BRITISH JOURNAL OF PSYCHIATRY (2006), 188, 354^358

Relationship between antidepressant sales the data for Denmark were available up to 2000 and the data for Finland were and secular trends in rates available up to 2003. We were also able to obtain age-specific suicide trend data for and . Antidepressant in the Nordic countries sales data, expressed in terms of defined daily doses (DDDs) for SSRIs and other SVEIN RESELAND, ISABELLE BRAYandBRAY andDAVID GUNNELL antidepressants, were available up to 2003 for all four countries. Data on total anti- depressant and SSRI prescribing were available from 1974 and 1990 respectively for Norway, from 1977 and 1991 Background The effectof recent There are concerns that selective serotonin respectively for Sweden, from 1990 for increasesincreasesin in antidepressant prescribing on reuptake inhibitors (SSRIs) may precipitate both for Denmark and from 1985 and suicidal behaviour in some individuals 1989 respectively for Finland. population suicide rates is uncertain. (Healy, 2003). In the UK, the Medicines We plotted separate graphs for each of Aims ToinvestigateTo investigate the relationship and Healthcare products Regulatory the four countries to enable us to compare Agency (MHRA) recently concluded that the time trends in levels of antidepressant between antidepressant sales and trends a modest increase in the risk of self-harm prescribing with the trends in overall and in suicide rates. in SSRI users could not be ruled out, but gender-specific suicide rates. For Norway that there was too little evidence available and Sweden we also plotted age- and MethodMethod Graphical and quantitative to assess the suicide risk (Medicines and gender-specific suicide rates for three age assessment of trends in suicide and Healthcare products Regulatory Agency, groups (15–24, 25–44 and 4445 years), as antidepressant salesin Norway,Sweden, 2004). In contrast, an analysis of secular data from other countries suggest that time Denmark and Finland. trends in antidepressant prescribing and trends in suicide rates vary with age suicide in the Nordic countries, based (Cantor, 2000). ResultsResults Suiciderates declinedin allfour mainly on data for the period 1990–1996, To estimate the years (with 95% CI) in countries during the1990s, whereas suggests that reductions in the suicide rate which changes in trends in suicide rates coincided with increased antidepressant occurred we used Joinpoint software antidepressant salesincreased by 3- to 4- prescribing (Isacsson, 2000). However, version 2.7 (available from http://srab.cancer. fold.Decreasingsuicide ratesinrates in Sweden time trends in antidepressant prescribing gov/joinpoint). Join-point regression is a and Denmark preceded the rise in anti- and suicide in other countries provide con- form of analysis in which trend data are depressant salesbyover10 years, although flicting evidence for this hypothesis described by a number of contiguous linear the reductions accelerated between1988 (Gunnell & Ashby, 2004). In this paper segments and ‘join points’ where trends we update Isacsson’s analysis with more change (Kim et aletal, 2000). Permutation tests and1990.In Norway,Norway,a a modest but short- recent data on antidepressant sales and are used to determine the minimum number lived declinedeclinein in suicidesuiciderates rates began around suicide rates in the Nordic countries. In of join points required to provide an the time ofthe increase in antidepressant addition, we extend his time series further adequate fit to the data. sales.sales.In In Finland, decreasesdecreasesin in male suicide back in time to investigate whether the reductions in suicide in these countries rates and to a lesser extentin female RESULTSRESULTS coincided with or pre-dated the increases suicide rates began around the time of in antidepressant sales. In all four countries, SSRI sales rose rapidly increased antidepressant sales.In allfour from around 1991–1993 onwards (Fig. 1). countries decreasesdecreasesin in suicide rates In Sweden and Norway there was some appeared to precede the widespread use METHOD evidence of a brief compensatory decline of SSRIs. in the sales of other antidepressants, but Data on suicide rates from the year 1961 this effect was short-lived. The net effect Conclusions We found mixed evidence and levels of antidepressant prescribing on overall antidepressant sales of the for as long a time period as was available thatthatincreasesin increasesin antidepressant sales have increase in the use of SSRIs in the 1990s (in all cases before 1990) were obtained was that by 2000 overall levels of anti- coincided with a reduction inthe number from Statistics Norway, the Norwegian depressant sales in all four countries were of suicidessuicidesin in Nordic countries. Institute of Public Health, Statistics 3- to 4-fold greater than in 1990. Sweden, the National Board of Health Declaration of interest D.G. wa s aaD.G. and Welfare (Sweden), the National Board member ofthe Medicines and Healthcare of Health (Denmark), the Danish Associa- NorwayNorway products Regulatory Agency Expert tion of the Pharmaceutical Industry (Lif), The Norwegian statistics show that after a Statistics Finland and the National Agency steady rise in suicide rates throughout the WorkingGroupontheSafetyofSSRIs.HeWorking Group on the Safety of SSRIs.He for Medicines (Finland). 1970s and 1980s, the rates began to decline was an independent advisor, receiving The data on suicide rates for Norway around 1990, which partly coincided with expenses and an attendance fee. and Sweden were available up to 2002, the period when SSRIs were introduced

