Objectives Although gluten avoidance is thought to be common among New Zealanders, the prevalence of gluten avoidance and of actual coeliac disease (CD) in children is uncertain. Our aims were: (1) to determine the prevalence of doctor-diagnosed CD and of gluten avoidance in New Zealand children; and (2) among children without CD, to identify independent predictors of gluten avoidance.
Design The New Zealand Asthma and Allergy Cohort Study has detailed information on participants' demographic, pregnancy-related and neonatal factors. The authors surveyed parents regarding their child's history of lactose intolerance and gluten-related issues (eg, gluten avoidance, history of wheat or gluten allergy in first degree relatives, testing and doctor diagnosis of CD). After excluding children with doctor-diagnosed CD, the authors identified independent predictors of gluten avoidance.
Results Among 916 children, most (78%) were of European ethnicity. The authors identified nine (1.0%, 95% CI 0.5% to 1.9%) who had doctor-diagnosed CD, while 48 (5.2%, 95% CI 4.0% to 6.9%) avoided gluten. Among children without diagnosed CD, significant independent predictors for gluten avoidance were Christchurch site (OR 2.2, 95% CI 1.02 to 4.7), prior testing for CD (OR 9.0, 95% CI 4.1 to 19.5) and doctor-diagnosed lactose intolerance (OR 5.2, 95% CI 2.0 to 13.9).
Conclusions CD affected 1% of these New Zealand children, but 5% reported gluten avoidance. The predictors of gluten avoidance in children without doctor-diagnosed CD suggest important regional differences in community belief or medical practice regarding implementation of gluten avoidance and the contributory role of non-specific subjective abdominal complaints.
Statistics from Altmetric.com
Coeliac disease (CD) is a genetically determined autoimmune enteropathy with a known trigger of gluten and related proteins, which are found mainly in wheat, barley and rye.1 2 Population screening studies have suggested a prevalence of approximately 1%, which makes CD one of the most common genetically determined diseases in humans.3,–,5 CD is also an often cited example of the ‘iceberg’ phenomenon because of the assumed large number of undiagnosed CD cases in the general population.6 7 Although a study of biopsy-proven CD among adult New Zealanders8 9 found a similar prevalence (1%) as observed in other industrialised countries, a retrospective study during 1985–1992 suggested a CD prevalence of only 0.014% among New Zealand children.10 The current prevalence of CD in New Zealand children is uncertain.
What is already known on this topic
▶ Population screening studies on coeliac disease have suggested a prevalence of approximately 1% among European and US children.
▶ Adult New Zealanders have similar prevalence, but paediatric data are unavailable.
▶ Gluten avoidance among people without diagnosed coeliac disease is thought to be common among New Zealanders but the actual prevalence is unknown.
What this study adds
▶ The prevalence of doctor-diagnosed coeliac disease in New Zealand children is approximately 1%, similar to that among European and US children.
▶ The prevalence of gluten avoidance is five times higher than doctor-diagnosed coeliac disease.
▶ Predictors of gluten avoidance among children without coeliac disease suggest regional differences in gluten avoidance implementation and the contributory role of non-specific abdominal complaints.
Testing for CD in children usually follows the occurrence of non-specific abdominal symptoms such as abdominal pain, bloating or diarrhoea.11 These abdominal symptoms are also commonly found among other gastrointestinal disorders, such as irritable bowel syndrome and lactose intolerance.12,–,15 Despite highly sensitive and specific serological markers of testing for CD, the criterion standard for CD diagnosis remains small intestinal biopsy, which requires invasive upper endoscopic examination.
Management of CD requires life-long commitment to a gluten-free diet.1 Strict avoidance of gluten requires remarkable effort from both the child and their family, which can be costly and difficult in settings where gluten-free products are not widely available. Because life-long commitment to gluten-free products is required to maintain disease remission and a good quality of life,14 16,–,18 accurate and reliable tests are crucial for the diagnosis of CD. Undiagnosed cases may lead to adverse long-term consequences such as small bowel malignancies,19 osteoporosis20 and other autoimmune diseases.21 22 On the other hand, incorrect recommendation for strict gluten avoidance can potentially lead to significant adverse effects on psychosocial dynamics and quality of life23 as well as an increased public health burden.24
While anecdotal evidence suggests that up to 5% of the Australians without a CD diagnosis avoid gluten, the prevalence of gluten avoidance in New Zealanders is uncertain. Furthermore, the predictors of gluten avoidance among a population without a CD diagnosis have not been studied. Our objectives were: (1) to determine the prevalence of doctor-diagnosed CD and gluten avoidance in a birth cohort of New Zealand children; and (2) to identify predictors of gluten avoidance among children without doctor-diagnosed CD.
