Background. Surveys of Behçet’s disease (BD) have shown substantial geographic variations in prevalence, but some of these differences may result from methodological inconsistencies. This meta-analysis explored the effect of geographic location and study methodology on the prevalence of BD.

Methods. We systematically searched the literature in electronic databases and by handsearching to identify population-based prevalence surveys of BD. Studies were eligible if they provided an original population-based prevalence estimate for BD with the number of prevalent cases identified in the study area. Pooled prevalence proportions across all studies were computed by using random effects models based on a Poisson normal distribution. Pre-defined subgroup analyses and meta-regression were used to investigate the effect of covariates on the prevalence proportions.

Results. We included 45 reports published from 1974 to 2015 and covering worldwide areas. The pooled estimates of prevalence proportions (expressed as cases/100 000 inhabitants) were 10.3 (95% CI 6.1, 17.7) for all studies and 119.8 (59.8, 239.9) for Turkey, 31.8 (12.9, 78.4) for the Middle East, 4.5 (2.2, 9.4) for Asia and 3.3 (2.1, 5.2) for Europe. Subgroup analyses showed a strikingly greater prevalence for studies with a sample survey design than a census design [82.5 (95% CI 47.3, 143.9) vs 3.6 (2.6, 5.1)]. Metaregression identified study design as an independent covariate significantly affecting BD prevalence proportions.

Conclusions. Differences in BD prevalence proportions likely reflect a combination of true geographic variation and methodological artefacts. In particular, use of a sample or census study design may strongly affect the estimated prevalence.

Rheumatology key messages

  • Pooled prevalence of Behçet’s disease from 45 international population-based surveys was 10.3/100 000 inhabitants.

  • Reported prevalence estimates for Behçet’s disease vary substantially by geographic area but also by the case-finding strategy.

  • These findings reconcile the geo-epidemiological data with the view of a high genetic burden in Behçet’s disease.

Introduction

Behçet’s disease (BD) is a chronic multisystem inflammatory disorder of predominantly young and middle-aged adults characterized by recurrent oral ulcers and a broad range of other manifestations such as genital ulcers, uveitis and articular, gastrointestinal, neurologic or vascular involvement. The aetiology of BD is unknown, although genetic factors, particularly HLA-B51 carriage, may play an important role. The environmental risk determinants are not well understood [1].

The geospatial distribution of BD may hold important clues to the aetiology of BD. The current understanding is that BD occurs in many populations across the world but that it is much more common in countries along the Silk Road and seemingly rare in northern European countries. Epidemiological surveys have reported remarkable variations in prevalence proportions that differ at extremes by a factor of 1000 [2]. The large differences in the reported occurrence strongly suggest the intervention of geography-specific aetiological factors and point to a major role of yet unknown environmental risk determinants. Indeed, genetic differences across ethnic/racial groups are not likely strong enough to explain extreme variations in the frequencies of complex diseases [3]. Thus, differences in frequencies of diseases may reflect geographic variations and also methodological inconsistencies, changes over time or random fluctuation. Meta-analysis is increasingly being used to summarize prevalence data for diseases [4, 5] with the aim of exploring the sources of variations in disease frequency and assessing their effect magnitudes [6].

Here we report on our meta-analysis of the effect of geographic location and methodological and other study characteristics on the global prevalence of BD.

Methods

We performed a systematic literature review and meta-analysis that followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statements [7]. Deviations from these statements, when deemed inappropriate to the context of comprehensively summarizing prevalence proportions, are specified in the corresponding Methods sections.

Systematic literature review process

We searched Medline (from 1960 onwards) and EMBASE (from 1950 onwards) for reports of population-based studies of the prevalence of BD by using combinations of the MeSH term Behçet syndrome with incidence or prevalence, with no restrictions on language or publication date. The main literature search was performed in March 2015 and updated in June 2016. We also handsearched reference lists of relevant publications, review articles, conference proceedings from international BD meetings and textbooks by using a snowballing strategy. We included peer-reviewed publications (full original articles and letters to the editor) and grey literature (e.g. conference abstracts, conference proceedings, edited book chapters and dissertations) since the results of meta-analyses are considered more accurate when also including literature not published in peer-reviewed journals [8]. Publications written in languages not mastered by any of the authors were analysed with the help of native speakers.

To be included in the meta-analysis, studies had to report a point estimate for BD prevalence and had to have a population-based survey design defined by a study population that could be considered representative of the general population. Studies were eligible if the reported estimate referred to only adult populations; surveys focusing exclusively on paediatric populations were not eligible. Pre-specified eligibility criteria also included the requirement that the publication had to provide the number of identified prevalent BD cases to guarantee that the reported prevalence proportion was based on an exact case count and not simply a subjective approximation; consequently, we excluded studies providing prevalence estimates but not the number of prevalent BD cases. One author (C.M.) reviewed the titles and abstracts of the studies identified and selected potentially eligible publications for full assessment.

Data extraction

Two authors (K.D. and C.M.) independently extracted the following data from all studies selected by using a study-specific data extraction form: publication reference (name of first author, study period), language and type; geographic location of the study area; study period; age limits of the study population (if relevant); study methodology; criteria used to classify BD; number of prevalent cases (numerator); size of the study population (denominator) and prevalence estimates. For studies involving multi-ethnic populations living in the same study area, the numbers extracted were for the entire population and not those for only the indigenous subpopulation. For studies not reporting the denominator population size, we first attempted to obtain that information by sending e-mails to the corresponding authors. If this was unsuccessful, the denominator population was back-calculated by using the reported numerators and crude prevalence proportions. If a study provided only a standardized prevalence proportion (with reference to the age or age–sex distribution of an external standard population), this figure was used for the back-calculation.

According to the case-finding strategy used, two main categories of study design were discerned: census and sample surveys. Census surveys were defined as investigations in which cases were found indirectly in one or more sources of medical information (e.g. hospital data, ad hoc surveys among community physicians, registries, memberships in patient support groups). Sample surveys were defined as studies in which cases were identified by directly approaching individuals (e.g. telephone interviews, household visits). The more intricate process of sample surveys involves the approach being applied to relatively small population samples selected from the background populations. Sample surveys were further characterized by whether they used a multistage algorithm, which typically includes a physical examination to identify previously undiagnosed BD. The reference type was classified as peer-reviewed publications or grey literature.

The extracted data were cross-checked and between-assessor disagreements were resolved by discussion and, when necessary, with a third author (A.M.). No formal quality assessment was performed because of the lack of consensus and specific instruments for grading the quality of studies determining the frequency of human diseases in the general population [9].

Meta-analyses of prevalence data and metaregression

Prevalence proportions, expressed as the number of BD cases per 100 000 inhabitants, were calculated for each study by dividing the number of prevalent BD cases by the size of the study population. The prevalence proportions were log transformed and included in a random effects meta-analysis, assuming a Poisson normal distribution to calculate pooled prevalence proportions and 95% CIs [10]. For studies with prevalence proportions of zero, we replaced the numerator counts of zero with 0.5 to allow for taking the log transformation. Inverse-variance weighting was used for data pooling and the between-study variance (τ2) was estimated by maximum likelihood estimators [11]. Pooled proportions and 95% CIs were then back-transformed to the original scale. Statistical heterogeneity of estimates among studies was quantified by using the I2 index and classified as low (25%), moderate (50%) or high (75%) [12]. Publication bias was not examined because it was deemed irrelevant in the framework of summarizing descriptive data on the frequency of a disease.

