Objectives To investigate the impact of coeliac disease (CD) diagnosis on anthropometric measures at late adolescence and to assess trends in the prevalence of diagnosed CD over time.
Design A population based study.
Patients Prior to enlistment, at the age of 17 years, most of the Israeli Jewish population undergoes a general health examination. Subjects' medical diagnoses are entered into a structured database.
Interventions The enlistment database was thoroughly searched for CD cases between the years 1988 and 2015. Medical records of 2 001 353 subjects were reviewed.
Main outcome measures Anthropometric measures at the age of 17 years.
Results Overall, 10 566 CD cases (0.53%) were identified and analysed. Median age at data ascertainment was 17.1 years (IQR, 16.9–17.4). Multivariable analysis demonstrated that boys with CD were leaner (Body Mass Index 21.2±3.7 vs 21.7±3.8, p=0.02) while girls with CD were shorter (161.5±6 cm vs 162.1±6 cm, p=0.017) than the general population. The prevalence of diagnosed CD increased from 0.5% to 1.1% in the last 20 years with a female predominance (0.64% vs 0.46%). CD prevalence was significantly lower in subjects of lower socioeconomic status and those of African, Asian and former Soviet Union origin.
Conclusions Adolescent boys with CD were leaner and girls with CD were shorter compared with the general population. However, the clinical relevance of the small differences suggests that when CD is diagnosed during childhood, final weight and height are not severely impaired. Our cohort reinforces the observed increase in diagnosed CD.
- Adolescent Health
- Outcomes research
Statistics from Altmetric.com
What is already known on this topic?
Evidence regarding final adult height of patients with coeliac disease (CD) is inconclusive.
Incidence of diagnosed CD is increasing with distinct regional variability.
What this study adds?
Adolescent girls diagnosed with coeliac disease (CD) have lower height compared with the general population while boys have significantly lower weight and BMI.
However, final weight and height near the end of adolescence are not severely impaired in subjects with diagnosed CD.
The prevalence of diagnosed CD among Jewish adolescents is gradually increasing reaching 1.1% in 2015.
Coeliac disease (CD) is an autoimmune disorder precipitated by the ingestion of gluten in genetically susceptible individuals. CD occurs in about 1–3% of the population, in most areas of the world, with many cases being undetected.1
Anthropometric measures in subjects with CD were reported previously in several population based studies. Children with CD were shown to have lower weight and Body Mass Index (BMI) compared with the general population.2–4 In contrast, many adult patients with CD have a high or normal BMI.5–8
In contrast to data on weight, the evidence regarding final adult height of patients with CD is inconclusive and mostly relies on relatively small cohorts of patients. The adult height of children with CD is influenced by the compliance to gluten-free diet (GFD)9 while children diagnosed at early age exhibit slow and sometimes incomplete catch-up growth.10 While a study from 2002 reported that in patients with CD diagnosed as adults, both men and women were shorter than controls,11 in recent studies,12–14 final height of patients with CD was similar to the general population. However, few exceptions were noted in these studies: Adult height of male patients with CD was demonstrated to be inversely related to the age at diagnosis13 while in a subsequent study adult height was reduced only in a subgroup of an older birth cohort.14 The most recent study reported that adult men with CD, who had attained final height before diagnosis, are shorter relative to the general population.15
In the present study, we aimed to investigate the impact of CD on anthropometric measures, including BMI and height at late adolescence, and to determine whether patients with CD diagnosed during childhood are shorter than the general population. We also aimed to analyse the prevalence of CD over time and the socioeconomic, demographic and ethnic factors associated with diagnosis of paediatric CD.
Materials and methods
The majority of Jewish Israeli adolescents are required to attend army recruitment centres for an obligatory medical board examination close to the age of 17 years to assess their eligibility for military service. Comprehensive data were available from 1988 onwards. Strictly religious women who were exempt from duty were excluded. An estimate of potential female recruits not examined can be extracted from the numerical difference in the total number of men to women in the cohort. Assuming the genders are equally represented in the population, approximately 346 000 girls may not have been included during the 28 years of the study.
