Diagnostic Delay of Celiac Disease in Childhood

This cross-sectional study evaluates the diagnostic delay of celiac disease in childhood in Italy, and to assess factors associated with this delay.


Introduction
Celiac disease (CD) is an immune-mediated, gluten-sensitive enteropathy. 1 CD has been described worldwide, with a high prevalence in both childhood and adulthood, especially in White people (close to 1%); nonetheless, many patients remain undiagnosed in the real-life setting. 1,2Indeed, a strict and lifelong withdrawal from dietary gluten can revert villous atrophy in most cases and may improve quality of life. 3,4 has a wide clinical spectrum 5 that can make its diagnosis a great challenge, especially in children.In fact, CD in children may be asymptomatic, may cause mild gastrointestinal symptoms, or may present a severe clinical picture, including anemia, failure to thrive or weight loss, dysproteinemia, diarrhea, dehydration, and electrolyte imbalance.This wide heterogeneity may lead to a delayed diagnosis and may also increase the risk of misdiagnosis. 60][11] Hence, a timely diagnosis of CD is warranted. 33][14][15][16][17][18] A wide range of diagnostic delay has been reported, reaching up to 10 years in some cases.
The 2012 European Society for Paediatric Gastroenterology Hepatology and Nutrition (ESPGHAN) guidelines recommend using quantitative determination of antitissue transglutaminase IgA (TTG-IgA) as a first-line screening.If the antibody titer is above 10 times the upper limit of normal in a symptomatic patient, the guidelines recommend a second check of TTG-IgA, along with antiendomysial antibodies (EMA) and human leukocyte antigen (HLA)-DQ2/DQ8 detection.This strategy reduced the need for an upper gastrointestinal endoscopy by 30% to 50%. 19In 2020, even less stringent criteria were proposed, also including asymptomatic patients. 20Recently, a biopsyfree approach has even been suggested in the diagnosis of adult CD. 21arting from these premises, we hypothesized that: (1) there is still a substantial diagnostic delay for CD; (2) the presenting signs and symptoms can be scarce, tricky, and subtle, thus contributing to the diagnostic delay; and (3) the diagnostic delay in childhood is smaller than the one in adulthood, since in most cases there is no need for a biopsy in the former.Therefore, the main aim of this study was to investigate the diagnostic delay of CD in an Italian multicenter cohort of pediatric patients, highlighting their presenting symptoms, signs, and any concern potentially associated with the delay.

Study Design, Participants, and Collected Variables
This was a retrospective observational multicenter study.At first, 16 Italian gastroenterological, tertiary referral, outpatient pediatric clinics were invited to participate, and 13 of them accepted.The 13 participating centers are located across the country, from Northern to Southern Italy, thus providing a broad overview of pediatric CD in the whole country.The ethics committee of Pavia

Statistical Analysis
Univariable and multivariable logistic regression models were fitted using misdiagnoses as dependent variables and sign, symptom, and clinical data as independent variables.Parameters with a P value less than .20 at univariable analysis were included in multivariable models.The area under the model receiver operating characteristic curve was computed to assess discrimination.
Univariable and multivariable generalized linear regression models (with gaussian family and identity link) were fitted using diagnostic delays as dependent variables and sign, symptom, and clinical data along with clinical characteristics as independent variables.Delays were log-transformed for the purpose of the analysis.Variables with a P value less than .20 at univariable analyses were included in the multivariable models.Logistic regression was used to model extreme diagnostic delays; the same variables as for the corresponding delay model were used.Statistical analyses were performed using Stata statistical software, version 17 (StataCorp).A 2-sided P-value less than .05was considered statistically significant.Data were analyzed from January to June 2023.Interestingly, many differences according to the age group at diagnosis were found (eTable 1 in Supplement 1).Patients who received a first diagnosis when aged less than 3 years (650 patients [20.5%]), compared with all others, had a significantly higher rate of multiple symptoms, gastrointestinal symptoms, weight loss, failure to thrive, dysproteinemia, major presentation of CD, and need for hospitalization.Some differences in the clinical presentation were also found between sexes: the prevalence of autoimmune disorders, a history of CD in a first-grade relative, and the presence of abdominal pain were more frequent in female patients, while diarrhea was more common in male patients.Notably, the rate of nonatrophic duodenal lesions (Corazza-Villanacci 0-A) was, overall, relatively high, with a decreasing trend across the age groups.eTable 2 in Supplement 1

