A systematic review of eating disorder prevalencein the Nordic countries: 1994–2016

(1)

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=rnpy20

Nordic Psychology

ISSN: 1901-2276 (Print) 1904-0016 (Online) Journal homepage: http://www.tandfonline.com/loi/rnpy20

A systematic review of eating disorder prevalence in the Nordic countries: 1994–2016

Camilla Lindvall Dahlgren, Kristin Stedal & Line Wisting

To cite this article: Camilla Lindvall Dahlgren, Kristin Stedal & Line Wisting (2017): A systematic review of eating disorder prevalence in the Nordic countries: 1994–2016, Nordic Psychology, DOI:

10.1080/19012276.2017.1410071

To link to this article: https://doi.org/10.1080/19012276.2017.1410071

Published online: 11 Dec 2017.

Submit your article to this journal

View related articles

View Crossmark data

(2)

https://doi.org/10.1080/19012276.2017.1410071

ARTICLE

This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDeriva- tives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and

A systematic review of eating disorder

prevalence in the Nordic countries: 1994–2016

CAMILLA LINDVALL DAHLGREN , KRISTIN STEDAL & LINE WISTING

Correspondence address: Camilla Lindvall Dahlgren, Regional Department for Eating Disorders, Division of Mental Health and Addiction, Oslo University Hospital, Ullevål HF, Oslo, Norway. Email: [email protected]

Abstract

Objective: The objective of this study was to systematically review the literature on eating disorder prevalence in the Nordic countries (i.e. Denmark, Finland, Iceland, Norway and Sweden) between 1994 and 2016. Method: Three online databases, PubMed, PsycINFO and SveMed+, searches were conducted, targeting articles on prevalence, incidence and epidemiology of eating disorders (EDs). The review was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, and was limited to Internation- al Classification of Diseases 10th revision (ICD-10) and the Diagnostic and Statistical Manual of Mental Disorders 4th Edition (DSM-IV) ED diagnoses published between 1994 and 2016. Results:

A total of 19 studies fulfilled inclusion criteria and were included in the study. Results revealed that anorexia nervosa (AN) and binge eating disorder (BED) prevalence ranges in females are lower in the Nordic countries compared to Europe in general, and slightly higher for bulimia nervosa (BN). Discussion: The current review shows that there is an imperative need for studies investigating the prevalence of EDs in the Nordic countries by employing a methodology which allows for a formal diagnosis. The notable lack of two-stage designs in the studies reviewed renders concern regarding the validity of the obtained prevalence rates. It is essential that future studies employ standardized methodologies as these will allow for comparison across geographic regions, genders and ages, facilitating mapping of prevalence rates over time.

Keywords: review, prevalence, eating disorders, Nordic countries, DSM-IV, ICD-10, DSM-5

Introduction

Eating disorders (EDs) are serious mental illnesses characterized by irregular eating patterns and severe distress concerning body shape and weight. Anorexia nervosa (AN) is associated with a high risk of premature death (Kask et al., 2016) and is recognized by persistent efforts to lose weight despite being severely underweight, and significant concerns regarding body image.

Bulimia nervosa (BN) is characterized by episodes of binge eating followed by compensatory behaviours. Historically, AN and BN have been the most widely studied EDs. However, eating disorders not otherwise specified (EDNOS) was the most frequently reported ED using the fourth version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (American

1Regional Department for Eating Disorders, Division of Mental Health and Addiction, Oslo University Hospital, Ullevål HF, Postboks 4950 Nydalen, 0424 Oslo, Norway

Downloaded by [University of Oslo] at 03:33 12 December 2017

(3)

Psychiatric Association, 1994). Recently, the DSM-IV was replaced by the DSM-5 (American Psy- chiatric Association, 2013) which resulted in a number of changes to the diagnostic ED categories. A few of the most substantial revisions were to lower the diagnostic threshold for AN and BN, and to introduce binge eating disorder (BED) as an independent ED diagnosis. In addition, the diagnostic categories pica, rumination disorder and avoidant/restrictive food intake disorder (ARFID) were added to the DSM-5 “Feeding and Eating Disorders”. The changes in the DSM-5 were introduced to help minimize broad generalizations of EDs, and is expected to have important implications for ED prevalence rates in the years to come. Some of these implication have already been confirmed by studies investigating the prevalence of EDNOS (Machado, Gon- calves, & Hoek, 2013) as well as AN and BN (Mustelin et al., 2016; Smink, van Hoeken, Oldehinkel,

& Hoek, 2014; Trace et al., 2012).

For many years, EDs have been claimed to be represented predominantly in Western countries, and although there is evidence of potential biological markers (Cerasa et al., 2015) it is well known that cultural beliefs and attitudes influence the development of EDs (Makino, Tsuboi,

