change significantly after that. For significantly fromstandardized suicide mortality rate decreased significantlynot change but did per year), 0.6% 1999 (by 1981 to in plotted time are over changes Mean-specific that. after www.cmaj.ca/ at (available S1–S8 Figures 1, Appendix and lookup/doi/10.1503/cmaj.202378/tab-related-content). males, the age-standardized mortality rate of suicide by suffoca - by rate of suicide mortality the age-standardized males, than that was 3.2 times higher 100 000 population) tion (9.6 per times higher and 4.0 (3.0 per 100 000 population) for firearms For females, 100 000 population). poisoning (2.4 per than that for suffocation by suicide of rate mortality age-standardized the (3.0 per higher than was almost 1.8 times 100 000 population) times 30 and population) 100 000 per (1.7 poisoning for that firearms (0.1 per 100 000 population).higher than that for Age- mortality rates for males exceeded those standardized suicide of the 6 examined means of suicide. for females across each RESEARCH arms legislationin1977and1991. earlier decadeshavebeenattributed,inpart,topassageoffire- E334 drivers oftheobservedtrendsisspeculative,asanalysesare There areseverallimitationstothisstudy.Discussionofpotential Limitations unintentional. some intentionalpoisoningdeathsarebeingmisclassifiedas the USoverpastdecade, by firearmandtheoverallsuicideratehavebeenincreasingin significant changesinCanadaoverthesametimeperiod. Table 1(part 1of 2):Results from joinpointanalysesof suicidemortality rates inCanadaby means, sex andage, 1981–2018* All means strangulation hanging or Suffocation, Mean 9,22 Female Female Decreasesintherateofsuicidebyfirearm Both Both Male Male Sex group, 10–19 20–34 35–64 10–19 20–34 35–64 Age 65+ 65+ All All All All All All yr 24 theserateshavenotshownany 23 Althoughtherateofsuicide (–1.4 to –1.0) (–4.6 to –2.0) (–1.0 to –0.3) (–5.6 to –1.8) (–8.5 to –3.3) (–2.5 to –1.2) (–4.4 to –0.4) (–1.2 to –0.5) (9.5 to 17.2) (–2.2 to 0.8) (–2.9 to 1.3) 1981–2008 1981–1991 1981–1999 1981–1990 1981–1990 1981–1995 1981–2018 1981–1990 1981–2018 1981–1988 1981–1997 1981–1995 1981–1988 1981–2018 (3.3 to 4.1) (1.5 to 4.6) (3.1 to 5.3) (95% CI) Trend 1 Years –1.2* 13.3* –3.3* –0.6* –3.7* –5.9* –1.9* –2.4* –0.8* APC –0.7 –0.8 3.7* 3.0* 4.2* CMAJ

| MARCH 8,2021 (–9.0 to 40.3) (–3.7 to 30.4) (–1.0 to 0.01) (–2.2 to –0.2) (0.9 to 11.7) (–1.1 to 3.1) (–0.2 to 0.4) (–6.7 to 0.5) 2008–2015 1991–2018 1999–2004 1990–1995 1990–1993 1995–2018 1990–1994 1988–1996 1997–2018 1995–2018 1988–1999 (1.2 to 4.5) (1.9 to 6.4) (4.0 to 6.5) (95% CI) Trend 2 Years –1.2* –3.1 APC 13.0 12.1 –0.5 6.2* 2.8* 4.1* 5.2* 0.9 0.1 – – –

(–10.0 to 1.8) (0.004 to 0.8) (–0.8 to 0.3) 2015–2018 2004–2018 1995–2018 1993–2018 1994–2018 1996–2018 1999–2018 (0.4 to 1.1) (1.6 to 2.5) (0.9 to 2.1) (0.2 to 1.1) | (95% CI) VOLUME 193 Trend 3 Years APC –4.3 –0.3 0.7* 2.1* 1.5* 0.4* 0.6* some suicide deaths as unintentional poisoning, and changes in some suicidedeathsasunintentionalpoisoning,andchangesin ing intentinpoisoningdeaths,thepotentialmisclassificationof findings relatedtopoisoningbecauseofthedifficultyestablish- period of transition. Caution is also warranted when interpreting ably explainsomeofthetrendchangesobservedduring used, from 1981 to 1999 and from 2000 to 2018, which could argu- overfitting. AnotherlimitationisthedifferenteditionsofICD based onstatisticalsignificance,whichcouldhaveresultedin simplification. Moreover,theoptimumnumberofjoinpointswas to describingsegmentsoflineartrends,whichmaybeanover- purely descriptive.Thejoinpointregressionanalyseswerelimited – – – – – – –