To explore potential sources of between-study heterogeneity, we performed subgroup analyses and mixed-effects Poisson metaregression [13]. Pre-specified stratification variables were geographic study area (according to common geographic delineations), study design (census vs sample survey), criteria used to classify BD (International Study Group criteria [14] or not), study period (classified in four periods with approximately equal numbers of studies) and reference type (peer-reviewed publications vs grey literature). Because the study period was not consistently reported in publications, we used year of publication as a proxy. Multivariable Poisson regression analysis was used to test the association of explanatory covariates and BD prevalence after adjustment for potential confounders.

Meta-analysis and metaregression were performed using the metafor package of the statistical software R (R Project for Statistical Computing, Vienna, Austria).

Results

Literature search

The flow of study selection is shown in Fig. 1. The electronic database search yielded 1451 citations; 12 additional publications were found by handsearching; 50 potentially eligible studies were identified [15–64]. Five studies were excluded, three due to non-reporting of numerator data [27, 30, 51] and two that included replicate data [29, 60]. From one study that provided prevalence data for two study populations, with one nested inside of the other, we selected the data for the smaller population for which more detailed information was provided in the publication [53]. Therefore, all pooled analyses pertained to 45 study populations reported in 45 publications [15–26, 28, 31–50, 52–59, 61–64].

Fig. 1

Flow of selection of studies reporting population-based prevalence estimates for BD

Fig. 1

Flow of selection of studies reporting population-based prevalence estimates for BD

Characteristics of identified prevalence surveys

The 45 studies were published between 1974 [61] and 2015 [25, 39]. In six studies, BD was one of a number of diseases for which prevalence proportions were estimated [20, 23, 45, 58, 59, 62]. The studies covered 22 countries, grouped into six geographic areas: Asia (5 studies) [45, 50, 58, 61, 62], Middle East (11 studies) [15–17, 23, 28, 37, 41–43, 46, 49], North America/Caribbean Islands (4 studies) [21, 31, 35, 44], northern Europe (9 studies) [24, 34, 38–40, 48, 53, 55, 64], southern Europe (8 studies) [22, 26, 33, 47, 52, 54, 56, 57] and Turkey (8 studies) [18–20, 25, 32, 36, 59, 63]. Overall, 26 studies used the International Study Group classification criteria for BD and the remaining used varied established classification criteria or study-specific definitions.

Tables 1 and 2 summarize the main characteristics of the 45 studies retained for analyses, the 28 studies based on a census methodology and the 17 studies based on a sample methodology. Among the 28 census surveys, case-finding relied on a single source of information in 20 studies and multiple sources in 8 (Table 1). Among the 17 sample surveys, 15 used a multistage approach with sequential screening algorithms to identify people with both previously diagnosed or undiagnosed BD.

Table 1

Main characteristics of 28 selected population-based prevalence surveys of BD involving a census methodology

References Reference type Geographic study area Classification criteria Case identification No. of cases Population size 
Assaad-Khalil et al. [17GL Middle East (Alexandria, Egypt) ISG Single source 274 3 600 000 
Calamia et al. [21PRP North America (Olmsted County, MN, USA) ISG/rheumatologist diagnosis Single source 130 947 
Cartella et al. [22PRP Southern Europe (province of Brescia, Italy) ISG Multiple sources 43 1 056 063 
Chamberlain [24PRP Northern Europe (Yorkshire County, UK) Study specific Multiple sources 32 5 000 000 
Crespo et al. [26GL Southern Europe (central zone of Portugal) ISG Single source 29 1 900 000 
Deligny et al. [31GL Caribbean Islands (Martinique, French West Indies) ISG Multiple sources 28 403 795 
Eiroa et al. [33PRP Southern Europe (A Coruña Province, Spain) Study specific Single source 28 500 000 
Ek and Hedfors [34PRP Northern Europe (Stockholm region, Sweden) ISG Single source 345 000 
Hirohata et al. [35PRP North America (Hawaii, USA) 1974 JBDRC Single source 768 561 
Jankowski et al. [38PRP Northern Europe (Scotland, UK) ISG Single source 15 5 500 000 
Kappen et al. [39PRP Northern Europe (Rotterdam, The Netherlands) ISG Single source 100 1 319 680 
Kidd [40GL Northern Europe (Hertfordshire County, UK) ISG Single source 20 1 000 000 
Krause et al. [42PRP Middle East (Galilee region, Israel) ISG Single source 112 737 000 
Lannuzel et al. [44PRP Caribbean Islands (Guadeloupe, French West Indies) Study specific Single source 13 420 000 
Mahr et al. [47PRP Southern Europe (Seine-Saint-Denis County, France) ISG Multiple sources 79 1 094 412 
Mohammad et al. [48PRP Northern Europe (Skane County, Sweden) ISG Multiple sources 40 809 317 
Mousa et al. [49PRP Middle East (Kuwait nationwide) O’Duffy and Goldstein 1976 Single source 29 1 374 600 
Nakae et al. [50GL Asia (Japan nationwide) 1987 JBDRC Single source 16 750 124 074 074a 
Olivieri et al. [52PRP Southern Europe (Potenza, Italy) ISG Multiple sources 11 69 060 
Papoutsis et al. [53PRP Northern Europe (Berlin, Germany) ABD classification Tree/Cheng and Zhang/PD Single source 165 3 391 344 
Peñafiel Burkhardt et al. [54RPP Southern Europe (province of Granada, Spain) ISG Single source 44 880 000a 
Riewerts et al. [55GL Northern Europe (state of Baden-Württemberg, Germany) NR Multiple sources 200 8 466 241 
Salvarani et al. [56PRP Southern Europe (Reggio Emilia area, Italy) ISG Multiple sources 18 486 961 
Sanchez-Burson et al. [57GL Southern Europe (Seville, Spain) ISG Single source 30 316 000 
See et al. [58PRP Asia (Taiwan nationwide) ICD 9 Single source 43 1 000 000 
Yamamoto et al. [61PRP Asia (Japan nationwide) 1974 JBDRC Single source 4132 118 057 143a 
Yu et al. [62PRP Asia (Taiwan nationwide) ICD-9 Single source 13 963 355 
Zouboulis et al. [64PRP Northern Europe (Berlin, Germany) ABD classification tree Single source 49 2 170 411 
References Reference type Geographic study area Classification criteria Case identification No. of cases Population size 
Assaad-Khalil et al. [17GL Middle East (Alexandria, Egypt) ISG Single source 274 3 600 000 
Calamia et al. [21PRP North America (Olmsted County, MN, USA) ISG/rheumatologist diagnosis Single source 130 947 
Cartella et al. [22PRP Southern Europe (province of Brescia, Italy) ISG Multiple sources 43 1 056 063 
Chamberlain [24PRP Northern Europe (Yorkshire County, UK) Study specific Multiple sources 32 5 000 000 
Crespo et al. [26GL Southern Europe (central zone of Portugal) ISG Single source 29 1 900 000 
Deligny et al. [31GL Caribbean Islands (Martinique, French West Indies) ISG Multiple sources 28 403 795 
Eiroa et al. [33PRP Southern Europe (A Coruña Province, Spain) Study specific Single source 28 500 000 
Ek and Hedfors [34PRP Northern Europe (Stockholm region, Sweden) ISG Single source 345 000 
Hirohata et al. [35PRP North America (Hawaii, USA) 1974 JBDRC Single source 768 561 
Jankowski et al. [38PRP Northern Europe (Scotland, UK) ISG Single source 15 5 500 000 
Kappen et al. [39PRP Northern Europe (Rotterdam, The Netherlands) ISG Single source 100 1 319 680 
Kidd [40GL Northern Europe (Hertfordshire County, UK) ISG Single source 20 1 000 000 
Krause et al. [42PRP Middle East (Galilee region, Israel) ISG Single source 112 737 000 
Lannuzel et al. [44PRP Caribbean Islands (Guadeloupe, French West Indies) Study specific Single source 13 420 000 
Mahr et al. [47PRP Southern Europe (Seine-Saint-Denis County, France) ISG Multiple sources 79 1 094 412 
Mohammad et al. [48PRP Northern Europe (Skane County, Sweden) ISG Multiple sources 40 809 317 
Mousa et al. [49PRP Middle East (Kuwait nationwide) O’Duffy and Goldstein 1976 Single source 29 1 374 600 
Nakae et al. [50GL Asia (Japan nationwide) 1987 JBDRC Single source 16 750 124 074 074a 
Olivieri et al. [52PRP Southern Europe (Potenza, Italy) ISG Multiple sources 11 69 060 
Papoutsis et al. [53PRP Northern Europe (Berlin, Germany) ABD classification Tree/Cheng and Zhang/PD Single source 165 3 391 344 
Peñafiel Burkhardt et al. [54RPP Southern Europe (province of Granada, Spain) ISG Single source 44 880 000a 
Riewerts et al. [55GL Northern Europe (state of Baden-Württemberg, Germany) NR Multiple sources 200 8 466 241 
Salvarani et al. [56PRP Southern Europe (Reggio Emilia area, Italy) ISG Multiple sources 18 486 961 
Sanchez-Burson et al. [57GL Southern Europe (Seville, Spain) ISG Single source 30 316 000 
See et al. [58PRP Asia (Taiwan nationwide) ICD 9 Single source 43 1 000 000 
Yamamoto et al. [61PRP Asia (Japan nationwide) 1974 JBDRC Single source 4132 118 057 143a 
Yu et al. [62PRP Asia (Taiwan nationwide) ICD-9 Single source 13 963 355 
Zouboulis et al. [64PRP Northern Europe (Berlin, Germany) ABD classification tree Single source 49 2 170 411 
a