Ascertainment of coeliac cases
The diagnosis of CD was established and confirmed by a medical recruitment board composed of three physicians according to accepted criteria. The medical information including the diagnosis of CD was retrieved from detailed medical history documents provided by the subject's family physician for each individual case. The diagnosis of CD required a confirmation document signed by a gastroenterologist. Patients encountered from 1988 to 1990 were diagnosed according to the diagnostic criteria published in 1970.16 Patients encountered from 1991 to 2012 were diagnosed according to European Society for Pediatric Gastroenterology, Hepatology and Nutrition diagnostic criteria published in 199017 and from 2013 according to the revisited guidelines published in 2012.18
Covariate baseline adolescent data included date of birth, country of birth, parental country of origin and socioeconomic status. The socioeconomic status was ascertained from the Israeli Central Bureau of Statistics (CBS) based on a 1–10 scale and grouped into three categories: (low 1–3, medium 4–6 and high 7–10). The data collected by the CBS are ecological and include multiple variables (income, size of housing in m2/person, number of children in the family, number of cars, years of schooling and data about taxes). The definition of Western countries includes Europe (excluding the republics of the former Soviet Union), North America, Australia and New Zealand, and Eastern countries include Asian (mainly Arab) and African countries including Ethiopia. Subjects originated from the Republics of the former Soviet Union were analysed separately. Data for country of birth were obtained from the Israel Defense Forces enlistment database based on the Israeli Interior Ministry registry. For the 2 001 353 adolescents included in the analysis, data were missing for parental origin in 21% of the subjects.
During the medical encounter all adolescents had their weight and height measured and their BMI calculated. In order to assess the proportion of subjects with height and BMI <third centile, we used the Centers for Disease Control and Prevention growth charts.19
Continuous variables were evaluated for normal distribution using histogram, Q-Q plots and Kolmogorov-Smirnov test and reported as median (IQR) for non-normally distributed variables or mean (SD) for normally distributed variables. Trends in prevalence of CD, country of birth, country of origin, education, socioeconomic status and anthropometric measures were tested using a linear regression model. Continuous variables were compared using paired t-test while categorical variables were compared using χ2 test or Fisher's exact test. Correlation between continuous variables was evaluated using Pearson's correlation coefficient and for ordinal variables (non-normally distributed) Spearman's r correlation coefficient. Univariate analysis was used to evaluate the association between categorical variables. Multivariable logistic regression was used for analysing the association between CD and multiple variables which were identified in the univariate analysis as associated with CD with significance of <0.1. After controlling for potential confounders, the association between CD and other variables was analysed using multivariable logistic or linear regression as appropriate. The first block in each regression included age, gender and CD. The second block included potential confounders which were selected for inclusion in the regression using the forward method (country of origin (grouped as Israel, West, former USSR, Asia, Africa and Ethiopian), socioeconomic status (high, medium and low) and anthropometric measures). The linear regression was tested to meet the assumptions: normal distribution of the residuals, lack of multicollinearity and homoscedasticity. p<0.05 was considered as statistically significant. SPSS V.21.0 was used for all statistical analyses.
The sociodemographic characteristics of the entire cohort as well as the characteristics of the cases diagnosed with CD are provided in table 1. Overall, 10 566 cases of CD were identified out of 2 001 353 persons examined (0.53%). Median age at the time of examination was 17.1 years (IQR, 16.9–17.4 years). Boys accounted for 58% of the cohort. Distribution of age at examination between the entire cohort and the CD cohort was similar. Socioeconomic status of the whole population was distributed as follows: 3% low, 50% medium and 47% high.
Anthropometrics of subjects with CD
Univariate and multivariable analyses of anthropometric measures at median age of 17.1 years are shown in table 2. Prior to multivariable logistic regression both boys and girls diagnosed with CD had significantly lower weight and BMI compared with non-CD subjects (p<0.0001) (table 2). Following multivariable logistic regression the statistical significance was maintained only for boys (p=0.006 for weight, p=0.02 for BMI).
Univariate analysis demonstrated that both boys and girls with CD had significantly lower height compared with non-CD subjects (p<0.0001). Statistical significance was maintained following multivariable logistic regression only in girls (p=0.017).
The proportion of subjects with BMI <third centile was significantly higher in subjects with CD compared with non-CD subjects in both boys and girls (boys: 2.9% vs 4.5%, p<0.0001; girls: 3% vs 3.8%, p=0.005).