& Dennerstein, 2004) BN, in particular, seems to be a culture-bound syndrome (Keel & Klump, 2003). The influence of sociocultural factors on EDs was recently emphasized in a study by Nakai and colleagues (Nakai, Nin, & Noma, 2014) where a conflict between the influence of western culture and traditional Japanese values, as well as a growing media-oriented culture was put forth as an explanation for the dramatic increase in ED prevalence rates in Japan. Because EDs are influenced by sociocultural factors, both on a local and global level (Becker & Fay, 2006) it is important to obtain prevalence rates within targeted geographical areas. Unfortunately, population-based prevalence studies are both time-consuming and financially demanding to imple- ment. In addition, they will often be limited by small sample sizes and selection bias. Therefore, systematic reviews of prevalence studies can be particularly helpful. By synthesizing studies assessing prevalence rates across countries with similar standards of living, socio-economic status and culture, it is possible to obtain better estimates of the number of individuals affected by the illness in question. Lately, there has been a marked increase in studies reviewing the prevalence of EDs across geographic regions including Africa (van Hoeken, Burns, & Hoek, 2016), Asia (Pike & Dunne, 2015) and Europe (Keski-Rahkonen & Mustelin, 2016). In their recent European review, Keski-Rahkonen and Mustelin demonstrate great variability in prevalence rates across European countries. However, there are some limitations to the study methodology as the literature was not reviewed through a systematic process using specific review criteria. In addition, only studies published in 2015 and the first half of 2016 were included, leaving out a substantial proportion of studies of relevance. Finally, comparisons between countries included in the review are likely to be obscured by socio-economic differences. The authors compare prevalence rates from Finnish and German communities, as well as students from Serbia and the Canary Islands. Poverty rates and income inequalities in the Nordic countries have been reported to be lower, compared to some Southern European countries (Andersen et al., 2007), and correlations between socio-economic conditions and mental illnesses are well documented (Adler & Ostrove, 1999; Everson, Maty, Lynch, & Kaplan, 2002; Hudson, 2005). In sum, it is necessary to obtaining local prevalence rates to minimize the effects of sociocultural and economic factors. The Nordic region is one local area of interest.

The Nordic region consists of some of the most affluent countries in the world, including Den- mark, Finland, Iceland, Norway and Sweden. In addition to the close geographic proximity, the countries are also similar in terms of standard of living, socio-economic status and culture. The

Downloaded by [University of Oslo] at 03:33 12 December 2017

(4)

“Nordic Model” promotes health and welfare policies which aim to ensure good public health, provide a high level of social security and diminish social inequalities (Andersen et al., 2007). To the authors’ knowledge, no studies have previously investigated the prevalence of EDs in the Nordic region. A systematic review of the prevalence of EDs in the Nordic region could provide a valuable point of reference when planning ED treatment, allocating resources, evaluating out- comes and to be able to identify an aggregated estimate of the economic burden of EDs in this region. Further, a systematic review of the prevalence of EDs in the Nordic region during the DSM-IV period may help facilitate rate comparisons between existing and future studies, and has the potential of informing the development of new prevalence studies. The aim of this study was thus to systematically review the literature on ED prevalence in the Nordic countries between 1994 and 2016. The review was conducted in accordance with Preferred Reporting Items for Sys- tematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2015).

Methods

Search strategy

The literature was reviewed in November 2016 using the three databases PubMed, PsycINFO and SveMed+. A librarian at the Medical Library, Oslo University Hospital (OUS), developed this search strategy. To update the initial search, an additional search was performed in November 2016. All publication titles were screened initially by the first author. Eligibility was established by reading abstracts and full texts. A detailed search strategy is available in Appendix 1.

Eligibility criteria

Publications that examined prevalence, incidence and epidemiology of EDs in the Nordic countries (Denmark, Finland, Iceland, Norway and Sweden) were targeted, and were reviewed according to the following criteria:

(1) Articles published between January 1994 and November 2016 (initial search) (2) Articles that were written in English or had an available published English translation (3) Articles presenting ED prevalence rates using the ICD-10 or DSM-IV

(4) Articles that described methods used to assess prevalence

No exclusions for age or gender were applied. Both cross-sectional and longitudinal studies were included.

Exclusion criteria

Books, book chapters, editorials, commentaries, reviews, errata, prevalence studies in clinical samples (e.g. psychiatric outpatients), studies presenting data where diagnostic measures were unclear and articles using diagnostic systems preceding DSM-IV (e.g. DSM-III) or ICD-10 (World Health Organisation, 1992) (e.g. ICD-8 and ICD-9) or a combination of diagnostic systems (e.g.

DSM-III and DSM-IV) were omitted. All omitted review articles were screened for additional references. Inclusions and exclusions were tracked and registered (see Figure 1).

Title, abstract and article selection

Publication titles in studies sourced through the search strategy outlined in Appendix 1 were initially reviewed by the first author (CLD). Where titles were ambiguous, abstracts were also screened at this stage. Full texts were retrieved and assessed for eligibility by the three authors

Downloaded by [University of Oslo] at 03:33 12 December 2017

(5)

(CLD, KS & LW). Full texts were divided between the three authors, which, after having been examined were labelled “include”, “exclude” or “inconclusive”. Inconclusive texts were discussed.

Agreement was reached for all inconclusive texts. None of the reviewers were blind to the authors of the texts, nor their affiliations.

Figure 1. The PRISMA flow diagram describing the literature screening process.

Downloaded by [University of Oslo] at 03:33 12 December 2017

(6)

Results

The PRISMA literature search: 1994–2016

The PubMed, PsycINFO and SveMed+ searches produced 389 titles, which, after duplicates had been removed, were reduced to 347. In the title screening process, publications with titles corresponding to obviously ineligible texts (e.g. “Gastronomy and diet in 18-month-old children with congenital heart defects”) were initially removed, whereas equivocal (e.g. “Eating Disorders – an overview”) and plausible titles (e.g. “Epidemiology of eating disorders in children and adolescents) were retained. Consequently, 233 out of the 347 publications were omitted based on the lack of title relevance. 114 titles were retained. Retained titles were equally divided between the three authors (CLD, KS & LW) who first screened potential eligibility using inclusion and exclusion criteria based on abstracts, and subsequently, established eligibility by reading full texts.

An additional 51 texts were excluded in the abstract extraction phase, and after having read full texts, another 44 articles were omitted. No additional studies were found through the screening of review articles. All in all, 19 studies were included in the review. A more detailed overview of the screening process is presented in Figure 1.