Population size (denominator) back-calculated.

ABD: Adamantiades-Behçet disease; GL: grey literature; ICBD: International Criteria for Behçet’s Disease; ICD: International Statistical Classification of Diseases and Related Health Problems; ISG: International Study Group; JBDRC: Japanese Behçet’s Disease Research Committee criteria; ; NR: not reported; PD: physician diagnosis; PRP: peer-reviewed publication.

Table 2

Main characteristics of 17 selected population-based prevalence surveys of BD involving a sample methodology

References Reference type Geographic study area Classification criteria Population setting Case identification No. of cases (% known) Population size (% of respondersa
Al-Dalaan et al. [15GL Middle East (Al Quassim region, Saudi Arabia) ISG Households Questionnaire/PE 2 (NR) 10 267 (100) 
Al-Rawi et al. [16GL Middle East (Saglawia, Iraq) ISG Food distribution centre Questionnaire/PE 6 (NR) 12 617 (89) 
Azizlerli et al. [18PRP Turkey (Istanbul) ISG Households Questionnaire/PE 101 (46.5) 23 986 (100) 
Cakir et al. [19PRP Turkey (Havsa) ISG Households Questionnaire/PE 1 (NR) 4861 (85) 
Cakir et al. [20PRP Turkey (Havsa) ISG Households Questionnaire/PE 3 (NR) 15 280 (81) 
Chaaya et al. [23PRP Middle East (Lebanon nationwide) PD Households Questionnaire/PE 3 (NR) 3530 (100) 
Çölgeçen et al. [25PRP Turkey (Kayseri) ISG Medical records of primary health care centres Questionnaire/PE 9 (100) 5218 (100) 
Davatchi et al. [28PRP Middle East (Tehran, Iran) ICBD Households Questionnaire/PE 7 (NR) 10 291 (100) 
Demirhindi et al. [32PRP Turkey (Istanbul) O’Duffy Households Personal interview/ pathergy test/PE 4 (NR) 4940 (100) 
Idil et al. [36PRP Turkey (Ankara) ISG Households Personal interview/PE 16 (56.3) 13 894 (81) 
Jaber et al. [37PRP Middle East (Taibe, Israel) ISG Parents of children attending a paediatric centre Questionnaire/PE 6 (NR) 4876 (88) 
Klein et al. [41PRP Middle East (Dalyat El-Carmel, Israel) ISG Community clinics Personal interview/PE 2 (50.0) 4000 (100) 
Krause et al. [43GL Middle East (NR, Israel) ISG Patients attending a community health centre Personal interview/extensive interview and PE 0 (0) 1500 (100) 
Li et al.[45PRP Asia (Mentougou district of Beijing, China) ISG Households Questionnaire/PE 1 (NR) 10 556 (72) 
Madanat et al. [46GL Middle East (northern Jordan) NR Hospital workers Personal interviewb 12 (100) 2501 (100) 
Seyahi et al. [59PRP Turkey (Istanbul) PD Parents of children attending schools Questionnaire and validation of BD by medical interview, hospital or physician records, home visit or telephone interviewb 4 (100) 4462 (83) 
Yurdakul et al. [63PRP Turkey (Camas) O’Duffy Households Personal interview/PE 19 (5.3) 5131 (100) 
References Reference type Geographic study area Classification criteria Population setting Case identification No. of cases (% known) Population size (% of respondersa
Al-Dalaan et al. [15GL Middle East (Al Quassim region, Saudi Arabia) ISG Households Questionnaire/PE 2 (NR) 10 267 (100) 
Al-Rawi et al. [16GL Middle East (Saglawia, Iraq) ISG Food distribution centre Questionnaire/PE 6 (NR) 12 617 (89) 
Azizlerli et al. [18PRP Turkey (Istanbul) ISG Households Questionnaire/PE 101 (46.5) 23 986 (100) 
Cakir et al. [19PRP Turkey (Havsa) ISG Households Questionnaire/PE 1 (NR) 4861 (85) 
Cakir et al. [20PRP Turkey (Havsa) ISG Households Questionnaire/PE 3 (NR) 15 280 (81) 
Chaaya et al. [23PRP Middle East (Lebanon nationwide) PD Households Questionnaire/PE 3 (NR) 3530 (100) 
Çölgeçen et al. [25PRP Turkey (Kayseri) ISG Medical records of primary health care centres Questionnaire/PE 9 (100) 5218 (100) 
Davatchi et al. [28PRP Middle East (Tehran, Iran) ICBD Households Questionnaire/PE 7 (NR) 10 291 (100) 
Demirhindi et al. [32PRP Turkey (Istanbul) O’Duffy Households Personal interview/ pathergy test/PE 4 (NR) 4940 (100) 
Idil et al. [36PRP Turkey (Ankara) ISG Households Personal interview/PE 16 (56.3) 13 894 (81) 
Jaber et al. [37PRP Middle East (Taibe, Israel) ISG Parents of children attending a paediatric centre Questionnaire/PE 6 (NR) 4876 (88) 
Klein et al. [41PRP Middle East (Dalyat El-Carmel, Israel) ISG Community clinics Personal interview/PE 2 (50.0) 4000 (100) 
Krause et al. [43GL Middle East (NR, Israel) ISG Patients attending a community health centre Personal interview/extensive interview and PE 0 (0) 1500 (100) 
Li et al.[45PRP Asia (Mentougou district of Beijing, China) ISG Households Questionnaire/PE 1 (NR) 10 556 (72) 
Madanat et al. [46GL Middle East (northern Jordan) NR Hospital workers Personal interviewb 12 (100) 2501 (100) 
Seyahi et al. [59PRP Turkey (Istanbul) PD Parents of children attending schools Questionnaire and validation of BD by medical interview, hospital or physician records, home visit or telephone interviewb 4 (100) 4462 (83) 
Yurdakul et al. [63PRP Turkey (Camas) O’Duffy Households Personal interview/PE 19 (5.3) 5131 (100) 
a