The proportion of subjects with height <third percentile differed in girls (2.9% in non-CD vs 3.8% in CD, p=0.003), with a similar proportion in boys (3% in non-CD and 3.1% in CD, p=0.2).
The prevalence of diagnosed CD
The prevalence of diagnosed CD at age 17.1 years was available from the year 1998 throughout 2015 as data prior to 1998 did not consistently include the entire population (figure 1). The prevalence increased from approximately 0.5% to 1.1% over the analysed time period (table 3).
Factors associated with diagnosis of CD
Table 1 presents the univariate and multivariable analyses for the sociodemographic variables associated with the diagnosis of CD among the whole cohort (1 990 787 non-CD cases; 10 566 cases of CD). The multivariable analysis shows that independent factors associated with diagnosis of CD were country of birth: Compared with subjects born in Israel, CD diagnosis was significantly less prevalent in subjects born in Asia (OR, 0.54, 95% CI 0.35 to 0.84, p=0.005), Africa (OR, 0.42, 95% CI 0.35 to 0.51, p<0.0001) and former Soviet Union (OR, 0.66, 95% CI 0.64 to 0.73, p<0.0001). Parental country of origin showed similar trends with significant lower prevalence of CD in subjects with Asian, African and former Soviet Union maternal or paternal origin. Lower socioeconomic status was significantly associated with reduced CD diagnosis (Low: OR, 0.54, 95% CI 0.43 to 0.73, p<0.0001; medium: OR, 0.68, 95% CI 0.64 to 0.72, p<0.0001; Reference: high).
In this study, we had the opportunity to assess anthropometric measures, prevalence trends and associated sociodemographic factors in a national cohort that included the majority of the Jewish population in Israel near the end of adolescence.
Our most important observation, analysing this large population based cohort of adolescents diagnosed with CD, is that at the age of 17 years which represents near-adult height for the entire cohort, only female subjects with CD are shorter than their healthy counterparts with a mean difference of <1 cm which could be regarded as marginally meaningful, if at all. Similarly, stunted growth (defined as height <third centile for age) was significantly more common only in girls with CD. Even though the median height of our cohort may not represent true final height, the narrow IQR indicates that the vast majority of subjects was measured in a narrow time frame around 17 years thus representing measurements that should, at least, correspond to adult height, particularly for girls who complete puberty at younger age.
So far, data regarding final adult height of patients with CD are scarce and inconclusive. With few exceptions, most studies, based on relatively small cohorts, did not demonstrate that final height is affected by CD diagnosis.12–14 In contrast, Cosnes et al11 studied 184 patients with CD, aged >18 years, diagnosed as adults, and found both men and women to be shorter than the age-matched control group. Patients who had undiagnosed symptomatic CD during childhood had a higher prevalence of short stature. Similarly, Sonti et al15 demonstrated that men with CD diagnosed in adulthood were shorter than men in the general population. So far, only one small study performed by Carbone et al20 showed that adolescent patients with CD were shorter than controls irrespective of gender. Overall, some reports showed reduced height only in men,11 ,15 ,21–22 whereas others showed reduced height only in women.23 A possible explanation for the gender difference, found also in our study, is that girls complete puberty at early age, thus at the time of CD diagnosis have fewer years for catch-up growth in comparison to boys with CD.
The pathogenesis of short stature in paediatric CD remains obscure. A plausible explanation involves growth hormone (GH) axis dysfunction,24 which might be driven from malnutrition which by itself can lead to abnormal GH levels. Furthermore, increased serum concentrations of the proinflammatory cytokines such as interleukin 6 and TNF α may also play a role.25 However, these changes should normalise upon the institution of GFD, coinciding with catch-up growth thus diminishing detrimental effects on growth.25
Our national cohort enabled us to examine other anthropometric findings in adolescents with CD and to demonstrate that male adolescents with CD have significantly low weight and BMI as compared with non-CD whereas female adolescents do not. Studies looking at weight and BMI in patients with CD imply on an apparent distinction between paediatric and adult populations. While data from paediatric studies clearly show a tendency towards low weight and BMI,3 ,4 adult studies indicate that many adult patients with CD have a high or normal BMI,2 ,5–8 especially in those who adhere closely to the diet.26 Our study cohort may represent a point in time which lies in-between childhood and adulthood; nevertheless it is evident that these variables are low in male patients with CD while entering adulthood. Another interesting observation is that underweight (defined as BMI <third centile for age) was more common in subjects with CD of both sexes. This finding is in disagreement with the report by van der Pals et al2 who demonstrated a lower rate of underweight in patients with CD at the age of 12 years but in concordance with the data showed by Brambilla et al3 delineating that a significant proportion of underweight children seem to remain underweight when on a GFD. Low BMI in a subset of patients with CD might be attributed to either low adherence to GFD with its deleterious consequences or to catch-up failure in some patients who were diagnosed at younger age.
