Factors influencing diagnostic delays in celiac disease

Abstract

BACKGROUND

Celiac disease (CeD), an autoimmune disorder triggered by gluten ingestion, is characterized by non-specific clinical manifestations such as fatigue, abdominal pain, and nutritional deficiencies, often leading to substantial diagnostic delays. Prolonged delays (≥ 2 years from symptom onset) are associated with increased risks of complications like osteoporosis, small intestinal lymphoma, and reduced quality of life.

AIM

To estimate diagnostic delay prevalence and identify risk factors in Chinese CeD patients.

METHODS

We reviewed clinical records of 166 patients diagnosed with CeD from 2017 onward. Patient-attributed delays were measured from symptom onset to first consultation, while physician-related delays were measured from initial visit to diagnosis/treatment. Data on demographics, symptoms, time from onset to diagnosis, and laboratory results were analyzed. Logistic regression models were used to identify associations, while restricted cubic splines explored nonlinearities. Mediation analysis assessed the roles of intermediate factors in delayed diagnosis.

RESULTS

Delayed diagnosis (over 2 years from symptom onset) was observed in 42.2% of patients. Patients with diagnostic delay exceeding 5 years accounted for 18.7%. The mean interval from symptom onset to the first medical visit was 12.32 months, with an average of 20.57 months from the first visit to diagnosis. The time from first consultation to diagnosis significantly increased with prolonged delay (P < 0.001). Multivariate analysis showed that blood urea nitrogen (BUN) was an independent risk factor (OR = 1.29, 95%CI: 1.01–1.65, P = 0.038). A nonlinear association was observed between BUN and delayed diagnosis, with a threshold of 4.3 mmol/L; the risk significantly increased above this threshold (OR = 1.39, P = 0.04). Subgroup analyses indicated that the risk effect of BUN was stronger in females, non-classical CeD patients, Kazak ethnic group members, individuals without vitamin D deficiency/anemia, and those with Marsh III pathology (all P < 0.05). Mediation analysis revealed that folic acid deficiency and anemia mediated 11.9% (P = 0.028) and 13.0% (P = 0.044) of the effect of BUN on diagnostic delay, respectively.

CONCLUSION

Elevated BUN levels are independent predictors of diagnostic delay in CeD, with heterogeneity observed across gender, disease subtype, ethnicity, and pathological type. Clinicians should prioritize high-risk populations with BUN ≥ 4.3 mmol/L, particularly female patients with non-classical CeD and Kazak individuals, to reduce diagnostic delay.

Key Words: Celiac disease; Delayed diagnosis; Restricted cubic splines; Independent risk factors; Blood urea nitrogen

Core Tip: This study presents the largest single-center cohort of Chinese celiac disease (CeD) patients (Marsh ≥ II histopathology) leveraging Xinjiang’s regional referral advantage. Among 166 CeD patients, 42.2% experienced diagnostic delay (> 2 years). Blood urea nitrogen (BUN) was identified as an independent risk factor (OR = 1.29), with risk significantly increasing at ≥ 4.3 mmol/L. Subgroup analyses showed stronger associations of BUN with diagnostic delay in females, patients with non-classical CeD phenotypes, Kazak individuals, and other subgroups. These findings highlight BUN as a novel predictor and provide evidence to prioritize high-risk subgroups for timely screening.

INTRODUCTION

Celiac disease (CeD) is an immune-mediated disorder triggered by gluten in genetically predisposed individuals, affecting about 1% of the global population[1]. Early diagnosis primarily depends on clinical symptoms, while serological testing and histological examination serve as key diagnostic tools, with the latter being the gold standard for diagnosis. Despite established diagnostic algorithms, many cases remain undiagnosed or misdiagnosed due to silent, asymptomatic, or atypical presentations. In developed countries, there are an estimated average of five undiagnosed cases for every diagnosed case[2]. In developing countries, diagnosis often depends on symptomatic presentation. The diagnostic delay for CeD ranges from 9.7 to 13.3 years[3], highlighting the significant time many patients endure symptoms before receiving a diagnosis.