Proportion of the original sample target population.

b

No active screening for previously undiagnosed BD.

GL: grey literature; ICBD: International Criteria for Behçet’s Disease; ISG: International Study Group; NR: not reported; PE: physical examination; PRP: peer-reviewed publication.

Summary results for prevalence

The study-specific prevalence proportions (per 100 000 inhabitants) across the 45 studies ranged from 0.1 (95% CI 0, 1.0) [35] to 479.8 (272.9, 843.7) [46] (Fig. 2). Table 3 shows the results of the pooled prevalence proportions (per 100 000 inhabitants) for all studies combined and within subgroups. The pooled overall prevalence proportion (per 100 000 inhabitants) was 10.3 (95% CI 6.1, 17.7) and featured high between-study heterogeneity (I2 = 99.9%) (Fig. 2). Weights attributed to the 45 individual studies ranged from 1.44 to 2.35%. Prevalence proportions were highest for Turkey [119.8 (95% CI 59.8, 239.9)] and lowest for northern Europe [2.1 (95% CI 1.1, 4.0)]. Between-study heterogeneity remained mostly high within geographic strata.

Fig. 2

Forest plot of prevalence estimates for BD reported by 45 population-based studies and pooled estimates

Pooled estimates are shown for all studies combined and within subgroups of studies involving census or sample surveys. The dashed vertical line represents the summary prevalence proportion for all studies combined.

Fig. 2

Forest plot of prevalence estimates for BD reported by 45 population-based studies and pooled estimates

Pooled estimates are shown for all studies combined and within subgroups of studies involving census or sample surveys. The dashed vertical line represents the summary prevalence proportion for all studies combined.

Table 3

Pooled prevalence proportions (per 100 000) for BD calculated by random effects meta-analysis

Variables No. of studies Prevalence (95% CI) τ2 I2 
All studies 45 10.3 (6.1, 17.7) 3.163 99.9 
By geographic area 
    Turkey 119.8 (59.8, 239.9) 0.805 91.0 
    Middle East 11 31.8 (12.9, 78.4) 2.053 98.2 
    Asia 4.5 (2.2, 9.4) 0.589 99.9 
    Southern Europea 5.3 (3.4, 8.2) 0.359 92.4 
    Northern Europea 2.1 (1.1, 4.0) 0.921 98.3 
    North America/Caribbean Islands 3.8 (2.2, 6.8) 0.183 60.9 
By study design 
 Census surveys 
        All studies 28 3.6 (2.6, 5.1) 0.818 99.6 
        Middle East and Asia 5.0 (2.7, 9.3) 0.672 99.9 
        Middle East 6.3 (2.5, 15.8) 0.637 98.5 
    Sample surveys     
        All studies 17 82.5 (47.3, 143.7) 1.060 90.3 
        Middle East and Asia 58.1 (26.3, 128.3) 1.066 81.2 
        Middle East 68.4 (30.9, 151.3) 0.965 81.1 
Classification criteria 
    International Study Group criteria 26 11.5 (6.1, 21.5) 2.463 98.9 
    Other criteria 19 8.8 (3.4, 22.6) 4.211 99.9 
Year of publication 
    1974–93 11 3.7 (1.0, 13.9) 4.748 99.9 
    1994–2003 10 16.9 (5.6, 50.7) 3.047 99.2 
    2004–10 12 11.1 (5.5, 22.3) 1.286 98.3 
    2011–14 12 14.7 (5.7, 38.2) 2.672 98.4 
Reference type 
    Peer-reviewed publications 34 10.3 (5.4, 19.7) 3.469 99.5 
    Grey literature 11 10.2 (4.0, 25.7) 2.279 99.6 
Variables No. of studies Prevalence (95% CI) τ2 I2 
All studies 45 10.3 (6.1, 17.7) 3.163 99.9 
By geographic area 
    Turkey 119.8 (59.8, 239.9) 0.805 91.0 
    Middle East 11 31.8 (12.9, 78.4) 2.053 98.2 
    Asia 4.5 (2.2, 9.4) 0.589 99.9 
    Southern Europea 5.3 (3.4, 8.2) 0.359 92.4 
    Northern Europea 2.1 (1.1, 4.0) 0.921 98.3 
    North America/Caribbean Islands 3.8 (2.2, 6.8) 0.183 60.9 
By study design 
 Census surveys 
        All studies 28 3.6 (2.6, 5.1) 0.818 99.6 
        Middle East and Asia 5.0 (2.7, 9.3) 0.672 99.9 
        Middle East 6.3 (2.5, 15.8) 0.637 98.5 
    Sample surveys     
        All studies 17 82.5 (47.3, 143.7) 1.060 90.3 
        Middle East and Asia 58.1 (26.3, 128.3) 1.066 81.2 
        Middle East 68.4 (30.9, 151.3) 0.965 81.1 
Classification criteria 
    International Study Group criteria 26 11.5 (6.1, 21.5) 2.463 98.9 
    Other criteria 19 8.8 (3.4, 22.6) 4.211 99.9 
Year of publication 
    1974–93 11 3.7 (1.0, 13.9) 4.748 99.9 
    1994–2003 10 16.9 (5.6, 50.7) 3.047 99.2 
    2004–10 12 11.1 (5.5, 22.3) 1.286 98.3 
    2011–14 12 14.7 (5.7, 38.2) 2.672 98.4 
Reference type 
    Peer-reviewed publications 34 10.3 (5.4, 19.7) 3.469 99.5 
    Grey literature 11 10.2 (4.0, 25.7) 2.279 99.6 

aThe ppooled estimate for Europe (combining southern and northern Europeans studies) was 3.3 (95%CI 2.1, 5.2).