354354

Downloaded from https://www.cambridge.org/core. 23 Sep 2021 at 23:26:22, subject to the Cambridge Core terms of use. ANTIDEPRESSANT SALES AND SECULAR TRENDS IN SUICIDE

Fig. 11Fig. Number of v. SSRI and other antidepressant sales in (a) Norway (1961^2002), (b) Sweden (1961^2002), (c) Denmark (1961^2000) and (d) Finland (1961^2003). SSRI, selective serotonin reuptake inhibitor; DDD, defined daily dose.,total; , total; ,male;,male;...... ,female;&&, DDD (SSRI); xx, DDD (other antidepressant).

and sales of antidepressants increased Sweden decreases in suicide rates continued in markedly (Fig. 1a). However, after 3 years men and women aged over 25 years of decline (1992–1994), suicide rates then In Sweden, in contrast to Norway, suicide throughout the period of increased SSRI stabilised, despite large increases in SSRI rates have been declining steadily since sales, any such declines were less markedmarked sales (Fig. 1a). In age-specific analyses the 1970s, many years before the rise or absent in individuals aged 15–2415–24 (Fig. 2a and b) it is clear that this reduction in SSRI sales in the early 1990s (Fig. (Fig. 2c and d). As in Norway, there was and subsequent levelling out of previously 1b), but the decline (again in contrast a reduction in the sales of rising suicide rates occurred in all age/gen- to Norway) also continued after the other antidepressants in the years after der categories except for female introduction of SSRIs in Sweden. It is the introduction of SSRIs in Sweden individuals aged 15–24 years. noteworthy that although age-specific (Fig. 1b).(Fig.1b).

Fig. 22Fig. Number of suicides (per(per100 100 000 inhabitants, smoothed by 3-year moving averages) vv. SSRI sales by age group in Norway for (a) males and (b) females, and in Sweden for (c) males and (d) females, for the period1961^2002. SSRI, selective serotonin reuptake inhibitor; DDD, defined daily dose. &&, DDD (SSRI);, 5545 years45years (3-year moving average);, 25 ^ 44 years (3-year moving average); ...... , 15 ^24 years (3-year moving average).

355

Downloaded from https://www.cambridge.org/core. 23 Sep 2021 at 23:26:22, subject to the Cambridge Core terms of use. RESELAND ET AL