The New Zealand Asthma and Allergy Cohort Study25 is a prospective birth cohort study comprised of 1105 infants recruited in Wellington and Christchurch between 1997 and 2001. Briefly, expectant mothers were recruited by a random sample of midwives, health professionals who provide almost all maternity care in New Zealand. At birth, midwives or study nurses collected the newborn's anthropometric details. Study nurses conducted face-to-face, structured interviews shortly after birth and at 3 months of age. The study was conducted according to the guidelines in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Wellington and Canterbury Regional Ethics Committees. All mothers gave written informed consent for their children to participate in this study to monitor their health and development and to undergo regular assessments in the form of interviews, questionnaires and a variety of specific tests.
Variables of interest
In addition to the study site of either Wellington or Christchurch, demographic information included the child's birth date, sex and ethnicity (European, Maori, Pacific and Others). Socio-economic status was assessed using the New Zealand deprivation index 2001 based on the child's home address at 3 months of age; the index ranges from 1 (affluent) to 10 (poor).26 Pregnancy-related factors (eg, maternal age at birth, smoking status during pregnancy), neonatal factors (eg, gestational age at birth, birth weight) and history of breastfeeding were collected.
In late 2009, we surveyed parents regarding gluten-related issues (eg, gluten avoidance in the child, history of wheat or gluten allergy in first degree relatives (mother, father and siblings), history of testing for CD including a blood test, intestinal biopsy and other forms of testing, and doctor-diagnosed CD) and their child's history of doctor-diagnosed lactose intolerance. The degree of gluten avoidance was assessed by parent-reported questionnaire with a four-point scale: 1, seldom; 2, sometimes; 3, often; 4, always. If the child was previously diagnosed with CD, responses regarding age of symptom onset and of diagnosis were also obtained.
We performed all analyses using STATA statistical software 11.1 (StataCorp, College Station, Texas, USA). The sample was described using mean (SD), median (IQR) and proportion (with 95% CI). Logistic regression was used to examine unadjusted associations between variables of interest and gluten avoidance (yes/no). For each characteristic, we reported OR with 95% CI. To adjust for potential confounders, we also performed a multivariable logistic regression using gluten avoidance as a dependent variable. A two-tailed p<0.05 was considered statistically significant.
Of 1105 initial participants, 916 (83%) provided complete responses to the gluten-related questions. The median age of children at the time of the survey was 10.6 years (IQR 9.9–11.2 years). The two study sites, Wellington and Christchurch, contributed similar numbers of participants. Most children were of European ethnicity (78%), with Maori children accounting for 13%, Pacific 4% and Others 5%. This ethnic breakdown is representative of New Zealand in the late 1990s when the cohort was assembled. Almost all children were born at gestational age ≥37 weeks (97%). With regard to the demographic, pregnancy-related and neonatal factors, the children in the present study (n=916) are similar to those in the overall cohort (n=1105).25 Table 1 shows the baseline characteristics and gluten-related issues of the cohort participants.
With regard to prior testing for CD, 60 (6.6%) children had at least one form of CD testing. All but one of the 60 children received at least one conventional test for CD (either a blood test or intestinal biopsy); some of these children received more than one form of testing. Five (0.5%) children had a biopsy performed, 57 (6.2%) had specific serological testing and seven (0.8%) had other forms of testing, that is, diet testing (one), hair testing (one), genetic testing (one), muscle biopsy (one), skin prick test (one) and unknown testing during a related hospitalisation (two).
Among 916 participants, we identified nine (1.0%, 95% CI 0.5% to 1.9%) children with doctor-diagnosed CD. The median child age at the time of doctor-diagnosed CD was 7.0 years (IQR 5.0–8.0 years), and eight of the nine children were of European ethnicity; the ninth child with CD was of Maori origin. All patients with doctor-diagnosed CD avoided gluten and had received a blood test for CD, but only 44% (4/9) had an intestinal biopsy performed.
Overall, we identified 48 (5.2%, 95% CI 4.0% to 6.9%) children who avoided gluten. Among 907 children without doctor-diagnosed CD, 39 (4.3%, 95% CI 3.1% to 5.8%) avoided gluten. Unadjusted predictors of gluten avoidance among children without diagnosed CD are shown in table 2.