Search summary

19 studies met inclusion criteria and were included in the review. No studies meeting inclusion criteria were published prior to 1999. Table 1 summarizes all included studies between 1999 and 2016. Studies are grouped based on study design (i.e. 2-stage design studies, interviews only and self-reports) and listed in chronological order. One study did not report study design, and was listed under the table heading “OTHER” (Goodman, Heshmati, Malki, & Koupil, 2014). Assessments of the following full-threshold diagnoses were reported: ED, AN, BN, BED and EDNOS. Although not recognized as its own disorder in the DSM-IV, prevalence of BED was assessed in the majority of the included studies. As it now is, i.e. in the DSM-5, considered a distinct diagnostic category, BED was included in the current review for comparative purposes.

Sample characteristics Demographics

Nine (47.4%) of the included studies were Finnish, six (31.6%) were Swedish, three (15.8%) were Norwegian, and one (5.3%) study reported prevalence rates in Iceland. No studies from Denmark fulfilled inclusion criteria. Out of the nine Finnish studies, four were based on the same study sample (Keski-Rahkonen et al., 2009, 2007, 2006; Mustelin et al., 2016). This was also true for two of the Swedish studies (Ghaderi & Scott, 1999, 2001). A total of ten studies reported prevalence for females only, whereas one study reported rates for males only. The eight remaining studies reported prevalence rates for both females and males. Ages ranged from 12 to 55, with 13 the vast majority (68.4%) of studies reporting ED prevalence rates in individuals 18 years and above.

Five studies (26.3%) reported rates in adolescents and young adults ranging in age from 14 to 20, and the final study had a mixed sample including children, adolescents and adults (ages 12–36).

Sample size and setting

Sample sizes ranged from 595 to 2 015 862, with the majority of studies (N = 15) reporting prevalence rates from samples ranging from 500 to 4500 participants. Three studies presented data

Downloaded by [University of Oslo] at 03:33 12 December 2017

(7)

Table 1. Overview of studies assessing eating disorder prevalence in the Nordic countries 1994–2016. Studies are grouped by study design and listed in chronological order. Author (Year)N (Total)CountryGender (N or %)Age range mean (SD)Sample characteristicsAssessmentPrevalence typePrevalence rate (%) 2-STAGE DESIGN Mustelin et al. (2016)a2825Finland♀22-27Sample drawn from FinnTwin16 - A nation- wide population based cohort Screening: Self-report questionnaire developed for the study and three subscales from the EDI-2

LifetimeAN: 2.2 24.4 (.9) Diagnosis: SCID-I/NP Lähteenmäki et al. (2014)b1863Finland♀20-35 A nationally representative sample drawn from The Mental Health in Early Adulthood in Finland (MEAF) study

Screening: SCOFFLifetime (L)ED (L, P): 6.0, 2.3 Diagnosis: SCID-IPoint (P)AN (L, P): 2.1, .3 BN (L, P): 2.3, 1.4 EDNOS (L, P): 2.0, .5 Isomaa et al. (2010)c595Finland♀ (283)♀: 15.3 (.3)15-year old 9th grade students Screening: Self-report questionnaire based on DSM-IV

Lifetime (L)♀ ♂ (312)♂: 15.4 (.4) Point (P) ED (L, P): 7.0, 2.8 AN (L, P): 1.8, .7 Diagnosis: RAB-T BN (L, P): .0, .0 ♂ ED (L, P): .0, .0 AN (L, P): .0, .0 BN (L, P): .0, .0 Raevuori et al. (Raevuori et al., 2009)d

2122Finland♂Screened at age: Male sample drawn from FinnTwin16 - A nation- wide population based cohort

Screening: Unspecified self-report questionnaire LifetimeAN: .2 22-27 24.4 (.8)Diagnosis: SCID-I/NP Interviewed at age: 25-30, Mean = 28.0 (Continued)

Downloaded by [University of Oslo] at 03:33 12 December 2017

(8)

Table 1. (Continued). Author (Year)N (Total)CountryGender (N or %)Age range mean (SD)Sample characteristicsAssessmentPrevalence typePrevalence rate (%) Keski-Rahkonen et al.(2009)e2545Finland♀21-27Sample drawn from FinnTwin16 - A nation- wide population based cohort Screening: Self-report questionnaire developed for the study and three subscales from the EDI-2

LifetimeBN: 1.7 24.4 (.9) Diagnosis: Short version of the SCID-I/NP Raevuori et al. (2008)f4388Finland♀ (2426)22-28Sample drawn from FinnTwin16 - A nation- wide population based cohort

Screening: Three subscales from the EDI-1Lifetime AN MZ (♀, ♂): 2.1, .2 ♂ (1962)24.4 (.9)AN DZ (♀, ♂): 2.9, .4 Diagnosis: Short version of the SCID-I/NPBN MZ (♀, ♂): 1.5, .0 BN DZ (♀, ♂): 1.6, .1 AN OS (♀, ♂): 1.6, .1 BN OS (♀, ♂): 2.1, .1 Thorsteinsdottir & Ulfarsdottir (2008)g

3052Iceland♀ (1818)15-20 College studentsScreening: ESPPointAN (♀, ♂): 1.1, .0 ♂ (1234)17.8 (1.2)Diagnosis: EDDSBN (♀, ♂): 5.6, .8 BED (♀, ♂): .6, .2 Keski-Rahkonen et al. (2007)2881Finland♀22-28 Sample drawn from FinnTwin16 - A nation- wide population based cohort Screening: Self-report questionnaire developed for the study and three subscales from the EDI-2

Lifetime (L)AN: 2.2 24.4 (.9) Diagnosis: Short version of the SCID-I/NP Keski-Rahkonen et al. (2006) 2881Finland♀22-28 Sample drawn from FinnTwin16 - A nation- wide population based cohort

Screening: Screening questionnaire, single item questions and three subscales from the EDI-2

Lifetime (L)AN (L, P): 1.9, .5 24.4 (.9)Point (P)BN (L, P): 2.1, .9 Diagnosis: Short version of the SCID-I/NP (Continued)