Demographic and Clinical Data
shows the characteristics of patients who were diagnosed with a biopsy compared with those who
variables included in the multivariable analysis on the diagnostic delay are reported in Figure 1.
Having received a first diagnosis of CD when aged less than 3 years rather than between 3 and 18 years was associated with a lower diagnostic delay, both overall (median

Associations With Extreme Diagnostic Delay
We observed an extreme diagnostic delay in 587 patients (18.5%

Discussion
The diagnostic delay of CD in pediatric patients in Italy found in this cross-sectional study seems lower than that reported in the literature.Our study, compared with those already published (Table 1) is currently the largest in the pediatric setting that we know of, and includes a well-characterized cohort of pediatric patients with CD.We have herein reported the diagnoses made after the introduction of the biopsy-sparing ESPGHAN guidelines of 2012.Moreover, this is one of the few studies that distinguishes preconsultation and postconsultation delay and focuses on factors associated with cases of extreme diagnostic delay.Our results confirm and corroborate previous findings on major symptoms as they frequently present in childhood.Also, the biopsy-sparing approach was found to be burdened by a lower overall delay, possibly due to time saved by making an appointment for the upper gastrointestinal endoscopy and for the histopathological report.
In fact, the diagnostic delay of CD in children is here confirmed 18 to be shorter than in adults by comparison with data from a national multicenter study by our group 11 that includes the same period of years and a similar method of data description.In the adult study, overall, preconsultation, and postconsultation diagnostic delays were 8, 3, and 4 months, respectively, compared with 5, 2, and 1 in the pediatric study.The diagnosis of CD in children is usually simpler because the general pediatrician and the hospital pediatrician would lead the process and the patient does not usually consult a series of other specialists for different symptoms.A broad adherence to the international guidelines for the diagnosis of CD should probably be considered another important element of this success.The biopsy-sparing policy was likely advantageous in lowering the mean time to diagnose CD, as the time needed for making the appointment and analyzing the biopsy is completely absent.
The counterpart of the application of ESPGHAN guidelines is that a retrospective study based on clinical records, such as the present one, would be lacking many theoretically interesting data, such as HLA-typing, EMA and AGA levels, endoscopic findings, and histological grading.
Given the shortness of diagnostic delay found in this study, we consider the results regarding extreme diagnostic delays more important.Moreover, there are some factors that affect diagnostic delay of CD both in adults and in children, such as a previous misdiagnosis. 11,26The rate of false diagnoses before CD diagnosis is lower in the pediatric population as compared with the adult population.Luckily, the diagnostic delay of CD in children is not burdened by the risk of typical complications of CD (ie, refractoriness and enteropathy-associated T-cell lymphoma), but can imply a severe malabsorption syndrome anyway, with the need for hospitalization and parenteral nutrition.
This study, for its magnitude, has the merit of depicting the current clinical presentation of Italian pediatric patients with CD.The fairly high rate of detection of nonatrophic lesions in the group of patients who underwent endoscopy is similar to those reported from a review of the literature. 20ere are also some clinical similarities in the presentation of CD in adult and pediatric Italian populations, such as a very high prevalence of red blood cell count alterations and particularly microcytic anemia and gastrointestinal symptoms.The prevalence of autoimmune diseases seems to be lower in the pediatric population compared with adults, as expected, and this tended to be associated with shorter diagnostic delay, though at the limit of significance.This result may suggest the need for enforcing screening in patients with multiple autoimmune comorbidities.A family history of CD and an asymptomatic presentation were associated with shorter delays in both age ranges, because patients with these characteristics are likely to have been diagnosed by a casefinding strategy (due to the family history or the presence of conditions associated with CD).
A very interesting finding is that younger patients reached the diagnosis in less time, especially patients younger than 3 years for whom extreme diagnostic delay was rare.These data confirm a similar trend observed in another recent European study 19 and would probably have been an unexpected result in the past, when serological tests were not as reliable as they are today.
The higher prevalence of failure to thrive in patients who were burdened by a higher diagnostic delay may have different explanations.Failure to thrive could be attributed to diseases other than CD, and milder cases could be overlooked.Notably, it has already been reported that the presence of a classical malabsorption syndrome in adult patients was associated with a longer delay. 9,26Given the fact that the delay depends mostly on patients, this means that some patients and their families may delay seeking medical care.It should be stressed that in Italy there is universal coverage of health care expenses, which can also be full coverage in cases of medium to low family income, so pediatric treatments are universally available to anyone.In other countries, priority in the health care system may depend on or be influenced by the economic status of the individual, and hence our results may not be applicable to other settings.
The difference found between sexes on overall and preconsultation delays did not seem clinically relevant (5 vs 4 months and 2 vs 1 month).On the other hand, it is easy to hypothesize that gastroesophageal reflux or other signs or disease associations, all associated with a longer diagnostic delay, can be initially misinterpreted and prompt the assessment of specialists other than pediatricians, with a consequently longer diagnostic process.It is notable that osteopenia, which is more typically related to adult CD, was reported in 44 cases.Indeed, older children are more likely to get a workup for osteopenia than younger children, and some rare cases of osteopenia in infants could be missed.