In the multivariable model, as shown in table 3, independent predictors of gluten avoidance are Christchurch site (OR 2.2), prior testing for CD (OR 9.0) and doctor-diagnosed lactose intolerance (OR 5.2).
In an ongoing birth cohort of New Zealand children, we found that approximately 1% had doctor-diagnosed CD. The median age at the time of diagnosis was 7.0 years. All nine children with doctor-diagnosed CD had undergone blood testing for CD but only 44% had an intestinal biopsy performed. The prevalence of CD in our study was similar to that in other industrialised countries,3,–,5 as well as a previous study in the adult New Zealand population.9 The only prior publication on CD prevalence among New Zealand children,10 which was conducted more than two decades ago, showed a significantly lower prevalence of CD (0.014%) compared to our study, a difference that suggests significant under-ascertainment in the earlier study or significant increases in the incidence and general awareness of CD. For our study, midwives were randomly selected from various facilities and encouraged to enrol all expectant mothers. We believe that this approach limited selection bias, represents most pregnancies and improves generalisability to the wider New Zealand population.
We also found that approximately 5% of the cohort avoided gluten, a prevalence that was five times higher than that of actual doctor-diagnosed CD. After excluding children with doctor-diagnosed CD (all of whom avoided gluten), 4.3% of the remaining children avoided gluten. We determined that among the participants who avoided gluten without doctor-diagnosed CD, that Christchurch site, prior testing for CD and doctor-diagnosed lactose intolerance were significant predictors in both unadjusted and adjusted logistic regression models.
In an outpatient setting, the child who presents with non-specific abdominal symptoms may undergo testing for CD. Currently, serology and intestinal biopsy remain crucial for CD diagnosis. All children with doctor-diagnosed CD in our study had a blood test performed and avoided gluten but less than half of the group underwent intestinal biopsy. Furthermore, among 907 children without diagnosed CD, 26/39 (67%) children who avoided gluten had no prior testing for CD. The lack of a definitive diagnosis via intestinal biopsy may have led to premature or inappropriate implementation of gluten avoidance in the absence of biopsy-confirmed CD. Thus, the prevalence of gluten avoidance among children without biopsy-confirmed CD (as compared to doctor-diagnosed CD) might be higher than we report due to potential false-positive diagnoses of CD.
Non-specific abdominal symptoms are recognised in CD and lactose intolerance.13 The prevalence of lactose intolerance in the world population varies between geographical locations, with estimates of 4%–20% in most Europeans to 80%–98% in most Asians.27 For our study, the survey was obtained by parental report when the child was aged around 10 years, and we found most children with doctor-diagnosed lactose intolerance had a history of diarrhoea. Most children were of European ethnicity with a relatively similar prevalence of lactose intolerance (4.4%) compared to prior studies,27 28 although we have no information regarding testing. Therefore, we believe that children with doctor-diagnosed lactose intolerance could be a combination of temporary lactose intolerance (after enteritis) and genetically predisposed hypolactasia individuals.
Our study also showed doctor-diagnosed lactose intolerance as one of the significant predictors of gluten avoidance. Prior studies have shown a higher prevalence of lactose intolerance among CD cases29 and vice versa.30 However, among 40 children with doctor-diagnosed lactose intolerance in our study, only six of them had at least one prior test for CD. Some of these children could have had underlying CD; most were not tested before they initiated gluten avoidance. Both gluten and lactose are commonly found in the routine diet of children in industrialised countries (eg, in bread, breakfast cereals and pre-prepared meals). Therefore, if children with doctor-diagnosed lactose intolerance avoid gluten-containing products as well, this might potentially cause an over-estimation of gluten avoidance by parental report.
CD and lactose intolerance also are associated with irritable bowel syndrome-type symptoms.12,–,15 Recently, Biesiekierski and colleagues described non-coeliac gluten intolerance in their randomised, double blind, placebo-controlled re-challenge trial.31 The authors showed that gluten caused gastrointestinal symptoms among 34 adults with irritable bowel syndrome and without CD. With this caveat in mind, we still believe that implementation of gluten avoidance was more common in children with doctor-diagnosed lactose intolerance because of the commonly shared non-specific abdominal symptoms among these gastrointestinal disorders.