Downloaded by [University of Oslo] at 03:33 12 December 2017

(9)

Table 1. (Continued). Author (Year)N (Total)CountryGender (N or %)Age range mean (SD)Sample characteristicsAssessmentPrevalence typePrevalence rate (%) Rosenvinge et al. (Rosenvinge, Borge, & Börresen, 1999)h

678Norway♀ (464)15.4 (.3)A randomly selected sample of 15-year old students in public schools Screening: Three subscales from the EDI + SCANS

PointAN (♀, ♂): .4, .0 ♂ (214)BN (♀, ♂): 1.1, .0 Diagnosis: DSEDBED (♀, ♂): 1.5, .0 INTERVIEW Bulik et al. (Bulik et al., 2006)i31 406Sweden♀ (50.6%)54.6 (inter-quartile range, 49.0-57.6 years)

A Swedish Twin Registry subsampleDiagnosis: SCIDPointAN (♀): .6 ♂ (49.4%)AN (♂): .1 SELF-REPORT Litmanen et al. (2016)jT1: 1423FinlandT1 ♀ (673)T1 ♀: 15.5 (.4)15-year old 9th grade studentsDiagnosis: Self-report questions based on DSM-IV criteria

PointT1 AN (♀, ♂): .3, .1 T2: 1125T1 ♂ (750)T1 ♂: 15.5 (.4)T1 BN (♀, ♂): 3.0, .3 T2 AN (♀, ♂): .5, .2 T2 ♀ (575)T2 ♀: 15.6 (.4)T2 BN (♀, ♂): 2.3, .4 T2 ♂ (550)T2 ♂: 15.7 (.4) Trace et al. (2012)k13 295Sweden♀20-47A Swedish Twin Registry subsampleAn expanded SCID based instrumentLifetimeBED 5.8 Dellava et al. (Dellava, Thorn- ton, Lichtenstein, Pedersen, & Bulik, 2011)l

23 818Sweden♀20-47A Swedish Twin Registry subsampleDiagnosis: Online SCID- based questionnairePointAN: .4 (Continued)

Downloaded by [University of Oslo] at 03:33 12 December 2017

(10)

Author (Year)N (Total)CountryGender (N or %)Age range mean (SD)Sample characteristicsAssessmentPrevalence typePrevalence rate (%) Zachrisson et al. (2008)m1467Norway♀40.4 (12.7)A representative sample of the adult Norwegian female population

A modified version of SEDsLifetime (L)ED (L, P): 10.0, 4.0 Point (P)AN (L, P): .2, 1.0 BN (L, P): 4.1, 1.8 BED (L, P): .7, .3 EDNOS (L, P): 5.0, 1.8 Kjelsås et al. (2004)n1960Norway♀ (1026) 14-15Secondary school (9th and 10th grades) students Diagnosis: Modified version of SEDsLifetime (L)♀ ♂ (934) ♀: 14.5 (.5)Point (P)ED (L, P): 17.9, 8.0 ♂: 14.5 (.5) BN (L, P): 1.2, .5 EDNOS (L, P): 14.6, 6.5 BED (L, P): 1.5, .4 ♂ ED (L, P): 6.5, 2.5 BN (L, P): .4, .3 EDNOS (L, P): 5.0, 1.7 BED (L, P): .9, .4 Ghaderi & Scott (2001) 1157Sweden♀18-30Sample selected randomly from the female general population

Diagnosis: SEDsPoint ED: 3.2 AN: .1 BN: 1.3 BED: 1.2 EDNOS: .5 Ghaderi & Scott (1999) 1157Sweden♀18-30Sample selected randomly from the female general population

Diagnosis: SEDsLifetime (L)ED (L, P): 7.8, 2.6 Point (P)AN (L, P): .9, .0 BN (L, P): 2.9, 1.7 BED (L.P): 1.0, .5 EDNOS (L,P): 2.4, .3 (Continued)

Table 1. (Continued).

Downloaded by [University of Oslo] at 03:33 12 December 2017

(11)

Author (Year)N (Total)CountryGender (N or %)Age range mean (SD)Sample characteristicsAssessmentPrevalence typePrevalence rate (%) OTHER Goodman et al. (2014)o2 015 862Sweden♀ 48.7% 12-36Total-population Swedish cohort studyDiagnosis: Cases identified from the Swedish national inpatient, outpatient, or death registers based on ICD-10 diagnoses from 1997–2010