DenmarkDenmark the best-fitting model included the follow- effects were most marked in men, despite The Danish statistics (Fig. 1c) clearly show ing three join points: 1977 (95% CI the fact that women are the greatest consu- a steady rise in male and female suicide 1972–1981), after which suicide rates rose; mers of antidepressants. The decline may rates throughout the 1960s and 1970s, with 1983 (95% CI 1979–1987), when rates be partly explained by Finland’s vigorous a marked peak around 1980. Subsequently, rose again, and 1990 (95% CI 1988– national programme, suicide rates declined in both genders, 1993), when theybegan to fall. In each which was initiated around this time around 10 years before the introduction of the final (best-fitting)(best-fitting) models, PP-values-values (Annals of Internal Medicine, 2004), of SSRIs in Denmark and the associated provide strong evidence (evidence(PP<0.005) for a dif- although more recent suicide prevention in- increase in levels of antidepressant use. ference in slopesat the most recent join itiatives in Norway (in 1992) and Sweden The decline in suicide rates has continued point (1988–1990). (in 1995) do not appear to have influenced over the period of increased SSRI sales. trends in suicide in those countries (Calgary However, the rates of decline appeared Centre for Suicide Prevention, 2004). to increase somewhat in the 1990s, DISCUSSION Statistical analysis suggests that in all particularly in women. four countries a decline (or an acceleration Trends in suicide rates in Nordic countries of a pre-existing decline) in suicide rates in the 10-year time period before and after began around 1988–1990. This period FinlandFinland the introduction of SSRIs provide mixed pre-dates the introduction of SSRIs (and In Finland, after increases in suicide rates in evidence that increased sales have resulted certainly their widespread use), although the 1960s, 1970s and 1980s, declines in the in a reduction in suicides. In Norway, the the 95% confidence intervals for the year overall and male suicide rates, and to a period when the greatest increases in anti- in which suicide rates began to decline lesser extent in the female suicide rates, depressant sales occurred was characterised extend to 1999 for Norway, to 1991 for coincided with the introduction and by relatively stable suicide rates. The Sweden, to 1995 for Denmark and to increased sale of SSRIs (Fig. 1d). decline in suicide rates in Denmark and 1993 for Finland. These results are derived Sweden pre-dated the introduction of the from models that fit linear segments to non- SSRIs by more than 10 years, and the linear data. Although the positions of the Join-point analysis suicide rates in these countries continued join points of these segments provide useful For Norway, Sweden and Denmark the to decrease thereafter. The strongest estimates of the years in which trends in most appropriate model of secular trends evidence of an association between suicide rates changed significantly, they in suicide rates included two join points increases in antidepressant sales and a represent a simplification of the observed whereas for Finland three join points decrease in suicide rates was seen in temporal trends, and should therefore be provided the best fit to the data. The Finland, where reductions in suicide rates treated with caution. estimated join points and their 95% confi- coincided with the introduction and Our findings contrast with a previous dence intervals are shown in Table 1 for increasing use of SSRIs. However, such assessment of trends in antidepressant models with one, two and three join points, Ta b l e 11Tab Results of join-point analysis of suicide rates in the Nordic countries from1961to 2000^200211 and the model of best fit is denoted in bold type.type. PP-values for a test of the difference in slopes at each join point in the fitted model Number of Join point (95% CI) PP22 Join point (95% CI) PP Join point (95% CI) PP are also shown. join pointspointsjoin For the best-fitting model for Norway, the 2 years in which changes in trends Norway occurred were 1967 (95% CI 1963– 33 19671967(1962^1990) 0.04 19881988(1963^1998) 550.01 19951995(1986^2001) 0.04 1991), when suicide rates began to rise, 22 1967(1963^1991) 0.091988 (1985^1999) 550.010.01 and 1988 (95% CI 1985–1999), when sui- 11 19881988(1986^1989) 550.010.01 cide rates began to fall. A third change, SwedenSweden namely the levelling out of rates described 31970 (1962^1971)(1962^1971)33 1971(1966^1992) 33 1988(1982^2001) 550.01 above, was also identified in 1995 (95% 22 1968(1965^1970) 550.01 1988(1985^1991) 550.01 CI 1986–2001) in a model with three join 111969 (1967^1972) 550.01 points, although this is not the model of best fit. Similarly, for Sweden 2 years were Denmark identified, namely 1968 (95% CI 1965– 319793 1979(1976^1980) 33 1980(1979^1982) 33 1989(1985^1994) 550.01 1970), when rates began to decline and 22 1982(1978^1986) 0.011989 (1985^1995) 550.01 1988 (95% CI 1985–1991), when there 11 19851985(1983^1988) 550.01 was an accelerated rate of decline as described previously by Carlsten et aletal Finland (2001). The two join points that were 33 19771977(1972^1981) 0.041983 (1979^1987) 0.011990 (1988^1993) 550.01 identified in the Danish suicide data were 22 19861986(1962^1993) 0.45 19901990(1988^2002) 0.05 1982 (95% CI 1978–1986) and 1989 11 19921992(1990^1994) 550.01 (95% CI 1985–1995), the latter year corre- 1. Bold typeface denotes best fit. sponding to an increase in the rate of de- 2.2. PP values are for a difference in slope at each join point. cline that began around 1982. In Finland, 3. PP values not calculated because the information matrix is singular.