(95% CI) Trend 4 | Years ISSUE 10 APC – – – – – – – – – – – – – –

(95% CI) Trend 5 Years APC – – – – – – – – – – – – – –

(95% CI) Trend 6 Years APC – – – – – – – – – – – – – –

(–1.7 to –0.4) (–1.2 to –0.4) (–1.4 to –0.3) (–2.5 to –1.2) (–1.2 to –0.5) (–0.3 to 3.3) (–1.0 to 2.5) (–0.2 to 1.5) (0.3 to 1.9) (5.0 to 8.3) (3.3 to 4.1) (0.3 to 1.6) (0.8 to 1.7) (0.8 to 2) AAPC –1.0* –0.8* –0.8* –1.9* –0.8* 1.1* 6.7* 3.7* 0.9* 1.2* 1.4* 1.5 0.7 0.6 RESEARCH E335 –3.1* –4.0* –4.7* –3.1* –5.2* –3.3* –0.7* –1.8* –3.1* –5.2* –2.1* –1.0* –2.1* –3.1* –3.2* –2.2* –2.1* –2.4* AAPC (–3.7 to –2.4) (–3.7 to (–4.9 to –3.2) (–4.9 to (–6.6 to –2.8) (–6.6 to (–3.6 to –2.6) (–3.6 to (–7.1 to –3.4) (–7.1 to (–3.8 to –2.8) (–3.8 to (–1.2 to –0.3) (–1.2 to (–2.5 to –1.1) (–2.5 to (–4.1 to –2.1) (–4.1 to (–5.9 to –4.5) (–5.9 to (–3.9 to –0.3) (–3.9 to (–3.1 to –1.1) (–3.1 to (–3.5 to –2.6) (–3.5 to (–5.0 to –1.4) (–5.0 to (–3.1 to –1.3) (–3.1 to (–2.9 to –1.4) (–2.9 to (–2.7 to –2.1) (–2.7 to (–2.1 to –0.02) (–2.1 to

– – – – – – – – – – – – – – – – – APC Years –10.4* Trend 6 Trend (95% CI) 2015–2018 (–17.8 to –2.3) (–17.8 to

– – – – – – – – – – – – – – – – – APC –1.5* Years Trend 5 Trend (95% CI) 2000–2015 (–2.3 to –0.7) (–2.3 to

– – – – – – – – – – – – – – – – APC –7.8* –5.9* ISSUE 10 ISSUE Years | Trend 4 Trend (95% CI) 2015–2018 1995–2000 (–13.2 to –2.1) (–13.2 to (–11.0 to –0.6) (–11.0 to

– – – – – – – – – 0.9 4.3 2.4* –1.5 –0.7 APC –1.6* –2.2* Years –15.8* –12.6* Trend 3 Trend VOLUME 193 VOLUME (95% CI) | (0.3 to 4.6) (0.3 to 2008–2015 2008–2018 2004–2018 2007–2018 2005–2018 2005–2018 1990–1995 2015–2018 2015–2018 (–1.2 to 2.9) (–1.2 to (–4.1 to 1.3) (–4.1 to (–1.7 to 0.3) (–1.7 to (–2.4 to –0.9) (–2.4 to (–3.7 to –0.6) (–3.7 to (–1.3 to 10.1) (–1.3 to (–24.9 to –5.5) (–24.9 to (–20.7 to –3.6) (–20.7 to