LifetimeBN (♀, ♂): .3, .0 ♂ 51.3% EDNOS (♀, ♂): .1, .1 Notes. ♀ = Females; ♂ = Males. Abbreviations: ED = Eating disorders; AN = Anorexia nervosa; BN = Bulimia nervosa, BED = Binge eating disorder; EDNOS = Eating disorder not otherwise specified; L = Lifetime prevalence; P = Point prevalence. aAN rates have been previously reported in Keski-Rahkonen et al. (2007). EDI-2 = Eating Disorder Inventory-2 (Garner, 1991); SCID-I/NP = Structured Clinical Interview for DSM-IV-TR Axis I Disor- ders, Research Version, Non-Patient Edition (First, Spitzer, Gibbon, & Williams, 2002). bSCOFF = Sick, Control, One stone, Fat, Food Questionnaire (Morgan, Reid, & Lacey, Morgan, Reid, & Lacey, 1999); SCID = Structured Clinical Interview for DSM-IV-TR Axis I Disorders (First et al., 1996). cRAB-T = Rating of Anorexia and Bulimia Interview-Teenage version (Clinton & Norring, 1999). dDSM-IV criteria excluding amenorrhea was used. eThe BN prevalence rate refers to the BN narrow category (i.e. according to the DSM-IV criteria). fEDI/EDI-1 = Eating Disorder Inventory (Garner, Olmsted, & Polivy, 1983); MZ = Monozygotic; DZ = Dizygotic; OS = Opposite Sex; Prevalence rates are reported for DSM-IV AN and DSM-IV BN only. Broad diagnoses are not reported. gESP = Eating Disorder Screen for Primary care (Cotton, Ball, & Robinson, 2003); EDDS = Eating Disorder Diagnostic Scale (Stice, Telch, & Rizvi, 2000). hSCANS = Setting Conditions for Anorexia and Bulimia Scale (Slade & Dewey, 1986); DSED = Diagnostic Survey for Eating Disorders (Johnson, 1993). iPrevalence rates are reported for narrow DSM-IV only (i.e. individuals who met full DSM-IV criteria by means of the SALT interview (excluding amenorrhoea for males). jT1 = Data collected 2002–2003, T2 = Data collected 2012–2013. kAn expanded SCID-based instrument. lPrevalence rates are reported for group 1 only (i.e. current DSM-IV criteria. All DSM-IV criteria for AN (AN-DSM-IV) is required). The prevalence rate reported is calculated as follows: 19.5% (92 people) of the 473 women with AN met criteria for AN-DSM-IV. 92/23818 = .0038 ≈ .4%. mOnly results from the data collection in 2004 is reported. SEDs = Survey for Eating Disorders (Ghaderi & Scott, 2002). nSame sample as in the included study from 1999 written by the same authors (Ghaderi &Scott) i.e. a follow-up study. oPrevalence rates are only reported for BN and EDNOS as a combination of ICD-9 and ICD-10 was used to diagnose AN (1987–2010).

Table 1. (Continued).

Downloaded by [University of Oslo] at 03:33 12 December 2017

(12)

from samples ranging in size from 10 000 to 30 000. One study had a sample size just above two million (N = 2 015 862) with ED cases identified using the Swedish national inpatient, outpatient, or death registers based on ICD-10 between 1997 and 2010. Nearly fifty per cent (N = 9) of the included studies had collected data from twin registries. These studies were either Finnish or Swedish. Five study samples were drawn from the general population, and the remaining five studies had assessed prevalence in student populations.

Data collection Design

Ten (52.6%) of the included studies followed what is known to be the gold standard in assessing prevalence rates; the two-stage design. Here, one or more self-report questionnaires were used to screen for ED risk, followed by an interview to establish ED diagnosis. One (5.8%) study used an interview alone to assess ED prevalence, and seven (36.8%) studies employed one or more self-reports. The final study had collected diagnostic data from national inpatient, outpatient, or death registers based on ICD-10, but without specifying the design of the study.

Assessment

In sum, 13 different assessment instruments had been used for either screening purposes or to assign specific ED diagnoses. The instrument most commonly applied to determine ED diagnoses was the SCID (First, Spitzer, Robert, Gibbon, & Williams, 1996) (all SCID interview formats included), which was used in eight (42.1%) out of the 19 included studies. All but one of these was conducted in Finland. The SEDs self-report questionnaire (Ghaderi & Scott, 2002) was used for diagnostic assessment in four (21.1%) studies. The Survey for Eating Disorders (SEDs) was only used in Swedish or Norwegian prevalence studies. A wide array of additional instruments was used to determine ED diagnoses, all of which were based on DSM-IV (American Psychiatric Asso- ciation, 1994) or ICD-10 (World Health Organisation, 1992) criteria, but without being specified any further (see Table 1).

Point- and lifetime prevalence rates

Point prevalence refers to the proportion of a population presenting with a condition at a speciﬁc point in time, whereas lifetime prevalence refers to the proportion of individuals who, at any point in their life up to the time of assessment, has suffered from the condition in question. In studies where overall lifetime ED prevalence differentiated between females and males (regardless of assessment method), rates ranged from 7.0% (Isomaa, Marttunen, Kaltiala-Heino, &

Isomaa, 2010) to 17.9% (Kjelsas, Bjornstrom, & Gotestam, 2004) in females, and from .1% (Isomaa et al., 2010) to 6.5% (Kjelsas et al., 2004) in males. AN and/or BN lifetime prevalence rates were most commonly reported in 2-stage design studies, with AN rates ranging from .2 to 2.9% in females (Raevuori et al., 2008) and from .0% (Isomaa et al., 2010) to .4% (Raevuori et al., 2008) in males. BN rates ranged from .0% (Isomaa et al., 2010) to 2.3% (Lähteenmäki et al., 2014) in females, and from .0% (Isomaa et al., 2010; Raevuori et al., 2008) to .4% (Kjelsas et al., 2004) in males. Point prevalence was most commonly reported in studies using the self-report design, with all but one study (Kjelsas et al., 2004) assessing ED prevalence in females, or females and males. Here, female point prevalence for overall ED rates ranged from 2.6% (Ghaderi & Scott, 1999) to 8.0%

(Kjelsas et al., 2004) and AN rates ranged from .0% (Ghaderi & Scott, 1999) to .5%. (Litmanen,

Downloaded by [University of Oslo] at 03:33 12 December 2017

(13)

Frojd, Marttunen, Isomaa, & Kaltiala-Heino, 2016) BN prevalence rates ranged from .5% (Kjelsas et al., 2004) to 3.0% in females (Litmanen et al., 2016). BED rates were only reported in two out of the ten 2-stage design studies, with point prevalence rates being .6 and 1.5% for females, and .2 and .0% for males. In self-report studies, BED point prevalence rates ranged from .3 to 1.2% in females, whereas the corresponding range for lifetime prevalence rates was 1.0–5.8%.