τ2: estimate of between-study variance.

Subgroup analyses of the effect of study design showed striking differences in pooled BD prevalence proportions (per 100 000 inhabitants), with estimates of 3.6 (95% CI 2.6, 5.1) for census surveys vs 82.5 (95% CI 47.3, 143.7) for sample surveys. In light of the unequal geographic distributions of census and sample surveys and to rule out that the magnitude of differences in the computed BD prevalence proportions was artificially inflated by confounding with geographic location, we computed prevalence proportions within the areas where studies with either design were undertaken. The analyses of Middle East and Asian studies suggested an ∼12-fold greater pooled BD prevalence for sample surveys than census surveys (Table 3).

Further stratification of census surveys by whether case-finding occurred via single or multiple sources of medical information revealed similar pooled prevalence proportions across strata (data not shown). Classification criteria, study period or reference type had a negligible impact on BD prevalence (Table 3). Between-study heterogeneity remained high in all subgroups.

Metaregression analyses of prevalence proportions

Univariate metaregression analyses confirmed the strong effect of geographic area and study design on the pooled BD prevalence (both P < 0.0001). To separate the respective effects of geographic area and study design on BD prevalence, we performed multivariate metaregression analyses including both predictor variables (Table 4). When adjusting for geographic area, study design consistently predicted BD prevalence in the analyses of all studies (P = 0.0001) or only studies from the Middle East and Asia where both census and sample surveys were undertaken (P = 0.0003). In contrast, when adjusting for study design, the effect of geographic area did not significantly predict BD prevalence in the analyses of all geographic areas combined (P = 0.089) or the Middle East and Asian studies (P = 0.240). In addition, we studied the effect of geographic area for studies using census or sample methodologies. Geographic area did not significantly predict BD prevalence in studies with census surveys (P = 0.127) or sample surveys (P = 0.153).

Table 4

Uni- and multivariable metaregression analyses studying the effect of covariates on the pooled prevalence of BD

Explanatory variables No. of studies df P-valuea τ2 
None (all studies combined) 45  — 3.163 
Geographic area 
    All studies (six areas) 45 <0.0001 1.027 
        Adjusted for study design (six areas) 45 0.089 0.719 
    Middle East and Asian studies (two areas) 16 0.008 1.593 
        Adjusted for study design (two areas) 16 0.240 0.796 
    Census surveys (five areasb28 0.127 0.636 
    Sample surveys (three areasb17 0.153 0.849 
Study design (census vs sample surveys) 
    All studies 45 <0.0001 0.895 
        Adjusted for geographic area 45 0.0001 0.719 
    Middle East and Asian studies 16 <0.0001 0.849 
        Adjusted for geographic area 16 0.0003 0.796 
    Middle East studies 11 0.0003 0.850 
Classification criteria (ISG vs other criteria) 
    All studies 45 0.637 3.154 
        Adjusted for geographic area 45 0.833 1.025 
        Adjusted for study design 45 0.924 0.894 
        Adjusted for geographic area and study design 45 0.692 0.720 
Year of publication (across four calendar periodsc
    All studies 45 0.182 2.873 
        Adjusted for geographic area 45 0.049 0.864 
        Adjusted for study design 45 0.413 0.834 
        Adjusted for geographic area and study design 45 0.153 0.632 
Reference type (peer-reviewed vs grey literature) 
    All studies 45 0.995 3.162 
        Adjusted for geographic area 45 0.509 1.014 
        Adjusted for study design 45 0.738 0.891 
        Adjusted for geographic area and study design 45 0.519 0.710 
Explanatory variables No. of studies df P-valuea τ2 
None (all studies combined) 45  — 3.163 
Geographic area 
    All studies (six areas) 45 <0.0001 1.027 
        Adjusted for study design (six areas) 45 0.089 0.719 
    Middle East and Asian studies (two areas) 16 0.008 1.593 
        Adjusted for study design (two areas) 16 0.240 0.796 
    Census surveys (five areasb28 0.127 0.636 
    Sample surveys (three areasb17 0.153 0.849 
Study design (census vs sample surveys) 
    All studies 45 <0.0001 0.895 
        Adjusted for geographic area 45 0.0001 0.719 
    Middle East and Asian studies 16 <0.0001 0.849 
        Adjusted for geographic area 16 0.0003 0.796 
    Middle East studies 11 0.0003 0.850 
Classification criteria (ISG vs other criteria) 
    All studies 45 0.637 3.154 
        Adjusted for geographic area 45 0.833 1.025 
        Adjusted for study design 45 0.924 0.894 
        Adjusted for geographic area and study design 45 0.692 0.720 
Year of publication (across four calendar periodsc
    All studies 45 0.182 2.873 
        Adjusted for geographic area 45 0.049 0.864 
        Adjusted for study design 45 0.413 0.834 
        Adjusted for geographic area and study design 45 0.153 0.632 
Reference type (peer-reviewed vs grey literature) 
    All studies 45 0.995 3.162 
        Adjusted for geographic area 45 0.509 1.014 
        Adjusted for study design 45 0.738 0.891 
        Adjusted for geographic area and study design 45 0.519 0.710 
a

Statistical significance of the regression coefficient.

b

Census surveys were from North America/Caribbean Islands, Asia, Middle East, northern Europe and southern Europe; sample surveys were from Asia, Middle East and Turkey.

c

1974–93, 1994–2003, 2004–10 and 2011–14.

df: degree of freedom, ISG: International Study Group; τ2: estimate of between-study variance.

Classification criteria, study period and reference type had no significant effect on BD prevalence proportions in unadjusted analyses and mostly no significant effect after adjustment for geographic area and/or study design.

Discussion

Our analyses pooled prevalence data for BD reported from 45 population-based surveys. The overall prevalence estimate of 10.3/100 000 inhabitants should be interpreted as the best global summary estimate of available frequency data for BD from countries and areas where such studies were undertaken. More importantly, our findings suggest that in addition to geographic location, the case-finding methodology used to generate prevalence estimates strongly affected the outcome of such studies. Conversely, the criteria used to classify BD had no perceptible impact on disease prevalence estimates, which corroborates the only marginally different performance characteristics of the various classification systems for BD [65].

Meta-analysis of prevalence data offers the opportunity to disentangle true variations in disease frequency from the effects of confounders, but it also has some conceptual challenges [5]. In particular, different mathematical methods have been suggested to transform individual prevalence estimates (i.e. log or arcsine) and eventually summarize them [10]. Some meta-analyses of prevalence data used negative binomial models, which is essentially a Poisson model with a gamma-distributed random effect [66]. However, we used a Poisson normal model because the assumed normal distribution of the random effects component better suits the use of between-study heterogeneity statistics such as I2. Even though these choices likely result in slight differences in absolute estimates, they should not substantially bias the relative differences across subgroups and thus our conclusions in terms of the factors predicting BD prevalence.