– – – – – 1.0 0.5 0.3 1.8* –6.5 APC –4.9* –8.7* –5.7* –6.2* –5.5* –7.7* –9.3* Years –11.2* Trend 2 Trend (95% CI) (0.2 to 3.4) (0.2 to 1993–2008 1992–2008 1993–2004 1993–2007 2008–2018 1993–2005 2010–2018 1995–2005 1987–1990 2000–2018 1998–2015 1998–2015 2015–2018 (–5.3 to 7.7) (–5.3 to (–0.3 to 1.4) (–0.3 to (–0.4 to 1.1) (–0.4 to (–5.5 to –4.4) (–5.5 to (–9.7 to –7.7) (–9.7 to (–6.4 to –5.0) (–6.4 to (–7.1 to –5.3) (–7.1 to (–8.2 to –2.6) (–8.2 to (–15.2 to –7.0) (–15.2 to (–10.2 to –5.1) (–10.2 to (–21.4 to 11.2) (–21.4 to (–17.3 to –0.5) (–17.3 to MARCH 8, 2021 |

CMAJ 1.2 –2.3 –0.6 APC –2.4* –1.8* –2.8* –2.1* –7.4* –2.1* –0.7* –0.7* –5.2* –3.6* –2.2* –3.1* –3.2* –2.8* –1.5* Years Trend 1 Trend (95% CI) 1981–2018 1981–1993 1981–1992 1981–1993 1981–1993 1981–2008 1981–1993 1981–2018 1981–2010 1981–1995 1981–2018 1981–1987 1981–2000 1981–1998 1981–2018 1981–2018 1981–1998 1981–2015 (–5.6 to 1.2) (–5.6 to (–2.0 to 0.8) (–2.0 to (–1.7 to 4.2) (–1.7 to (–2.7 to –2.1) (–2.7 to (–2.5 to –1.0) (–2.5 to (–4.5 to –1.0) (–4.5 to (–3.0 to –1.3) (–3.0 to (–8.8 to –6.1) (–8.8 to (–3.0 to –1.3) (–3.0 to (–1.2 to –0.3) (–1.2 to (–1.2 to –0.3) (–1.2 to (–5.9 to –4.5) (–5.9 to (–5.0 to –2.2) (–5.0 to (–2.9 to –1.4) (–2.9 to (–3.5 to –2.6) (–3.5 to (–5.0 to –1.4) (–5.0 to (–3.4 to –2.1) (–3.4 to (–1.7 to –1.2) (–1.7 to yr All All All All All All 65+ 65+ 65+ Age Age 35–64 20–34 10–19 35–64 20–34 10–19 35–64 20–34 10–19 group, group, Sex Male Both Male Both Female Female Mean Poisoning explosives Firearms or Firearms Note: AAPC = average annual percent change between 1981 and 2018, APC = annual percent change within each temporal trend, CI = confidence interval. CI = confidence trend, temporal within each change 1981 and 2018, APC = annual percent between change annual percent = average AAPC Note: significance. *Statistical Table 1 (part 2 of 2): Results from joinpoint analyses of suicide mortality rates in Canada by means, sex and age, 1981–2018* age, and sex means, in Canada by rates suicide mortality of joinpoint analyses from 2): Results 2 of 1 (part Table

RESEARCH E336 sexes, 1981–2018. Figure 3: Age-standardized mortality rates of suicide by suffocation, hanging or strangulation; poisoning; and firearms or explosives in Canada for both Figure 2:Age-standardizedsuicidemortalityratesbysexinCanada,1981–2018.