Discussion

The current study systematically reviewed studies of ED prevalence in the Nordic countries between 1994 and 2016. In addition to discrepant results regarding reported prevalence rates, we found significant variability across studies in terms of study design, measures applied to determine diagnoses and sample characteristics. Finland was the country with most studies (N = 9), all drawn from the same register (FinnTwin16). A majority of the reviewed studies uti- lized the gold standard 2-stage design; all of which were Finnish. The majority of the Norwegian (N = 3) and Swedish (N = 6) studies had assessed ED prevalence using self-report measures. None of the included studies were from Denmark.

As mentioned in the introduction, it is necessary to account for the impact of cultural aspects when evaluating the extent of ED pathology and reported prevalence rates. Although historically, EDs have been considered a Western phenomenon, ED prevalence rates are rising in non-Western societies, possibly due to increasing levels of industrialization and urbanization (Pike & Dunne, 2015). Nevertheless, EDs are still more common in Western countries than Asian countries (Qian et al., 2013). Keski-Rahkonen and Mustelin reviewed the epidemiology of EDs in Europe (Keski-Rahkonen & Mustelin, 2016) and reported prevalence among European females to be < 1–4% for AN,<1–2% for BN, and < 1–4% for BED with considerable variations depending on geographic area, age and ethnic origin. An ED prevalence of .3% was reported among Euro- pean males. In comparison, the present review of ED point prevalence (P) and lifetime prevalence (LP) among females in the Nordic region, regardless of assessment method, yielded rates in the range of .0–1.1% (P) and .2–2.9% (LP) for AN, .0–5.6% (P) and .0–4.1% (LP), respectively, for BN, and .4–1.5% (P) and .7–5.8% (LP) for BED. The ED point prevalence range for males was .0–2.5%

(2.3–8.0% for females). Thus, the reported prevalence range of AN and BED in females is lower in our review of the Nordic countries compared to Europe in general, and slightly higher for BN. However, accurate comparisons of prevalence rates are challenging due to methodological issues such as study design, age, sample size and variations in the diagnostic assessment. Also, as this review (Keski-Rahkonen & Mustelin, 2016) was descriptive rather than systematic, it is highly likely that a number of relevant studies have been omitted. Review studies of ED prevalence have also been conducted in other regions of the world. Kolar, Rodriguez, Chams, & Hoek, (Kolar, Rodriguez, Chams, & Hoek, 2016) conducted a systematic review and meta-analysis of 17 studies in Latin America, and found a mean point prevalence of .1% for AN, 1.16% for BN, and 3.53% for BED in the general population. These results show that the prevalence of AN is lower in Latin America compared to Western countries, whereas prevalence of BN, and especially BED, seems to be higher. Furthermore, a four-study meta-analysis on ED prevalence in Africa reported no cases of AN in females (Hoek, 2016), whereas BN had a combined point prevalence of .87%

which is slightly higher than the rate reported in the current review (.3%), and an EDNOS rate of 4.45%. The EDNOS rate in Africa is thus substantially higher than the rate reported in Nordic country females (.1%). In sum, these results indicate that the prevalence of BN among females

Downloaded by [University of Oslo] at 03:33 12 December 2017

(14)

in Africa was within the range of rates reported for Western, African-American and Latin Ameri- can populations. In contrast, the prevalence of EDNOS was higher than in Western populations and comparable to that in Latin America. Finally, Thomas et al. (Thomas, Lee, & Becker, 2016) conducted a review of ED prevalence in Asia and the Pacific as measured by disability-adjusted life years (DALYs). The authors suggest the prevalence of EDs to be comparable to those reported in Europe and North America. It should be highlighted that the studies mentioned above often report different prevalence estimates. For example, Hoek and colleagues (2016) report combined point prevalence, Kesti-Rahkonen and Mustelin (2016) report prevalence rates and the current study reports both point and lifetime prevalence.

Variations in the prevalence rates are affected by which type of population is investigated. The current review excluded studies which assessed the prevalence of EDs in psychiatric institutions, (Gotestam & Agras, 1995; Pagsberg & Wang, 1994; Steinhausen & Jensen, 2015; Taraldsen, Eriksen,

& Gotestam, 1996) hospitals, (Gotestam, Eriksen, Heggestad, & Nielsen, 1998) and athletes ( Torstveit, Rosenvinge, & Sundgot-Borgen, 2008). However, prevalence rates from representative community samples can also be biased by variables such as age and gender. The average age onset of EDs is somewhere between 14 and 19 years (Herpertz-Dahlmann, 2015), but due to great variation in age ranges across the included studies, it is difficult to establish whether there are any age trends in prevalence rates. However, significant gender differences in prevalence rates were found in the reviewed studies. This is not surprising as EDs are more frequent in females compared to males, and as ED in males still is an underreported phenomenon in the extant literature. Typically, the female-to-male ratio is defined as 10:1 for AN and BN, and approximately 2.5:1 for BED (Jacobi et al., 2004). Some epidemiological studies suggest that these ratios warrant further investigation though, proposing that EDs in males may be under-detected. (Raevuori, Keski-Rahkonen, & Hoek, 2014). This could be partly due to gender-bias in current measures of ED psychopathology. Whereas males are typically more concerned with muscularity, current ED measures have generally been developed to capture female-specific ED pathology such as drive for thinness and thin-ideal internalization. Also, these measures have predominantly been tested and evaluated with female samples, further contributing to the notion of a gender bias in available measures of ED psychopathology. The DSM-5 with its broadened diagnostic criteria for AN and BN (i.e. removal of the amenorrhoea criteria for AN), may contribute to more precise identifi- cation of males with EDs, and male specific assessment instruments such as the Eating Disorder Assessment for Men (EDAM) (Stanford & Lemberg, 2012) may prove fruitful in more accurately capturing ED pathology in males.