In many cases, the atypical nature of symptoms affects a significant proportion of patients with undiagnosed CeD. The wide spectrum of CeD symptoms complicates timely diagnosis. Many patients present with atypical or non-gastrointestinal manifestations, such as anemia, growth retardation, dermatitis herpetiformis, infertility, unexplained elevated liver enzymes, or osteomalacia. These diverse presentations often lead patients to seek care from various medical specialties, further delaying diagnosis.

Delayed diagnosis prolongs patient suffering, reduces quality of life, and increases the risk of complications. Delayed diagnoses correlate with poorer mental health, greater anxiety, and worse self-perceived health outcomes[4]. In pediatric and adolescent patients, delays can severely impair growth and development[5]. Additionally, prolonged diagnostic delays result in increased healthcare utilization, higher medical costs, and elevated risks of autoimmune diseases and certain cancers, including intestinal lymphoma[6].

Thus, timely diagnosis and treatment of CeD are essential to improve patients' quality of life, reduce their risk of health complications, and improve their overall health. This study aimed to explore factors associated with diagnostic delays in CeD by analyzing clinical data from patients with previously diagnosed CeD.

MATERIALS AND METHODS

Study design and participants

This retrospective study included patients aged 18 years or older with a primary diagnosis of CeD who were admitted to the Department of Gastroenterology, People’s Hospital of Xinjiang Uygur Autonomous Region from March 2016 to October 2024. Patients without a pathological diagnosis, relevant laboratory data, or complete clinical records were excluded. A diagnostic delay was defined as a period exceeding 2 years between symptom onset and CeD diagnosis.

Diagnostic criteria

CeD was diagnosed based on the World Gastroenterology Organization’s 2013 guidelines[7]. Patients with a positive serum anti-tissue transglutaminase antibody-IgA test and a small intestinal biopsy showing Marsh ≥ II pathological changes were confirmed to have CeD. For individuals with total IgA deficiency, deaminated gliadin peptide antigen-IgG was used for diagnosis. Patients with a lack of pathological diagnosis and laboratory examination indexes and incomplete clinical medical records were excluded from the analysis.

Patient-attributed delays were determined by analyzing the time from the onset of symptoms reported by patients to the first consultation with a healthcare provider. In contrast, physician-related delays were calculated from the initial consultation to the definitive diagnosis or treatment initiation. We utilized patient interviews and medical records to gather these data points.

Clinical data collection

Demographic and clinical data, including age, gender, ethnicity, height, weight, body mass index (BMI), disease duration, clinical symptoms, complications, gastroscopy findings, pathology results, complete blood count, anemia-related markers, and bone mineral density, were retrieved from medical records. Additional details were supplemented through follow-up assessments.

Duodenal biopsy and pathology

Biopsies were performed with at least one specimen from the duodenal bulb (at 9 or 12 o’clock positions) and four specimens from the descending duodenum. Specimens were fixed in 3.7% neutral formaldehyde solution and examined for pathological changes based on the Marsh-Oberhuber classification system[8].

Relevant definitions and criteria

Clinical classification of CeD: Based on the Oslo definition, there are three common types: (1) Classical CeD, which is characterized by symptoms of malabsorption, such as diarrhea, weight loss, or growth delay; (2) Non-classical CeD, characterized by symptoms other than classical features, excluding significant malabsorption; and (3) Subclinical CeD, characterized by symptomatic cases below the clinical detection threshold.

Anemia: Defined by the World Health Organization as hemoglobin (Hb) < 130 g/L for males and < 120 g/L for females at sea level.

Frequent hospitalizations: Defined as three or more hospital admissions.

Complications: Included thyroid disease, elevated serum aminotransferase levels, ulcerative colitis, nephropathy, connective tissue disease, IgA deficiency, osteomalacia, epilepsy, tumors, and lymphoma.