Age-standardized rate per 100 000 population Age-standardized rate per 100 000 population 0 1 2 3 4 5 6 7 8 1980 10 15 20 25 30 0 5 1980 1984 1984 1988 198 CMAJ 81 | 1992 MARCH 8,2021 992 1996 Female 1996 Year | VOLUME 193 Year 2000 200 Both sexe 02 Poisoning Su ocation, hangingorstrangulation 2004 004 Firearm orexplosive s | ISSUE 10 Ma 2008 le 2008 201 2012 22 01 2016 62 02 2020 0 RESEARCH ​ ​

40. 34:​ 196:​ E337 Epide-

2017; 2005;

2010; BMC Public Natl Vital Stat Rep Int Epidemiol J Int [Epub ahead of print]. doi: Br J Psychiatry What do we know about the The epidemiology and preven- Comparability of cause of death death of of cause Comparability et al. Suicide prevention strategies revis- A, Classification of undetermined et al. Classification of undetermined 2020 Apr. 10 et al. , et al. et , J et al. The gender paradox in suicidal behavior behavior suicidal in paradox gender The K . D, Permutation tests for joinpoint regression Permutation tests for joinpoint regression , DL, Suicide mortality and coronavirus disease Suicide mortality and coronavirus disease Increase in /suffocation in Toxicity of antidepressants: rates of suicide rates of suicide of antidepressants: Toxicity . BGC Villaveces TE. S et al. et al. M, et al. Classification of undetermined deaths by poi- Hawton Hoyert Long-term trends of suicide by choice of method in , EJ, Subaskaran Sabbe O L. SR, HC, , Joiner , JAMA Psychiatry Wasserman , AM, Simkin H J , Cerdá Suicide acts in 8 states: incidence and case fatality rates fatality case and incidence states: in 8 acts Suicide K, IH, H Projected increases in suicide in Canada as a consequence J, Feuer Y. TR. Wilcox McFaull Mehlum Miniño Bollen MP, G, , Lee P, M, Bennewith Hawton Bergen https://proceedings.cmbes.ca/index.php/proceedings/article/ ISSUE 10 ISSUE Stanley Miller | DL , , RN, Hu , Fay RS, G, D Martirosyan K Qin , Frize RS, MA, SP, (accessed 2020 (accessed 2020 www.cdc.gov/nchs/products/databriefs/db373.htm M . Q, H-J, A, 42. 8 -548-X (accessed 2020 Apr. 7). tion of suicide by hanging: a systematic review. 433- Zalsman ited: 10-year systematic review. Lancet Psychiatry 2016;3:646-59. 10.1001/jamapsychiatry.2020.1060. McIntyre of COVID-19. Psychiatry Res 2020;290:113104. Gunnell relative to prescribing and non-fatal overdose. 2019: A perfect storm? 354- Reger Table 13-10-0098-01: Mental health characteristics and suicidal thoughts. thoughts. and suicidal Mental health characteristics Table 13-10-0098-01: www150.statcan.gc.ca/t1/tbl1/en/tv. Canada. Available: Ottawa: Statistics Apr. 6). (accessed 2020 action?pid=1310009801 Schrijvers . Geneva: World Health Organization; Organization; World Health . Geneva: a global imperative suicide: Preventing www.who.int/mental_health/suicide-prevention/world_ 2014. Available: Apr. 6). (accessed 2020 report_2014/en/ Available: view/608 (accessed 2020 Apr. 7). Frize soning: comparison of homogeneous and heterogeneous databases. Ottawa: Ottawa: databases. heterogeneous and homogeneous of comparison soning: Canadian Medical and Biological Engineering Society. CMBES Proc html services/publications/healthy-living/suicide-canada-key-statistics-infographic. ICD-9 for mortality statistics in Canada. Ottawa: Comparability of ICD-10 and www150.statcan.gc.ca/n1/en/catalogue/ Statistics Canada; 2005. Available: 84​ Anderson and its impact on the suicidal process. J Affect Disord 2012;138:19-26. and its impact on the suicidal [infographic]. Ottawa: Public Health Agency ofSuicide in Canada: key statistics 4. Available: www.canada.ca/en/public-health/ Canada; modified 2020 Mar. (accessed 2020 Dec. 9). association between firearm legislation and firearm-related injuries? firearm-related and legislation firearm between association miol Rev 2016;38:140-57. and sex by rates, suicide in differences Urban-rural SC. Curtin K, Pettrone NCHS Data Brief No three leading methods: United States, 2000–2018. Statistics; 2020. Avail- 373. Hyattsville (MD): National Center for Health able: Dec. 1). Esty between ICD-9 and ICD-10: preliminary estimates. between ICD-9 and ICD-10: 1st, by age and sex. Table 17-10-0005-01: Population estimates on July Ottawa: Statistics Canada. Available: www150.statcan.gc.ca/t1/tbl1/en/tv. action?pid=1710000501 (accessed 2020 May 4). Kim 2001;49:1-32. by cause, Chapter XX: External causes of mor- Table 13-10-0156-01: Deaths, to Y89). Ottawa: Statistics Canada. Available: bidity and mortality (V01 (accessed www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1310015601 2020 May 4). . Ottawa: Statistics Canada; 2019. Available: Causes of death, 2017. The Daily. Ottawa: Statistics Canada; 2019. Available: www150.statcan.gc.ca/n1/daily-quotidien/190530/dq190530c-eng.htm (accessed 2020 Apr. 4). Santaella-Tenorio ;90:1885-91. by demographics and method. Am J Public Health 2000 deaths by poisoning: suicidal or unintentional. In: Kyriacou E, Christofides S, Pattichis C, editors. XIV Mediterranean Conference on Medical and Biological Engineering and Computing 2016. IFMBE Proceedings, Vol 57. Cham (Switzer- land): Springer. Hawton -51. with applications to cancer rates. Stat Med 2000;19:335 Health 2016;16:255. Spicer Baker 1969 to 2012. Norway: a joinpoint regression analysis of data from the U.S., 2000–2010. Am J Prev Med 2013;44:146-9. Puzo