Issues related to sample characteristics also needs to be taken into consideration when inter- preting findings from the synthesized material in this review. The majority of included studies were conducted in either Finland (N = 9) or Sweden (N = 6). Further, with the exception of two studies, all Finnish studies investigated samples drawn from the FinnTwin16 cohort. Out of the 19 studies included, only 10 unique samples were investigated. These results represent significant sample biases. Furthermore, basing prevalence estimates on a particular target group, in this particular case twins, represent selection bias. Finally, none of the included studies were conducted in Denmark. One possible explanation is that Danish studies are published in Danish, and therefore excluded in our review. These particular results stress the importance of disseminating findings across national and international arenas, facilitating both national and cross-regional comparisons.

Downloaded by [University of Oslo] at 03:33 12 December 2017

(15)

Methodological variables can have a substantial impact on reported prevalence rates. One such variable is the inconsistent use of diagnostic tools to assess ED prevalence, an issue common in the field of EDs, and discussed in depth in the recent review by Lindvall Dahlgren and Wisting (Lindvall Dahlgren & Wisting, 2016). The included studies in the current review investigate an array of populations, using a variety of different assessment methods and instruments. For example, in general, point prevalence of AN in females vary between .0 and .7. However, in a Norwe- gian study on adults (Zachrisson, Vedul-Kjelsas, Gotestam, & Mykletun, 2008) and in an Icelandic study on adolescents (Thorsteinsdottir & Ulfarsdottir, 2008), point prevalence rates are reported as high as 1.0 and 1.1%, respectively. One potential explanation of such elevated rates is the use of self-report questionnaires which commonly have been reported to yield higher ED prevalence compared to investigator driven examinations (Lindvall Dahlgren & Wisting, 2016). When investigating point and lifetime prevalence rates in other ED diagnoses in the current review, similar trends showing generally higher rates in studies adopting self-report questionnaires compared to investigator-driven assessment is observed. However, it is important to note that relatively few studies were included in the current review, which may influence the findings. Nevertheless, the trend of lower ED prevalence when measured by investigator based measures is in line with previous studies (Lindvall Dahlgren & Wisting, 2016). It may be explained by a trained interview- er’s ability to both define and convey the meaning of, and differences between diagnostically significant concepts such as overeating, subjective and objective binge eating etc. When core ED behaviours and cognitions are identified consistently and scored accurately, this may prevent over reporting and/or misdiagnosing. Finally, it appears as though certain assessment instruments are more or less popular in certain geographic regions. This may, of course, lead to skewed distributions of prevalence rates. Although this might be because certain instruments are trans- lated and/or validated in the various countries, one should strive for a more homogenous use of instruments – not based on convenience.

Another relevant methodological aspect in the assessment of prevalence is the difference between point- vs. lifetime prevalence, as reports of lifetime prevalence is usually higher than point prevalence (Lindvall Dahlgren & Wisting, 2016). The included studies in our review are, more or less, equally distributed in terms of reports of point- or lifetime prevalence. As for the reported prevalence rates, the authors observe that studies of lifetime prevalence generally yield somewhat higher rates than studies of point prevalence. The fact that more individuals have had an ED during their lifetime than at one specific point in time is to be expected, especially as age ranges increase.

This is the very first study to review the prevalence of EDs in the Nordic region. The thorough literature search using three databases represents the core strength of the current review. The rigorous methodology applied in the screening process increases the likelihood that all literature relevant to the scope and aim of this review is covered. There are, however, some limitations to note. Firstly, the study is limited by not addressing associated features such as psychiatric and somatic comorbidity. Also, the current review excluded studies which assessed the prevalence of EDs in psychiatric institutions, hospitals and athletes, which could affect variations in the prevalence rates. Research has shown that the occurrence of EDs are higher in both female and male athletes (Joy, Kussman, & Nattiv, 2016; Reardon, 2017), and the inclusion of this particular group in addition to clinical samples would most likely lead to inflated prevalence rates compared to community samples. Finally, it should be noted that the reviewed studies were conducted in a time span of 22 years, and we cannot rule out the possibility that this could somehow influence

Downloaded by [University of Oslo] at 03:33 12 December 2017

(16)

the results. However, as all studies have used the same diagnostic system (DSM-IV), we expect potential variations due to time of assessment to be minimal.

Conclusion

Precise prevalence rates of any disorder are crucial given the importance of prevalence to inform prevention- and treatment planning. The current review demonstrates that there is an imperative need for studies investigating the prevalence of EDs in the different Nordic countries. There is a notable lack of two-stage designs in the studies reviewed which renders concern regarding the validity of the obtained prevalence figures. Further, the optimal method to increase both the precision and the power of estimated prevalence rates would be a meta-analysis. However, with the current shortage of studies in the Northern region employing a methodology which allows for a formal diagnosis, i.e. a full diagnostic interview, performing a meta-analysis would be unfounded. Thus, it is essential that future studies employ standardized methodologies as these will allow for comparison across geographic regions, gender and age, allowing us to map occurrence trends over time.

Contributorship statement

CLD and LW conceived of the study and initiated the study design. CLD, LW and KS participated in the acquisition, analysis and/or interpretation of data. CLD, LW and KS drafted the manuscript, and revised it critically for important intellectual content. All three authors, CLD, LW and KS, approved of the version of the manuscript to be published.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the Regional Department for Eating Disorders, Oslo University Hospital, Ullevål HF.

ORCID

Camilla Lindvall Dahlgren http://orcid.org/0000-0002-6773-475X

REFERENCES

Adler, N. E., & Ostrove, J. M. (1999). Socioeconomic status and health: What we know and what we don't.

Annals of the New York Academy of Sciences, 896, 3–15.

American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders DSM-IV (4th ed.).

Washington, DC: American Psychiatric Association.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders DSM-5 (5th ed.).