Statistical analysis

Generalized linear regression models were used to evaluate relationships between clinical and biochemical indicators and diagnostic timing. Multiple logistic regression estimated ORs for these indicators. Restricted cubic splines (RCS) were employed to assess nonlinear relationships, with a two-stage analysis conducted to identify turning points in RCS curves. Subgroup analyses examined stratified relationships, and mediation analysis was performed to evaluate the role of specific factors in the association between diagnostic delay and clinical outcomes. All statistical analyses were conducted using R version 4.2.2, with two-sided P values < 0.05 considered statistically significant.

RESULTS

Baseline characteristics of patients with CeD

Clinical records of 166 patients with newly diagnosed CeD were analyzed. Of these, 116 patients (69.9%) were female, and 64 (38.6%) were of Kazakh ethnicity. Ninety-three patients (56.02%) received a diagnosis within 2 years of symptom onset, while 43.98% of the patients had delayed diagnosis, among which 31 patients (18.7%) were diagnosed 5 years later. The average time from symptom onset to the first clinical visit was 12.32 months, whereas the mean duration from the first visit to diagnosis was 20.57 months, reflecting a significantly longer delay attributable to physicians compared to patients. Notably, delayed diagnosis was associated with significantly longer time from first visit to diagnosis. Among the patients, 144 (86.7%) had at least one complication, and 138 (83.1%) presented with classic CeD. The frequency of hospitalizations was higher among those with delayed diagnoses. Most patients (63, or 38.0%) experienced disease onset between the ages of 45 and 55, whereas 81 patients (48.8%) received their diagnosis between 45 and 60 years of age. There were no statistically significant differences across groups in terms of the anti-tissue transglutaminase levels, BMI, education level, occupation, or pathological Marsh classification (Table 1).

Table 1 Baseline characteristics of patients with celiac disease, n (%).

Measures	Diagnosed within 2 years (n = 93)	Delayed diagnosis group
		Diagnosed within 2-5 years (n = 42)	Diagnosed over 5 years (n = 31)	P value
Gender				0.485
Female	66 (71.0)	31 (73.8)	19 (61.3)
Male	27 (29.0)	11 (26.2)	12 (38.7)
Ethnicity				0.241
Han	17 (18.3)	12 (28.6)	5 (16.1)
Uygur	30 (32.3)	17 (40.5)	13 (41.9)
Kazak	41 (44.1)	10 (23.8)	13 (41.9)
Other ethnic	5 (5.4)	3 (7.1)	0 (0.0)
tTG (CU), M (P25, P75)	928.1 (45.4, 2279.2)	243.4 (32.1, 1254.6)	1082.5 (82.7, 1988.5)	0.745
BMI (kg/m2), M (P25, P75)	20.6 (18.7, 24.7)	22.3 (20.2, 25.1)	22.7 (19.1, 25.3)	0.874
First onset of symptoms to visit (months), M (P25, P75)	1 (0.6, 15)	12.3 (2, 36)	12 (1, 21)	< 0.001
First visit to diagnosis (months), M (P25, P75)	0 (0, 0)	20.6 (0, 24)	96 (60, 114)	< 0.001
Patients with comorbidities	15 (16.1)	5 (11.9)	2 (6.5)	0.371
Patients without comorbidities	78 (83.9)	37 (88.1)	29 (93.5)
Non-classical CeD	15 (16.1)	9 (21.4)	4 (12.9)	0.605
Classical CeD	78 (83.9)	33 (78.6)	27 (87.1)
Number of hospitalizations ≤ 3	85 (91.4)	33 (78.6)	19 (61.3)	< 0.001
Number of hospitalizations > 3	8 (8.6)	9 (21.4)	12 (38.7)
Age at diagnosis				0.046
≤ 45 years old	48 (51.6)	14 (33.3)	9 (29.0)
45-60 years old	37 (39.8)	23 (54.8)	21 (67.7)
> 60 years old	8 (8.6)	5 (11.9)	1 (3.2)
Age of onset				0.308
≤ 45 years old	33 (35.5)	11 (26.2)	9 (29.0)
45-55	37 (39.8)	13 (31.0)	13 (41.9)
> 55 years old	23 (24.7)	18 (42.9)	9 (29.0)
Education	31 (33.3)	13 (31.0)	10 (32.3)	0.826
Below junior high school
High school	37 (39.8)	21 (50.0)	13 (41.9)
Bachelor's degree or above	25 (26.9)	8 (19.0)	8 (25.8)
Job	44 (47.3)	18 (42.9)	14 (45.2)	0.888
Freelance
Contract	49 (52.7)	24 (57.1)	17 (54.8)
Pathology	51 (54.8)	25 (59.5)	18 (58.1)	0.865
Marsh II
Marsh III	42 (45.2)	17 (40.5)	13 (41.9)