356356

Downloaded from https://www.cambridge.org/core. 23 Sep 2021 at 23:26:22, subject to the Cambridge Core terms of use. ANTIDEPRESSANT SALES AND SECULAR TRENDS IN SUICIDE

prescribing and suicide rates in the Nordic of non-fatal suicidal behaviour in children Implications countries (Isacsson, 2000). The limited and adults. As most of these trials were of Although there is broad consensus about period covered by the suicide data that short duration, it is uncertain whether such the effectiveness of SSRIs in treating were presented in Isacsson’s study meant increased risks may be offset by a longer- depression, the evidence that the decline that it was impossible to distinguish term reduction in risk among those taking in suicide rates which was seen in Norway, between short-term and longer-term trends antidepressants for the recommended Sweden, Denmark and Finland in the 1990s in suicide rate (Isacsson, 2000). Our period of up to 6 months. There were in- resulted from increased antidepressant analysis suggests that the favourable trends sufficient numbers of participants recruited prescribing is not clear-cut. A more detailed in two of the four countries studied reflect a to the trials to allow investigation of any understanding of the factors that have longer-term favourable trend in suicide beneficial or adverse effects of SSRIs on contributed to recent declines in suicide rates. However, we did find evidence that suicide deaths (Gunnell et aletal, 2005), high- rates in the Nordic countries is required. declining rates of suicide in Sweden acceler- lighting the importance of using observa- Elucidation of these factors will help to ated around 1988,1988, shortly before the wide- tional studies to investigate this issue. It is inform the development of evidence-based spread use of SSRIs in the 1990s. This noteworthy that observational studies suicide prevention policies. finding is consistent with another assess- provide no strong evidence that SSRIs differ ment of the association between anti- from tricyclic antidepressants with regard ACKNOWLEDGEMENTS depressant prescribing and suicide rates in to the risk of suicidal behaviour (Martinez Sweden up to 1997 which reported that et aletal, 2005).,2005). We thank the following for providing data from the the introduction of SSRIs coincided with national statistics: Wibeke Djume (Statistics an increased rate of decline in suicide Norway); Elisabeth Eriksen (Norwegian Institute of rates (Carlsten et aletal, 2001), but that this Limitations Public Health); Birgitta Chisena (Statistics Sweden); change pre-dated the large increases in Our analysis has several limitations. First, Andrejs Leimanis (National Board of Health and antidepressant sales. we employed an ecological study design Welfare, Sweden); Laila Christensen and Lone Mortensen (National Board of Health, Denmark); using national sources of prescribing and Jrgen Clausen (Danish Association of the Pharma- Prescribing trends and suicide suicide data. We have not investigated the ceutical Industry, Lif); Vibeke Dahl Jensen (Danish rates in other countries influence of antidepressant treatment on Medicines Agency); Mauno Huohvanainen (Statistics suicide risk. Furthermore, because the pre- Finland); and Tinna Voipio (National Agency for Evidence from other ecological investiga- scribing data are sales data rather than data Medicines, Finland). We also thank Margaret May tions is mixed. Studies in the USA (Olfson et on person-based consumption of anti- for statistical advice. alal, 2003; Grunebaum et aletal, 2004), Australia depressants, the number of individuals (Hall(Hall et aletal, 2003) and Hungary (Rihmer etet who took antidepressants will be over- REFERENCES alal, 2001) report trends which suggest a estimated. Second, we were unable to favourable effect of