As anticipated, the analyses showed geographic disparities in BD prevalence. The pooled prevalence estimate of 120/100 000 for Turkey, handled as a stand-alone geographic area because of the large number of studies (n = 8) identified for this single country, indicates a 4- to 57-fold higher prevalence in Turkey than in any other area. The results also showed a south to north decreasing frequency in Europe, which reinforces the rare occurrence of BD in northern Europeans. Because several of the northern European studies analysed multi-ethnic populations in which non-European migrants accounted for disproportionately high numbers among all identified cases [34, 39, 48, 53, 64], the occurrence of BD in people of genuine northern European background must be even lower than indicated by our estimate.

Pertinently, our study estimated a 22-fold higher pooled BD prevalence for sample surveys, at times also called field [63] or population surveys [4], than those based on a census methodology. To determine whether this result was confounded by geographic variations, we performed subgroup analyses restricted to geographic areas where surveys with either study design were undertaken. These analyses revealed a 12-fold higher disease frequency determined by sample than census surveys. Likewise, metaregression analyses adjusted for geographic areas consistently identified study design as a significant predictor of BD prevalence.

Several factors potentially explain the greater estimates with sample vs census surveys. By approaching people individually, sample surveys circumvent the risk of undercounting inherent to census surveys, which collect cases indirectly from sources of medical information that may provide incomplete data. As well, sample surveys often involve active screening for disease. Among the 14 sample surveys with active screening for BD, 5 provided the number of newly identified cases [18, 25, 36, 41, 63] and revealed that up to 95% of the identified cases were previously not known to have BD. The estimates from sample surveys may be further inflated by responder bias and skewed sampling of populations with demographic or other characteristics that bear an increased likelihood of having BD. Age and gender standardization or use of responders only as a denominator for prevalence calculations, which may prevent some of these biases, were only rarely implemented [20, 45].

The finding that prevalence proportions derived from census and sample surveys may not be directly comparable is of importance for illnesses other than BD. Indeed, recent data suggest that sample surveys result in 10- to 20-fold higher prevalence estimates for primary Sjögren's syndrome as compared with census surveys [4, 67]. The magnitude of difference in the estimates produced by census and sample surveys cannot be established as a general rule. In the context of BD, an indication is provided by the 12-fold difference found within the Asian and Middle Eastern studies, but this estimate is not implicitly generalizable to other areas or other diseases.

The fact that all BD studies from high-prevalence Turkey and none from low-prevalence Europe used a sample approach suggests that the true geographic differences in BD prevalence might be less extreme than suggested by individual studies and our pooled estimates for geographic strata. Although the incidence of BD has been much less studied, data from Asian and European populations also suggest more moderate geographic disparities, with ∼5-fold absolute differences in the incidence of BD [2]. A less extreme geographic gradient in BD prevalence reconciles the epidemiological data with the finding that 19–52% of BD risk is attributable to HLA-B51 carriage [68]. Indeed, the differences in frequencies of HLA-B51 in general background populations, which grossly parallel the geospatial distribution of BD [69], can explain, at most, moderate variations in BD risk across subpopulations.

Our study has limitations. Because of the relatively small number of studies, especially for some subgroup analyses, some results are prone to the effect of random error or lack of statistical power. In particular, the metaregression findings indicating BD prevalence significantly driven by study design but not geographic area could have resulted from low statistical power and should not lead to the conclusion that the reported geographic variations in the prevalence of BD are merely due to confounding. Also, the finding that between-study heterogeneity was not wholly explained by geographic area and study design leaves some uncertainty about other BD risk factors or study-specific confounders affecting BD prevalence.

To conclude, these findings shed a different light on understanding the geo-epidemiology of BD. By pinpointing methodological artefacts as a strong source of between-study heterogeneity, they indicate that true variations in BD prevalence may be less prominent than suggested by estimates reported from individual studies and better align the geo-epidemiology data with the suspected high genetic burden of BD. The conceptual differences in findings generated by census vs sample surveys urge a differential interpretation of the meanings of the two approaches, especially if the latter involves screening for previously undiagnosed illness. These observations also highlight the general need for better standards of conducting and reporting studies describing the frequencies of diseases.

Acknowledgements

We thank Drs M. Kremers (Rochester, MA, USA) and J. Sánchez-Bursón (Seville, Spain) for kindly providing additional information on published material and Dr J.Y. Mary (Paris, France) for critical reading of the manuscript.

Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declared no conflicts of interest.

References

1
Yazici
H
Fresko
I
Yurdakul
S.
Behçet’s syndrome: disease manifestations, management, and advances in treatment
.
Nat Clin Pract Rheumatol
 
2007
;
3
:
148
55
.
2
Mahr
A
Maldini
C.
Epidemiology of Behçet’s disease
.
Rev Med Interne
 
2014
;
35
:
81
9
.
3
Ioannidis
JP
Ntzani
EE
Trikalinos
TA.
Racial’ differences in genetic effects for complex diseases
.
Nat Genet
 
2004
;
36
:
1312
8
.
4
Qin
B
Wang
J
Yang
Z
et al.  .
Epidemiology of primary Sjogren’s syndrome: a systematic review and meta-analysis
.
Ann Rheum Dis
 
2015
;
74
:
1983
9
.
5
Wijnands
JM
Viechtbauer
W
Thevissen
K
et al.  .
Determinants of the prevalence of gout in the general population: a systematic review and meta-regression
.
Eur J Epidemiol
 
2015
;
30
:
19
33
.
6
Anello
C
Fleiss
JL.
Exploratory or analytic meta-analysis: should we distinguish between them?
J Clin Epidemiol
 
1995
;
48
:
109
16
; discussion 17–8.
7
Moher
D
Liberati
A
Tetzlaff
J
Altman
DG
Group
P.
Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement
.
BMJ
 
2009
;
339
:
b2535
.
8
Hopewell
S
McDonald
S
Clarke
M
Egger
M.
Grey literature in meta-analyses of randomized trials of health care interventions
.
Cochrane Database Syst Rev
 
2007
;
2
:
MR000010
.
9
Shamliyan
T
Kane
RL
Jansen
S.
Systematic reviews synthesized evidence without consistent quality assessment of primary studies examining epidemiology of chronic diseases
.
J Clin Epidemiol
 
2012
;
65
:
610
8
.
10
Stijnen
T
Hamza
TH
Ozdemir
P.
Random effects meta-analysis of event outcome in the framework of the generalized linear mixed model with applications in sparse data
.
Stat Med
 
2010
;
29
:
3046
67
.
11
Sutton
AJ
Abrams
KR
Jones
DR
Sheldon
TA
Song
F.
Methods for Meta-analysis in Medical Research
 .
New York
:
Wiley
;
2000
.
12
Higgins
JP
Thompson
SG
Deeks
JJ
Altman
DG.
Measuring inconsistency in meta-analyses
.
BMJ
 