tTG: The anti-tissue transglutaminase; BMI: Body mass index; CeD: Celiac disease.

Factors associated with delayed diagnosis

Univariate regression analysis showed that diagnostic age (OR = 1.02, 95%CI: 1.00-1.05, P = 0.063), white blood cell count (OR = 0.83, 95%CI: 0.70-0.99, P = 0.042), blood urea nitrogen (BUN) level (OR = 1.37, 95%CI: 1.10-1.71, P = 0.005), and serum iron (OR = 1.10, 95%CI: 1.02-1.19, P = 0.016) were associated with delayed diagnosis. Multivariate regression analysis revealed that after adjusting for confounders, only BUN level remained an independent risk factor (OR = 1.29, 95%CI: 1.01-1.65, P = 0.038; Table 2).

Table 2 Univariate and multivariate regression analyses of diagnostic delays in patients with celiac disease, mean ± SD.

Measures	Diagnosed within 2 years	Delayed diagnosis group		Univariate regression analysis, OR (95%CI), P value	Multivariate regression analysis, OR (95%CI), P value
		Diagnosed within 2-5 years	Diagnosed over 5 years
Age at diagnosis	46.8 ± 13.1	51.3 ± 14.2	49.6 ± 10.3	1.02 (1.00-1.05), P = 0.063	1.01 (0.98-1.04), P = 0.554
Leukocytes	6.4 ± 1.9	5.6 ± 1.6	6.1 ± 1.8	0.83 (0.70-0.99), P = 0.042	0.83 (0.68-1.00), P = 0.051
Hemoglobin	119.6 ± 23.6	121.8 ± 23.1	118.2 ± 28.4	1.00 (0.99-1.01), P = 0.856
Total protein	68.7 ± 10.5	68.1 ± 9.9	65.8 ± 9.5	0.98 (0.95-1.01), P = 0.307
Serum calcium	2.1 ± 0.2	2.2 ± 0.2	2.2 ± 0.2	3.85 (0.76-19.45), P = 0.103	1.44 (0.23-8.80), P = 0.696
Thyroid stimulating hormone	4.0 ± 5.0	3.1 ± 2.4	3.5 ± 4.6	0.96 (0.89-1.04), P = 0.295
Blood urea nitrogen	4.0 ± 1.4	4.7 ± 1.7	4.9 ± 2.1	1.37 (1.10-1.71), P = 0.005	1.29 (1.01-1.65), P = 0.038
Serum iron	8.2 ± 3.1	10.5 ± 5.5	10.4 ± 9.3	1.10 (1.02-1.19), P = 0.016	1.07 (1.00-1.15), P = 0.063

Correlation between urea nitrogen levels and diagnostic delay

A significant dose-response relationship was observed between BUN levels and diagnostic delay. Without covariate adjustment, each 1 mmol/L increase in BUN was associated with a 37% increased risk of delayed diagnosis (OR = 1.37, 95%CI: 1.12-1.73, P = 0.005). After stepwise adjustment for age, sex, BMI, comorbidities (vitamin D deficiency, anemia, thyroid disease), and hematologic parameters (Table 3), Model 4 showed a persistent significant independent effect of BUN (OR = 1.42, 95%CI: 1.09-1.91, P = 0.015), suggesting that BUN may influence diagnostic delay by reflecting systemic metabolic status or underlying comorbidities (Figure 1).