Our study showed significant regional/geographical differences in gluten avoidance between Christchurch and Wellington. Because these cities are located only 400 km apart, in the same country, this finding is difficult to explain on biological grounds. The cohort participants were enrolled in the same period and followed by the same methodology. Interestingly, access to gluten-free foods in these two areas may differ: anecdotal evidence suggests that Christchurch has a streamlined system with gluten-free products delivered directly to patients, while Wellington patients need to get their practitioners' prescription and take it to a pharmacy. We hypothesise that this finding might be due to important regional differences in community belief or medical practice regarding the implementation of gluten avoidance. Furthermore, we did not find that either ethnicity or socio-economic status was a significant predictor of gluten avoidance (table 2). Future population-based and surveillance studies on gluten avoidance, with a focus on the families' and healthcare practitioners' knowledge and beliefs about gluten avoidance as well as resource utilisation, would be required to further explain this finding.
Such a study would also need to focus on commonly used testing for CD, including serology and small intestinal biopsy. Among our study participants, 6.6% of the children had a history of at least one form of CD testing, especially serology (95% of this group). Five per cent of the children who underwent testing had non-validated tests for CD (eg, hair testing, muscle biopsy or skin prick test) which should be discouraged by medical professionals. Either way, this prior testing for CD significantly predicted gluten avoidance in New Zealand children without doctor-diagnosed CD. It would have been interesting to track the results of these tests over time, in conjunction with information regarding gluten intake, but the study survey only gathered the presence (yes/no) of CD testing. Unfortunately, we do not have details on which serological tests were carried out. In children with equivocal results, on either serological testing or small intestinal biopsy, clinicians may propose a trial of gluten avoidance. Furthermore, recent evidence showed that individuals with ‘silent CD’ (ie, asymptomatic cases with abnormal small intestinal biopsy) would attain benefits from early initiation of a gluten-free diet.32 An investigation on care giver's and healthcare professional's decisions to implement potentially long-term gluten avoidance, and how this relates to CD testing results, would be valuable.
This study has potential limitations. The cohort was originally designed to study risk factors for allergy and asthma in New Zealand children. The key data in the present report (ie, gluten-related issues, doctor-diagnosed CD and doctor-diagnosed lactose intolerance) were obtained in a single cross-sectional survey when participants were approximately 10 years old. Therefore, the time sequence of testing for CD and implementation of gluten avoidance could not be established with our survey. This study design may have led to some recall bias and limits our ability to determine the causality of observed associations. The relatively small number of cases who avoided gluten without doctor-diagnosed CD (n=39) limited our ability to perform multivariable modelling; the even smaller number of cases of doctor-diagnosed CD (n=9) precluded such analyses. Nevertheless, these data provide the first population-based data in New Zealand children on these important nutritional topics.
In summary, we found that the prevalence of doctor-diagnosed CD in New Zealand children is similar to that in adult New Zealanders and the population prevalence in other industrialised countries.3,–,5 9 We also found that gluten avoidance was five times more common than the prevalence of doctor-diagnosed CD. The independent predictors of gluten avoidance in New Zealand children without doctor-diagnosed CD included Christchurch site, prior testing for CD and doctor-diagnosed lactose intolerance. These results point to important regional differences in community belief or medical practice regarding implementation of gluten avoidance in the absence of doctor-diagnosed CD. The findings also suggest a contributory role of non-specific abdominal symptoms commonly shared with lactose intolerance and possible non-coeliac gluten intolerance. Our observational data support the initiation of future prospective, population-based studies to better define predictors of gluten avoidance among children without CD diagnosis, and the long-term costs and benefits of this practice.
The authors thank the midwives in Wellington, Porirua and Canterbury for assistance with recruitment, and all the families and children for their participation. The authors also thank the New Zealand Asthma and Allergy Cohort Study Group: J Crane, C Dench, M Duignan, MJ Epton, D Fishwick, P Fitzharris, T Ingham, V Irvine, R Kelly, P Lampshire, J Lane, P Leadbitter, C MacDonald, F McCartin, S McLeod, A Nicholson, P Pattemore, K Roff, G Sawyer, R Siebers, GI Town, K Wickens, H Wilson and K Withell.
Funding This work was supported by grants from the Health Research Council of New Zealand, the David and Cassie Anderson Bequest (Wellington, New Zealand) and the Food Allergy & Anaphylaxis Network (Fairfax, Virginia, USA).
Competing interests None.
Ethics approval The Wellington and Canterbury Regional Ethics Committees approved the New Zealand Asthma and Allergy Cohort 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.