antidepressants on determine the extent to which increases in Annals of Internal Medicine (2004) Finland pioneers suicide rates, whereas findings from Iceland international suicide prevention. Annals of Internal sales were a result of growing numbers of (Helgason(Helgason et aletal, 2004), Italy (Barbui et aletal,, MedicineMedicine,, 14 0, 853^856. each of the following: people who had been 1999) and England (Gunnell & Ashby, newly prescribed an SSRI; long-term users; Barbui, C., Campomori, A., D’Avanzo, B., et aletal 2004) do not support this. The main limita- (1999)(1999) Antidepressant drug use in Italy since the or a combination of the two. Third, we tion of those analyses, as well as those in introduction of SSRIs: national trends, regional have not taken into account other influ- differences and impact on suicide rates. Social Psychiatry the present study, is that there are many ences on temporal trends in suicide rates & Psychiatric Epidemiology,, 3434,152^156.,152^156. complex influences on national suicide (e.g. changes in the levels of unemploy- Calgary Centre for Suicide Prevention (2004) AA rates, including the availability of lethal ment, divorce or substance misuse) (Gun- Summary of National, State and Provincial Strategies for the methods of suicide, changing social and Prevention of Suicide. Calgary: Centre for Suicide nellnell et aletal, 2003). Fourth, we were unable economic conditions, changing levels of Prevention.Prevention. to obtain age- and gender-specific prescrib- http://www.suicideinfo.ca/csp/assets/Summaryof substance misuse and birth cohort effects ing data for the four countries, so we were NationalStateandProvStrategies.pdf (Gunnell(Gunnell et aletal, 2003). Distinguishing be- unable to investigate whether age-specific tween the discrete effects of changing levels Cantor,C. H. (2000) Suicide in the Western world. In trends in suicide rates mirrored those for The International Handbook of Suicide and Attempted of antidepressant prescribing and those of prescribing. However, analyses from other Suicide (eds K.K.Hawton Hawton & K. van Heeringen, pp. 9^28. these other influences on patterns of suicide Chichester: John Wiley & Sons. countries suggest that antidepressant pre- is challenging, as many of the factors are scribing increased by 2-fold or more in all Carlsten, A.,Waern, M., Ekedahl, A., et aletal (2001)(2001) strongly correlated with each other. age/gender categories in the 1990s Antidepressant medication and suicide in Sweden. Pharmacoepidemiology and Drug Safety,, 1010,525^530. (Middleton(Middleton et aletal, 2001; Hall et aletal, 2003),,2003), Person-based analyses which indicates that our use of all-age data Danish Medicines Agency (2005) Forbruget af of the association between Antidepressiva i Danmark i den Primre Sundhedssektor i will not seriously bias our interpretation Perioden 1994^2003.Copenhagen:. Copenhagen: Danish Medicines antidepressants and suicidal of the overall trends within specific age/ Agency. http://www.laegemiddelstyrelsen.dk/ behaviourbehaviour gender categories. Finally, our analysis publikationer/netpub/forbrugsanalyser/antidepressiva/ html/entire___publication.htm. Recent meta-analyses of placebo-controlled was restricted to four countries. As high- randomised trials of SSRIs (Whittington(Whittington etet lighted above, there is mixed evidence from Fergusson, D., Doucette, S., Cranley Glass, K., et al alal, 2004; Fergusson et aletal, 2005; Gunnell other countries concerning the effect of (2005)(2005) Association between suicide attempts and selective serotonin reuptake inhibitors: systematic et aletal, 2005) have provided evidence that increased levels of antidepressant prescrib- review of randomised controlled trials. BMJBMJ,, 330330,, SSRIs are associated with an increased risk ing on suicide rates. 396^399