Figure 1 Trend analysis of blood urea nitrogen levels and delayed diagnosis of celiac disease. BUN: Blood urea nitrogen.

Table 3 Association between blood urea nitrogen levels and delayed diagnosis of celiac disease.

Model	OR	95%CI	P value
1	1.37	1.12-1.73	0.005
2	1.34	1.07-1.72	0.015
3	1.37	1.09-1.78	0.013
4	1.42	1.09-1.91	0.015

Model 1: Unadjusted covariates. Model 2: Adjusted for age, sex, and body mass index. Model 3: Adjusted for age, sex, body mass index, presence or absence of vitamin D deficiency, presence or absence of anemia, and presence or absence of thyroid disease. Model 4: Age, sex, body mass index, presence or absence of vitamin D deficiency, presence or absence of anemia, and presence or absence of thyroid disease. Model 4: Adjusted for age, sex, body mass index, presence or absence of vitamin D deficiency, presence or absence of anemia, presence or absence of thyroid disease, white blood cells, creatinine, albumin, and serum iron.

Population characteristic analysis based on BUN tertiles demonstrated that the high BUN group (Q3) had a significantly older diagnostic age (P < 0.001), a longer time from first visit to diagnosis (P = 0.002), and higher proportions of male sex (44.4% vs Q1 15.8%, P = 0.004) and Uyghur ethnicity (46.3% vs Q1 29.8%, P = 0.001; Table 4). These findings suggest potential interactions between BUN and factors such as age and ethnicity.

Table 4 Characteristics of patients with celiac disease based on blood urea nitrogen tertiles, mean ± SD/M (P₂₅, P₇₅)/n (%).

Indicators	Q1	Q2	Q3	P value
Gender				0.004
Female	48 (84.2)	38 (69.1)	30 (55.6)
Male	9 (15.8)	17 (30.9)	24 (44.4)
Ethnicity				0.001
Han	5 (8.8)	13 (23.6)	16 (29.6)
Uygur	17 (29.8)	18 (32.7)	25 (46.3)
Kazak	30 (52.6)	24 (43.6)	10 (18.5)
Other ethnic	5 (8.8)	0 (0.0)	3 (5.6)
tTG (CU)	769.3 (29, 1896.5)	967.4 (71.7, 2673.4)	263.2 (38.8, 1660)	0.639
BMI (kg/m2)	20.1 (18.4, 24.2)	21.9 (19, 24.1)	22.5 (20, 25.9)	0.705
First onset of symptoms to visit (months)	1 (0.5, 12)	3 (1, 12)	6 (1, 12.3)	0.514
First visit to diagnosis (months)	0 (0, 1)	0 (0, 18)	14 (0, 48)	0.002
Patients with comorbidities (%)	11 (19.3)	7 (12.7)	4 (7.4)	0.18
Patients without comorbidities	46 (80.7)	48 (87.3)	50 (92.6)
Non-classical CeD	5 (8.8)	10 (18.2)	13 (24.1)	0.094
Classical CeD	52 (91.2)	45 (81.8)	41 (75.9)
Number of hospitalizations ≤ 3	48 (84.2)	49 (89.1)	40 (74.1)	0.109
Number of hospitalizations > 3	9 (15.8)	6 (10.9)	14 (25.9)
Age at diagnosis	43.89 ± 11.78	47.67 ± 12.79	54.02 ± 12.67	< 0.001
Age of onset	42.85 ± 12.74	46.95 ± 12.87	52.70 ± 13.52	0.001
Education				0.636
Below junior high school	19 (33.3)	15 (27.3)	20 (37.0)
High school	22 (38.6)	28 (50.9)	21 (38.9)
Bachelor's degree or above	16 (28.1)	12 (21.8)	13 (24.1)
Job				0.031
Freelance	19 (33.3)	32 (58.2)	25 (46.3)
Contract	38 (66.7)	23 (41.8)	29 (53.7)
Pathology				0.438
Marsh II	29 (50.9)	31 (56.4)	34 (63.0)
Marsh III	28 (49.1)	24 (43.6)	20 (37.0)

tTG: The anti-tissue transglutaminase; BMI: Body mass index; CeD: Celiac disease.