31. 29. 30. 28. 12. 10. 11. 26. 15. 13. 14. 24. 25. 18. 16. 17. 23. 22. 27. 21. 19. 20. VOLUME 193 VOLUME | ​ Can J More Global, Global, 9,25,26 28,29 We present We present Bull World Health Health World Bull 27 A range of sui- MARCH 8, 2021 | 30 7,31 CMAJ Methods of suicide: international sui- Additional studies could also could studies Additional 8 Health Promot Chronic Dis Prev Can Suicide in Canada: Is poisoning misclas- Trends in suicide mortality in Canada byTrends in - hospitaliza injury self-inflicted and Suicide et al. et al. et al. M, , E al. et AE, , , J Ring

Suicide prevention by limiting access to methods: a Suicide prevention by limiting access to methods: a . Suicide among children and adolescents in Canada: Suicide among children and adolescents in Canada: [Epub ahead of print]. doi: 10.1177/0706743720940565. MG, C . Draca Rhodes Palladino S , , S S,

Weiss HM, Crane Health at a Glance: Suicide rates — an overview. Cat no 82-624-X. V, . . Collaborators Self-Harm Disease of Burden Global T. 51. McFaull ; JB, McFaull McFaull 2020 July 14 , M , , R Orpana R, R 243-

, M 36: tions in Canada (1979 to 2014/15). 2016; 82-624-x/2012001/article/11696-eng.pdf (accessed 2020 Apr. 7). Florentine review of theory and practice. Soc Sci Med 2010;70:1626-32. Skinner Navaneelan Ottawa: Statistics Canada; 2012. Available: www150.statcan.gc.ca/n1/pub/ trends and sex differences, 1980–2008. CMAJ 2012;184:1029-34. Skinner Table 13-10-0394-01: Leading causes of death, total population, by age group. age by population, total death, of causes Leading 13-10-0394-01: Table Ottawa: Statistics Canada. Available: www150.statcan.gc.ca/t1/tbl1/en/tv. action?pid=1310039401 (accessed 2020 Apr. 6). Ajdacic-Gross sex and age group, 1981 to 2017: a population-based time series analysis. Psychiatry sification an issue? Can J Psychiatry 2016;61:405-12. Skinner cide patterns derived from the WHO mortality database. Organ 2008;86:726-32. regional, and national burden of suicide mortality 1990 to 2016: systematic to 2016: systematic regional, and national burden of suicide mortality 1990 BMJ 2019;364: l94. analysis for the Global Burden of Disease Study 2016. Varin Naghavi . . 7. 8. 6. 9. 3. 4. 1. 2 5

results for 4 age groups, but findings might have varied had differ- might have varied 4 age groups, but findings results for examined. been categories age ent research is needed to understand these potential issues and other these potential issues needed to understand research is that medications psychiatric of use increased (e.g., explanations are attempted). when overdoses are less lethal