Limitations
Our study has some limitations that must be mentioned.First, the retrospective nature and telephone interview are open to some biases; for example, the date of symptom onset may be inaccurate, and some physician consultations could have been missed.For some variables, the percentage of missing data was too high, so they were removed, such as those regarding socioeconomic aspects.Also, since White individuals were the most represented, our data cannot be generalized to other ethnicities.The generalizability of our results is also limited to those countries adhering to the ESPGHAN guidelines.Indeed, the setting is tertiary referral outpatient clinics, and therefore these findings cannot address primary care practice.A subanalysis depending on the actual modality of diagnosis could not be made either, as this would have introduced substantial biases, given that the precise reason for the endoscopy and the single diagnostic process could not be clearly retrieved in all cases; this was outside the scope of this research.The latest 2020 ESPGHAN guidelines were not applied here since patients diagnosed after 2020 were not included due to the COVID-19 pandemic.Additionally, we are aware that both the preconsultation and postconsultation delays could be affected by the time needed for making an appointment or other system-level factors that could not be captured.Regardless, this is still the largest study we know of looking at diagnostic delay in pediatric CD, in which a thorough analysis was made and patients were enrolled from almost all Italian regions, thus providing a global picture of the entire country.

Conclusions
In this cross-sectional study of pediatric CD, the diagnosis was usually made in a timely fashion, except for some cases in which both gastrointestinal and less common symptoms or disease associations led to a prior misdiagnosis and diagnostic delay.A better awareness of pediatric CD proteiform manifestations could be helpful in early recognition of CD.

Table 1 .
Diagnostic Delay (DD) in Children With Celiac Disease (CD) According to the Available Retrospective Studies Abbreviation: NA, not assessed.JAMA Network Open | PediatricsDiagnostic Delay of Celiac Disease in Childhood JAMA Network Open.2024;7(4):e245671.doi:10.1001/jamanetworkopen.2024.5671(Reprinted) April 9, 2024 3/14 Cura a Carattere Scientifico Policlinico San Matteo) approved the study protocol, and this approval was extended to all the ethics committees of the other participating centers.Either the patient or the caregiver, depending on the age at the time of enrollment, were asked to provide written informed consent to take part in the study.The necessity of a signed consent was waived for those patients who could not be reached, as all data were anonymous.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Downloaded from jamanetwork.comby guest on 04/15/2024 (Fondazione Istituto di Ricovero e

Table 2
reports the demographic characteristics and the clinical manifestations associated with the diagnosis of CD found in the whole cohort.Almost one-third of the patients (903 patients

Table 2 .
Demographic and Clinical Characteristics of the 3171 Patients With Celiac Disease (CD) (continued) compared with those who were diagnosed with a biopsy.Also, as shown in eTable 2 in Supplement 1, older patients were more likely to be biopsied, while those with gastrointestinal symptoms were less likely to be biopsied consistently with the guidelines.
a If the number of available data for a variable is different from 3171, it is reported.bBodymassindex is calculated as weight in kilograms divided by height in meters squared.cTheseinclude mood changes, learning disabilities, confusion, memory loss, depression, persecutory delusions, and psychosis.JAMA Network Open | PediatricsDiagnostic Delay of Celiac Disease in Childhood JAMA Network Open.2024;7(4):e245671.doi:10.1001/jamanetworkopen.2024.5671(Reprinted) April 9, 2024 6/14 Downloaded from jamanetwork.comby guest on 04/15/2024 get a biopsy For each level of each variable in the model, squares are the estimate of the difference in log delay, back transformed into ratios.Whiskers are 95% CIs.If these do not cross the null effect line, they are significant at the 5% level.R indicates ratio.Figure 2. Multivariable Analysis for Factors Associated With the Preconsultation, Postconsultation, and Overall Extreme DelayFor each level of each variable in the model, squares are the estimate of the odds ratio (OR).Whiskers are 95% CIs.If these do not cross the null effect line, they are significant at the 5% level.