Nonlinear relationship and two-stage analysis of urea nitrogen levels

RCS analysis revealed a nonlinear dose-response relationship between BUN levels and diagnostic delay (Figure 2). Two-stage piecewise linear regression showed that there was a significant inflection point when the BUN threshold was 4.3 mmol/L (P = 0.04). When BUN was ≤ 4.3 mmol/L, the risk of delayed diagnosis did not increase significantly (OR = 0.67, 95%CI: 0.23-1.93, P = 0.44); When BUN > 4.3 mmol/L, the risk significantly increased (OR = 1.39, 95%CI: 1.04-1.99, P = 0.04; Table 5). In the two-stage analysis, patients with BUN ≤ 4.3 mmol/L exhibited a significantly reduced risk of diagnostic delay (OR = 0.67, 95%CI: 0.23–1.93, P = 0.44) compared to those with BUN > 4.3 mmol/L. However, this effect was not statistically significant (P = 0.44), whereas BUN > 4.3 mmol/L independently increased the risk (OR = 1.39, P = 0.04). This result may reflect that mild abnormalities in renal function can serve as a marker of the severity of underlying diseases, leading to delayed clinical diagnosis.

Figure 2 Dose-response association between blood urea nitrogen levels with delayed diagnosis of celiac disease. BUN: Blood urea nitrogen.

Table 5 Two-stage analysis of the association between blood urea nitrogen levels and delayed diagnosis of celiac disease.

BUN	OR	95%CI	P value
BUN or less tendency for 4.3 mmol/L	0.67	0.23-1.93	0.44
BUN > tendency for 4.3 mmol/L	1.39	1.04-1.99	0.04

BUN: Blood urea nitrogen.

Subgroup analysis

Subgroup analyses (Table 6) showed heterogeneous effects of BUN across patient characteristics. The association was stronger in females (OR = 1.49, P = 0.008) and non-classical CeD patients (OR = 3.71, P = 0.009). Among ethnic groups, Kazak patients exhibited a significant association (OR = 1.64, P = 0.040). Patients without anemia (OR = 1.76, P = 0.001) or vitamin D deficiency (OR = 1.57, P = 0.002) had stronger BUN-delay relationships. Hospitalization history (≥ 3 times, OR = 1.31, P = 0.032) and lower education level (below junior high, OR = 1.74, P = 0.012) further modified the association, suggesting complex interactions between renal function, socioeconomic factors, and diagnostic processes in CeD.

Table 6 Effect of each stratification factor on the relationship between blood urea nitrogen and delayed diagnosis.

Variables	OR	95%CI	P value
Gender
Female	1.49	1.11-1.99	0.008
Male	1.26	0.88-1.8	0.206
Age (year)
< 45	1.26	0.89-1.69	0.197
45-65	1.67	0.91-3.06	0.099
> 65	1.24	0.88-1.75	0.215
Typing
Non-classical celiac disease	3.71	1.38-9.95	0.009
Classic celiac disease	1.26	1.01-1.57	0.042
Ethnic groups
Han nationality	1.45	0.8-2.65	0.224
Uygur	1.17	0.9-1.53	0.245
Kazak	1.64	1.02-2.64	0.040
Other ethnic	3.24	0.6-17.35	0.171
Combined vitamin D deficiency
No vitamin D deficiency	1.57	1.19-2.09	0.002
Have a vitamin D deficiency	0.86	0.53-1.38	0.528
Presence of anemia
No anemia	1.76	1.25-2.48	0.001
Having anemia	1.07	0.82-1.4	0.613
Number of hospitalizations
≤ 3	1.65	0.9-3.05	0.107
> 3	1.31	1.02-1.67	0.032
Level of education
Below junior high school	1.74	1.13-2.68	0.012
High school	1.2	0.83-1.75	0.339
Bachelor's degree or above	1.26	0.88-1.82	0.208
Job
Freelance	1.49	1.03-2.15	0.032
Contract	1.32	1-1.75	0.052
Pathology
Marsh II	1.29	0.98-1.7	0.073
Marsh III	1.5	1.05-2.16	0.027