Washington, DC: American Psychiatric Association.

Andersen, T. M., Holmström, B., Honkapohja, S., Korkman, S., Söderström, H. S., & Vartiainen, J. (2007). The Nordic model: Embrazing globalization and sharing risks. Helsinki: Yliopistppaino.

Downloaded by [University of Oslo] at 03:33 12 December 2017

(17)

Becker, A. E., & Fay, K. (2006). Socio-cultural issues and eating disorders. In S. Wonderlich, M. de Zwaan, H.

Steiger, & J. Mitchell (Eds.), Annual review of eating disorders (pp. 35–63). Chicago, IL: Academy for Eating Disorders.

Bulik, C. M., Sullivan, P. F., Tozzi, F., Furberg, H., Lichtenstein, P., & Pedersen, N. L. (2006). Prevalence, herit- ability, and prospective risk factors for anorexia nervosa. Archives of General Psychiatry, 63(3), 305–312.

doi:10.1001/archpsyc.63.3.305

Cerasa, A., Castiglioni, I., Salvatore, C., Funaro, A., Martino, I., Alfano, S., & Quattrone, A. (2015). Biomarkers of eating disorders using support vector machine analysis of structural neuroimaging data: Preliminary results. Behavioural Neurology, 2015, 924814. doi:10.1155/2015/924814

Clinton, D., & Norring, C. (1999). The rating of anorexia and bulimia (RAB) interview: Development and preliminary validation. European Eating Disorders Review, 7(5), 362–371. doi:10.1002/(SICI)1099- 0968(199911)7:5<362::AID-ERV296>3.0.CO;2-0

Cotton, M. A., Ball, C., & Robinson, P. (2003). Four simple questions can help screen for eating disorders. Jour- nal of General Internal Medicine, 18(1), 53–56.

Dellava, J. E., Thornton, L. M., Lichtenstein, P., Pedersen, N. L., & Bulik, C. M. (2011). Impact of broadening definitions of anorexia nervosa on sample characteristics. Journal of Psychiatric Research, 45(5), 691–698.

doi:10.1016/j.jpsychires.2010.10.003

Everson, S. A., Maty, S. C., Lynch, J. W., & Kaplan, G. A. (2002). Epidemiologic evidence for the relation between socioeconomic status and depression, obesity, and diabetes. Journal of Psychosomatic Research, 53(4), 891–895.

First, M. B., Spitzer, R. L., Robert, L., Gibbon, M., & Williams, J. B. W. (1996). Structured clinical interview for DSM-IV Axis I disorders, clinician version (SCID-CV). Washington, DC: American Psychiatric Press.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (2002). Structured clinical interview for DSM-IV-TR Axis I disorders. Research Version, Non-patient Edition (SCID-I/NP). New York, NY: Biometrics Research, New York State Psychiatric Institute.

Garner, D. M. (1991). Eating disorder inventory-2 manual. Lutz, FL: Psychological Assessment Resources.

Garner, D. M., Olmsted, M. P., & Polivy, J. (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2, 15–34.

Ghaderi, A., & Scott, B. (1999). Prevalence and psychological correlates of eating disorders among females aged 18–30 years in the general population. Acta Psychiatrica Scandinavica, 99(4), 261–266.

Ghaderi, A., & Scott, B. (2001). Prevalence, incidence and prospective risk factors for eating disorders. Acta Psychiatrica Scandinavica, 104(2), 122–130.

Ghaderi, A., & Scott, B. (2002). The preliminary reliability and validity of the survey for eating disorders (SEDs):

A self-report questionnaire for diagnosing eating disorders. European Eating Disorders Review, 10(1), 61–76.

doi:10.1002/erv.425

Goodman, A., Heshmati, A., Malki, N., & Koupil, I. (2014). Associations between birth characteristics and eating disorders across the life course: Findings from 2 million males and females born in Sweden, 1975–1998.

American Journal of Epidemiology, 179(7), 852–863. doi:10.1093/aje/kwt445

Gotestam, K. G., & Agras, W. S. (1995). General population-based epidemiological study of eating disorders in Norway. International Journal of Eating Disorders, 18(2), 119–126.

Gotestam, K. G., Eriksen, L., Heggestad, T., & Nielsen, S. (1998). Prevalence of eating disorders in Norwegian general hospitals 1990–1994: Admissions per year and seasonality. International Journal of Eating Disorders, 23(1), 57–64.

Herpertz-Dahlmann, B. (2015). Adolescent eating disorders: Update on definitions, symptomatology, epidemiology, and comorbidity. Child and Adolescent Psychiatric Clinics of North America, 24(1), 177–196.

doi:10.1016/j.chc.2014.08.003

Hoek, H. W. (2016). Review of the worldwide epidemiology of eating disorders. Current Opinion in Psychiatry, 29(6), 336–339. doi:10.1097/YCO.0000000000000282

Hudson, C. G. (2005). Socioeconomic status and mental illness: Tests of the social causation and selection hypotheses. American Journal of Orthopsychiatry, 75(1), 3–18. doi:10.1037/0002-9432.75.1.3

Isomaa, R., Marttunen, M., Kaltiala-Heino, R., & Isomaa, A.-L. (2010). Obesity and eating disturbances are common in 15-year-old adolescents. A two-step interview study. [Obesity and eating disturbances are common in 15-year-old adolescents. A two-step interview study]. Nordic Journal of Psychiatry, 64(2), 123–129.

Jacobi, F., Wittchen, H. U., Holting, C., Hofler, M., Pfister, H., Muller, N., & Lieb, R. (2004). Prevalence, co-morbid- ity and correlates of mental disorders in the general population: Results from the German health interview and examination survey (GHS). Psychological Medicine, 34(4), 597–611. doi:10.1017/S0033291703001399

(Continued)