Although not the primary outcome measure of our study, our cohort enabled us to examine trends in the diagnosis of CD as well as the effect of ethnicity and sociodemographic factors. We were able to show that the prevalence of diagnosed CD in Jewish adolescents increased gradually during the past 20 years reaching approximately 1.1% in 2013–2015. This finding is in accordance with previous reports demonstrating a continuous rise in the incidence of CD since the 1950s.27 ,28 The prevalence of CD is thought to be dependent on several factors including genetic predisposition, infant feeding patterns and other environmental risk factors. However, recent prospective observational studies have failed to confirm an association between breastfeeding patterns and CD29–31 or an association between age of gluten introduction and the development of CD.32 Thus, awareness of the disease among physicians and patients, the availability of serological tests and the changing of diagnostic criteria are speculated as the main determinants of this rising prevalence. We could also confirm the female gender dominance among diagnosed adolescents found in previous studies.33
Our findings demonstrate a clear association between Western origin and the likelihood for CD diagnosis. This linkage was maintained in respect to both country of birth and parental country of origin. Our results are in agreement with reports suggesting that CD is more common in Caucasians compared with African-Americans or Hispanics,1 ,34 and that CD is rare in people from sub-Saharan Africa.35
Another observation was the association of high socioeconomic status with CD diagnosis. This finding is consistent with several previous studies emerging from the UK36 and Sweden.37 Similar to our findings, Burger et al38 have recently described, that Dutch patients diagnosed in childhood were more often from higher socioeconomic status compared with patients diagnosed later in life. A possible explanation for this association is that individuals from more deprived areas are less likely to have high standard medical care and thus be potentially less likely to be tested for CD.39 Another potential explanation could be related to the hygiene hypothesis. In accordance with this hypothesis early exposure to infections and other antigens may protect children of lower socioeconomic groups from developing allergic and autoimmune diseases such as CD.40 Nevertheless an association between early infections and a reduced risk for CD in still debated.30
Our study has several limitations. First, the cohort was based on patients with diagnosed CD. Therefore, the true prevalence of CD in our population is higher than described, and the results are relevant only for diagnosed CD. Moreover, we could not retrieve valuable information such as age at diagnosis as well as adherence to GFD. Also, the prevalence data might be confounded by the fact that more than 300 000 girls may not have been included during the 28 years of the study.
Nevertheless, our study strengths stem from being a large cohort of more than 10 000 Jewish subjects with CD and the ability to compare anthropometric measurements in patients with CD with measurements in the majority of the Jewish population at the same age.
In conclusion, we found that adolescent boys diagnosed with CD have significantly lower weight and BMI whereas girls have lower height compared with the general population. However, the clinical relevance of the small differences suggests that when CD is diagnosed during childhood, the overall final weight and height are not severely impaired. Our cohort reinforces the observed increase in diagnosed CD, and the positive correlation with a higher socioeconomic status and Western origin.
Contributors AA: conceptualised and designed the study, participated in data analysis, drafted the initial manuscript and approved the final manuscript as submitted. YF-N and YL-W: participated in data analysis, participated in drafting the manuscript and approved the final manuscript as submitted. DT: extracted the epidemiological data, revised the manuscript and approved the final manuscript as submitted. AA and ZL: participated in conceptualising and designing the study, revised the manuscript, and approved the final manuscript as submitted. LHK: participated in conceptualising and designing the study, reviewed and revised the manuscript, and approved the final manuscript as submitted. RS: participated in conceptualising and designing the study, critically reviewed the manuscript, and approved the final manuscript as submitted.
Competing interests None declared.
Ethics approval The Institutional Review Board of the Israel Defense Forces Medical Corps approved the study.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.