2003
;
327
:
557
60
.
13
Bagos
P
Nikolopoulos
G.
Mixed-effects Poisson regression models for meta-analysis of follow-up studies with constant or varying durations
.
Int J Biostat
 
2009
;
5
:
21
.
14
Criteria for diagnosis of Behçet’s disease
. International Study Group for Behçet’s Disease.
Lancet
 
1990
;
335
:
1078
80
.
15
Al-Dalaan
A
Al-Ballaa
S
Al Sukati
M
et al.  . The prevalence of Behçet’s disease in Al Qassim region of Saudi Arabia. In:
Hamza
M
, ed.
Behcet’s Disease
 .
Tunis
:
Pub Adhoua
;
1997
:
170
2
.
16
Al-Rawi
ZS
Neda
AH.
Prevalence of Behçet’s disease among Iraqis
.
Adv Exp Med Biol
 
2003
;
528
:
37
41
.
17
Assaad-Khalil
S
Kamel
F
Ismail
E.
Starting a regional registry for patients with Behçet’s disease in North West Nile Delta region in Egypt. In:
Hamza
M
, ed.
Behcet’s Disease
 .
Tunis
:
Pub Adhoua
;
1997
:
173
6
.
18
Azizlerli
G
Kose
AA
Sarica
R
et al.  .
Prevalence of Behçet’s disease in Istanbul, Turkey
.
Int J Dermatol
 
2003
;
42
:
803
6
.
19
Cakir
N
Dervis
E
Benian
O
et al.  .
Prevalence of Behçet’s disease in rural western Turkey: a preliminary report
.
Clin Exp Rheumatol
 
2004
;
22
:
S53
5
.
20
Cakir
N
Pamuk
ON
Dervis
E
et al.  .
The prevalences of some rheumatic diseases in western Turkey: Havsa study
.
Rheumatol Int
 
2012
;
32
:
895
908
.
21
Calamia
KT
Wilson
FC
Icen
M
et al.  .
Epidemiology and clinical characteristics of Behçet’s disease in the US: a population-based study
.
Arthritis Rheum
 
2009
;
61
:
600
4
.
22
Cartella
S
Filippini
M
Tincani
A
Airo
P.
Prevalence of Behçet’s disease in the province of Brescia in northern Italy
.
Clin Exp Rheumatol
 
2014
;
32(4 Suppl 84)
:
S176
.
23
Chaaya
M
Slim
ZN
Habib
RR
et al.  .
High burden of rheumatic diseases in Lebanon: a COPCORD study
.
Int J Rheum Dis
 
2012
;
15
:
136
43
.
24
Chamberlain
MA.
Behçet’s syndrome in 32 patients in Yorkshire
.
Ann Rheum Dis
 
1977
;
36
:
491
9
.
25
Colgecen
E
Ozyurt
K
Ferahbas
A
et al.  .
The prevalence of Behçet’s disease in a city in Central Anatolia in Turkey
.
Int J Dermatol
 
2015
;
54
:
286
9
.
26
Crespo
J
Ribeiro
J
Jesus
E
et al.  . Behçet’s disease-particular features at the central zone of Portugal. In:
Wechsler
B
Godeau
P
, eds.
Behçet’s Disease
 .
Amsterdam
:
Excerpta Medica
;
1993
:
207
10
.
27
Davatchi
F
Shahram
F
Akbarian
M
et al.  .
Behçet’s disease-analysis of 3443 cases
.
APLAR J Rheumatol
 
1997
;
1
:
2
5
.
28
Davatchi
F
Jamshidi
AR
Banihashemi
AT
et al.  .
Prevalence of Behçet’s disease in Iran: a WHO-ILAR COPCORD stage I study
.
APLAR J Rheumatol
 
2007
;
10
:
239
43
.
29
Davatchi
F
Jamshidi
AR
Banihashemi
AT
et al.  .
WHO-ILAR COPCORD Study (Stage 1, Urban Study) in Iran
.
J Rheumatol
 
2008
;
35
:
1384
.
30
Davatchi
F
Sandoughi
M
Moghimi
N
et al.  .
Epidemiology of rheumatic diseases in Iran from analysis of four COPCORD studies
.
Int J Rheum Dis
 
2016
;
19
:
1056
62
.
31
Deligny
C
Antonio
L
Garnery
B
et al.  .
Épidémiologie et description à base de population de la maladie de Behçet en Martinique [abstract]
.
Rev Med Interne
 
2012
;
33(Supp 1)
:
S90
.
32
Demirhindi
O
Yazici
H
Binyildiz
P
et al.  .
The prevalence of Behçet’s disease in Fener village (Silivri, Istanbul) and its surroundings
.
Cerrahpasa Tip Fak Derg
 
1981
;
12
:
509
14
.
33
Eiroa
P
Sanchez
J
Rosales
M
et al.  .
Epidemiological study of Behçet’s disease in the sanitary area of la Coruña
.
Rev Esp Reumatol
 
1991
;
18
:
285
7
.
34
Ek
L
Hedfors
E.
Behçet’s disease: a review and a report of 12 cases from Sweden
.
Acta Derm Venereol
 
1993
;
73
:
251
4
.
35
Hirohata
T
Kuratsune
M
Nomura
A
Jimi
S.
Prevalence of Behçet’s syndrome in Hawaii. With particular reference to the comparison of the Japanese in Hawaii and Japan
.
Hawaii Med J
 
1975
;
34
:
244
6
.
36
Idil
A
Gurler
A
Boyvat
A
et al.  .
The prevalence of Behçet’s disease above the age of 10 years. The results of a pilot study conducted at the Park Primary Health Care Center in Ankara, Turkey
.
Ophthalmic Epidemiol
 
2002
;
9
:
325
31
.
37
Jaber
L
Milo
G
Halpern
GJ
Krause
I
Weinberger
A.
Prevalence of Behçet’s disease in an Arab community in Israel
.
Ann Rheum Dis
 
2002
;
61
:
365
6
.
38
Jankowski
J
Crombie
I
Jankowski
R.
Behçet’s syndrome in Scotland
.
Postgrad Med J
 
1992
;
68
:
566
70
.
39
Kappen
JH
van Dijk
EH
Baak-Dijkstra
M
et al.  .
Behçet’s disease, hospital-based prevalence and manifestations in the Rotterdam area
.
Neth J Med
 
2015
;
73
:
471
7
.
40
Kidd
D.
The prevalence of Behçet’s syndrome and its neurological complications in Hertfordshire, U.K
.
Adv Exp Med Biol
 
2003
;
528
:
95
7
.
41
Klein
P
Weinberger
A
Altmann
VJ
et al.  .
Prevalence of Behçet’s disease among adult patients consulting three major clinics in a Druze town in Israel
.
Clin Rheumatol
 
2010
;
29
:
1163
6
.
42
Krause
I
Yankevich
A
Fraser
A
et al.  .
Prevalence and clinical aspects of Behçet’s disease in the north of Israel
.
Clin Rheumatol
 