References The current analysis of suicide deaths from 1981 to 2018 yields of suicide deaths from 1981 to 2018 yields The current analysis of recent and long-term changes in meansgreater understanding Although the suicide rate is lower than it wasof suicide in Canada. who die by sui- in the 1980s, the significant number of Canadians suicide rate rein- cide each year and the increasing suffocation as a public health andforces the importance of suicide prevention restriction might have safety issue in Canada. Although means and poisoning sui- partially contributed to declines in the firearm suffocation suiciderates, and may be feasible for preventing cide and hos- facilities correctional as (such institutions within deaths of means of suffoca- pitals), the widespread physical accessibility a less realistic suicide tion in the community makes restriction prevention strategy for the general population. Conclusion the use of the undetermined-intent category over time. over category undetermined-intent the of use the examine variations in suicide rates by means over time by other by time over means by rates suicide in variations examine (e.g., marital status). Finally, sociodemographic characteristics available for analysis at the time of writingthe most recent data the we were therefore not able to examine were from 2018 and - disease 2019 on patterns of suicide in Can impact of coronavirus warrants urgent investigation. ada; its potential impact cide prevention strategies, including those aimed at reducing aimed at reducing cide prevention strategies, including those responsible and cognitive accessibility (e.g., safe messaging literacy and aware- media guidelines), improving mental health upstream risk andness of mental health supports, and addressing the worrying trend ofprotective factors may be needed to reverse increasing suffocation suicide mortality rates. RESEARCH E338 revised the manuscript critically for important revised themanuscriptcriticallyforimportant received fromallotherauthors.Alloftheauthors ses andrevisedanalysesbasedonfeedback review process.LiLiuconductedstatisticalanaly­ from allotherauthorsandduringthepeer the manuscriptbasedonfeedbackreceived Contributors: ColinCapaldidraftedandrevised (Tonmyr), CarletonUniversity,Ottawa,Ont. Department ofSociologyandAnthropology versity of California, Los Angeles, Calif.; Luskin SchoolofPublicAffairs(Kaplan),Uni- (Orpana), UniversityofOttawa,Ont.; School of Epidemiology and Public Health Tonmyr), PublicHealthAgencyofCanada; Applied Research(Liu,Capaldi,Orpana, Affiliations: CentreforSurveillanceand This articlehasbeenpeerreviewed. Competing interests:Nonedeclared. CMAJ licenses/by-nc-nd/4.0/ made. See:https://creativecommons.org/ use), andnomodificationsoradaptationsare noncommercial (i.e.,researchoreducational original publicationisproperlycited,theuse reproduction inanymedium,providedthatthe 4.0) licence,whichpermitsuse,distributionand Creative Commons Attribution (CC BY-NC-ND distributed inaccordancewiththetermsof Content licence: Thisisan Open Access article funding forthiswork. None of the other authors received specific Fulbright SpecialistProgram(FSP-P005326). was partiallysupportedbyagrantfromthe Funding: MarkKaplan’sworkonthisstudy accountable forallaspectsofthework. version tobepublishedandagreed intellectual content,gavefinalapprovalofthe | MARCH 8,2021 | VOLUME 193 | ISSUE 10 [email protected] Correspondence to:ColinCapaldi, Accepted: Dec.22,2020 Government ofCanada. necessarily reflect the official positions of the this articlearethoseoftheauthorsanddonot Disclaimer: safe messaging. ada for reviewing a draft of the manuscript for Golden fromthePublicHealthAgencyofCan- Acknowledgement: TheauthorsthankJodie should bemadetoStatisticsCanada. to accessthe19811999mortalitydata doi.org/10.25318/1310015601-eng. Requests 2018 arepubliclyavailableonlineathttps:// Data sharing:Mortalitydatafrom2000to The findings and conclusions in