357357

Downloaded from https://www.cambridge.org/core. 23 Sep 2021 at 23:26:22, subject to the Cambridge Core terms of use. RESELAND ET AL

Grunebaum, M. F., Ellis, S. P., Li, S., et aletal (2004)(2004) Antidepressants and suicide risk in the United States, 1985^1999.Journal of Clinical Psychiatry,, 65,1456^1462,1456^1462 CLINICAL IMPLICATIONS

Gunnell, D. & Ashby, D. (2004) Antidepressants and suicide: what is the balance of benefit and harm? BMJBMJ,, && The evidence that increased antidepressant sales have had a beneficial effect on 329329, 34^38.,34^38. population suicide rates is equivocal. Gunnell, D., Middleton, N.,Whitley, E., et aletal (2003) && Whyaresuicideratesrisinginyoungmenbutfallingin Explanations for the recent favourable trends in suicide rates in Norway, Sweden, the elderly? A time-series analysis of trends in England Denmark and Finland should be clarified.These may help to inform national suicide and Wales,Wales,1950^1998. 1950^1998. Social Science and Medicine,, 5757,, 595^611.595^611. prevention policies. Gunnell, D., Saperia, J. & Ashby, D. (2005) SelectiveSelective && Frequent and careful patient monitoring during the early stages of treatment with serotonin reuptake inhibitors (SSRIs) and suicide in adults: a meta-analysis of drug company data from antidepressants is important. placebo-controlled randomised controlled trials submitted to the MHRA’s safety review. BMJBMJ,, 330330,, LIMITATIONS 385^388. && Hall, W. D., Mant, A., Mitchell, P. B., et aletal (2003)(2003) This is an ecological study.We have not investigated the effect of antidepressant Association between antidepressant prescribing and treatment on suicide risk. ,1991^2000: trend analysis. BMJBMJ,, 326326,, 1008^1011. && We have not taken into account other influences on temporal trends in suicide Healy, D. (2003) Lines of evidence on the risks of rates (e.g. changes in the levels of unemployment, divorce or substance misuse). suicide with selective serotonin reuptake inhibitors. Psychotherapy and Psychosomatics,, 72, 71^79.,71^79. && We were unable to distinguish the extent to which increases in antidepressant Helgason,Helgason,T., T., ToTomasson,¤masson, H. & Zoega,T.Zoega, T. (2004) sales were a result of growing numbers of people who had been newly prescribed a Antidepressants and public health in Iceland.Time series selective serotonin reuptake inhibitor, long-term users, or a combination of the two. analysis of national data. British Journal of Psychiatry,, 184184,, 157157^162. ^ 162.

Isacsson, G. (2000) Suicide prevention ^ a medical breakthrough? Acta Psychiatrica Scandinavica,, 102,, SVEINSVEINRESELAND,KD-GConsulting,Hosle,Norway;ISABELLEBRAY,MSc,PhD,DAVIDGUNNELL,MB,ChB, RESELAND,KD-G Consulting, Hosle, Norway; ISABELLE BRAY, MSc, PhD, DAVID GUNNELL, MB, ChB, 113^117.113 ^ 117. MSc, PhD,FFPHM, Department of Social Medicine,University of Bristol, Bristol,UK Kim, H. J., Fay, M. P., Feuer, E. J., et aletal (2000) Permutation tests for joinpoint regression with CorrespCorrespondence:ondence: Svein Reseland,Hosletoppen 56,1362 Hosle,Norway.Tel/fax: +47 6714 5899; applications to cancer rates. Statistics in Medicine,, 1919,, e-mail: sreselan@@online.noonline.no 335^351.

Martinez, C., Rietbrock, S.,Wise, L., et aletal (2005)(2005) (First received 15 January 2005, final revision 12 April 2005, accepted 16 May 2005) Antidepressant treatment and the risk of fatal and non- fatal self-harm in first-episode depression^ a nested case^controlcase ^control study. BMJBMJ,, 330330, 389^393.,389^393. Rihmer, Z., Belso, N. & Kalmar, S. (2001) Medicines and Healthcare products Regulatory 1975^1998. Journal of Public Health Medicine,, 2323,, 262^267. Antidepressants and suicide prevention in Hungary. ActaActa Agency (2004) RReport of the Committee on Safety of Psychiatrica Scandinavica,, 103103, 238^239. Medicines Expert Working Group on the Safety of Selective Serotonin Reuptake Inhibitors. London:London:TSO TSO (The et aletal et aletal Stationery Office). Olfson, M., Shaffer, D., Marcus, S. C., (2003)(2003) Whittington, C. J., Kendall,T., Fonagy, P., (2004)(2004) Relationship between antidepressant medication Selective serotonin reuptake inhibitors in childhood Middleton, N., Gunnell, D.,Whitley, E., et aletal (2001)(2001) treatment and suicide in adolescents. Archives of General depression: systematic review of published versus Secular trends in antidepressant prescribing in the UK, PsychiatryPsychiatry,, 60, 978^982.,978^982. unpublished data. Lancet,, 363363,1341^1345.

358358

Downloaded from https://www.cambridge.org/core. 23 Sep 2021 at 23:26:22, subject to the Cambridge Core terms of use.