2007
;
26
:
555
60
.
43
Krause
I
Schlossberg
T
Weingarten
M
Weinberger
A.
Prevalence of Behçet’s disease in Israeli Yemenite Jews [abstract]
.
Clin Exp Rheumatol
 
2010
;
28(Suppl 60)
:
S158
9
.
44
Lannuzel
A
Lamaury
I
Charpentier
D
Caparros-Lefebvre
D.
Neurological manifestations of Behçet’s disease in a Caribbean population: clinical and imaging findings
.
J Neurol
 
2002
;
249
:
410
8
.
45
Li
R
Sun
J
Ren
LM
et al.  .
Epidemiology of eight common rheumatic diseases in China: a large-scale cross-sectional survey in Beijing
.
Rheumatology
 
2012
;
51
:
721
9
.
46
Madanat
W
Alawneh
K
Smadi
M
et al.  . The prevalence of Behçet’s disease in Jordan, a preliminary report [abstract]. 16th International Conference on Behçet’s Disease, Paris, 18–20 September
2014
, P38.
47
Mahr
A
Belarbi
L
Wechsler
B
et al.  .
Population-based prevalence study of Behçet’s disease: differences by ethnic origin and low variation by age at immigration
.
Arthritis Rheum
 
2008
;
58
:
3951
9
.
48
Mohammad
A
Mandl
T
Sturfelt
G
Segelmark
M.
Incidence, prevalence and clinical characteristics of Behçet’s disease in southern Sweden
.
Rheumatology
 
2013
;
52
:
304
10
.
49
Mousa
AR
Marafie
AA
Rifai
KM
Dajani
AI
Mukhtar
MM.
Behçet’s disease in Kuwait, Arabia. A report of 29 cases and a review
.
Scand J Rheumatol
 
1986
;
15
:
310
32
.
50
Nakae
K
Masaki
F
Hashimoto
T
Inaba
G.
Recent epidemiological features of Behçet’s disease in Japan. In:
Wechsler
B
Godeau
P
, eds.
Behçet’s Disease
 .
Amsterdam
:
Excerpta Medica
,
1993
,
145
51
.
51
O’Duffy
JD.
Summary of International Symposium on Behçet’s Disease. Istanbul, September 29–30, 1977
.
J Rheumatol
 
1978
;
5
:
229
33
.
52
Olivieri
I
Leccese
P
Padula
A
et al.  .
High prevalence of Behcet’s disease in southern Italy
.
Clin Exp Rheumatol
 
2013
;
31
:
28
31
.
53
Papoutsis
NG
Abdel-Naser
MB
Altenburg
A
et al.  .
Prevalence of Adamantiades-Behçet’s disease in Germany and the municipality of Berlin: results of a nationwide survey
.
Clin Exp Rheumatol
 
2006
;
24
:
S125
.
54
Peñafiel Burkhardt
R
Callejas Rubio
JL
Jimenez Alonso
JF
Ortego Centeno
N.
Behçet’s disease in Spain
.
Med Clin (Barc)
 
2007
;
128
:
717
.
55
Riewerts
F
Meisner
C
Peter
H
et al.  .
Prevalence of Behçet’s disease in south-west Germany among the adult Turkish and German population [abstract]
.
Clin Exp Rheumatol
 
2010
;
28(Suppl 60)
:
S150
.
56
Salvarani
C
Pipitone
N
Catanoso
MG
et al.  .
Epidemiology and clinical course of Behçet’s disease in the Reggio Emilia area of Northern Italy: a seventeen-year population-based study
.
Arthritis Rheum
 
2007
;
57
:
171
8
.
57
Sanchez Burson
J
Grandal
Y
Mendoza
M
et al.  . Clinical characteristics, HLA antigen and mortality in Behçet’s syndrome in Spain [abstract]. In:
Olivieri
I
Salvarani
C
Cantini
F
, eds.
8th International Congress on Behcet’s Disease: Program and Abstracts
 .
Milan
:
Milano Prex
,
1998
,
102
.
58
See
LC
Kuo
CF
Chou
IJ
Chiou
MJ
Yu
KH.
Sex- and age-specific incidence of autoimmune rheumatic diseases in the Chinese population: a Taiwan population-based study
.
Semin Arthritis Rheum
 
2013
;
43
:
381
6
.
59
Seyahi
E
Tahir Turanli
E
Mangan
MS
et al.  .
The prevalence of Behçet’s syndrome, familial Mediterranean fever, HLA-B51 and MEFV gene mutations among ethnic Armenians living in Istanbul, Turkey
.
Clin Exp Rheumatol
 
2010
;
28(4 Suppl 60)
:
S67
75
.
60
Tuzun
Y
Yurdakul
S
Cem Mat
M
et al.  .
Epidemiology of Behçet’s syndrome in Turkey
.
Int J Dermatol
 
1996
;
35
:
618
20
.
61
Yamamoto
S
Toyokawa
H
Matsubara
J
et al.  .
A nationwide survey on Behçet’s disease in Japan. I. Epidemiological survey
.
Jpn J Ophthalmol
 
1974
;
18
:
282
90
.
62
Yu
KH
See
LC
Kuo
CF
Chou
IJ
Chou
MJ.
Prevalence and incidence in patients with autoimmune rheumatic diseases: a nationwide population-based study in Taiwan
.
Arthritis Care Res
 
2013
;
65
:
244
50
.
63
Yurdakul
S
Gunaydin
I
Tuzun
Y
et al.  .
The prevalence of Behçet’s syndrome in a rural area in northern Turkey
.
J Rheumatol
 
1988
;
15
:
820
2
.
64
Zouboulis
CC
Kotter
I
Djawari
D
et al.  .
Epidemiological features of Adamantiades-Behçet’s disease in Germany and in Europe
.
Yonsei Med J
 
1997
;
38
:
411
22
.
65
International Team for the Revision of the International Criteria for Behçet’s Disease
.
The International Criteria for Behçet’s Disease (ICBD): a collaborative study of 27 countries on the sensitivity and specificity of the new criteria
.
J Eur Acad Dermatol Venereol
 
2014
;
28
:
338
47
.
66
Hilbe
J.
Negative Binomial Regression
 .
Cambridge, UK
:
Cambridge University Press
,
2011
.
67
Maldini
C
Seror
R
Fain
O
et al.  .
Epidemiology of primary Sjogren’s syndrome in a French multiracial/multiethnic area
.
Arthritis Care Res
 
2014
;
66
:
454
63
.
68
de Menthon
M
Lavalley
MP
Maldini
C
Guillevin
L
Mahr
A.
HLA-B51/B5 and the risk of Behcet's disease: a systematic review and meta-analysis of case-control genetic association studies
.
Arthritis Rheum
 
2009
;
61
:
1287
96
.
69
Verity
DH
Marr
JE
Ohno
S
Wallace
GR
Stanford
MR.
Behçet’s disease, the Silk Road and HLA-B51: historical and geographical perspectives
.
Tissue Antigens
 
1999
;
54
:
213
20
.

Comments

0 Comments