Mediation analysis

Analysis of the mediating role of folic acid deficiency and anemia in the association between BUN levels and the delayed diagnosis of CeD showed that the mediating effect of folic acid deficiency is significant. The total effect of BUN level on the delayed diagnosis of CeD is OR = 1.025 (95%CI: 1.008-1.03, P < 0.001), among which the mediating effect of folic acid deficiency accounted for 11.90% of the total effect (P = 0.028), and the effect of folic acid level itself on the delayed diagnosis of CeD was OR = 1.003 (95%CI: 1-1.008, P = 0.028). The mediating effect of anemia was also statistically significant. The total effect of BUN level on the delayed diagnosis of CeD was OR = 1.023 (95%CI: 1.01-1.03, P = 0.01), the mediating effect of anemia accounted for 13.00% of the total effect (P = 0.044), and the effect of anemia on the delayed diagnosis of CeD was OR = 1.003 (95%CI: 1-1.01, P = 0.044; Figure 3).

Figure 3 Mediation analysis of the relationship between blood urea nitrogen levels and delayed diagnosis of celiac disease. A: Mediation analysis of blood urea nitrogen (BUN) and folic acid; B: Mediation analysis of BUN and anemia. BUN: Blood urea nitrogen; CeD: Celiac disease.

DISCUSSION

Despite the increasing prevalence of CeD, diagnostic delays remain a significant issue, adversely affecting patient quality of life and contributing to complications. This study provides the first detailed analysis of diagnostic delays and contributing factors among patients with CeD in China. Our findings show that 43.98% of patients experienced diagnostic delays, with 18.67% diagnosed more than 5 years after symptom onset. The average patient-attributed delay was 12.32 months, while physician-related delays averaged 20.57 months. These delays highlight gaps in both patient awareness and healthcare delivery. Contributing factors include non-specific symptoms, limited physician expertise, cultural and socioeconomic barriers, and healthcare system inefficiencies.

CeD symptoms vary widely, complicating diagnosis. While some patients present with classic gastrointestinal (GI) symptoms such as abdominal pain, diarrhea, and weight loss, others exhibit non-digestive manifestations like fatigue, anemia, skin rashes, or bone pain. Our study revealed that asymptomatic cases were typically diagnosed within 2 to 5 years, yet 16.3% of patients with GI symptoms experienced delays exceeding 5 years. This variation often leads clinicians to explore alternative diagnoses, such as infections or irritable bowel syndrome, before considering CeD. Early symptoms may be mild and non-specific, evolving over time, which further obscures diagnosis and delays appropriate interventions. Our finding that non-classical CeD (OR = 3.71) has a stronger association with diagnostic delay than classical CeD (OR = 1.26) accords with atypical presentations hindering clinical recognition. Classical CeD overt GI symptoms prompt earlier referrals, while non-classical CeD extraintestinal/subtle symptoms are often misattributed, delaying diagnosis. This underscores the need to raise awareness of non-classical symptoms for early detection.

Comorbidities present another diagnostic challenge. Conditions like thyroid disease and type 1 diabetes often coexist with CeD, masking symptoms or producing overlapping signs such as weight changes, fatigue, or GI discomfort. These diseases may cause an abnormal immune response in the patient, which can mask the symptoms of CeD[9]. Singh et al[10] recently highlighted this interconnection, advocating for targeted CeD screening in patients with autoimmune disease. Our findings indicate higher rates of diagnostic delay among patients with comorbidities, likely due to the increased complexity of their clinical presentations. Mental health conditions, including depression and anxiety, further complicate diagnosis by affecting patients' symptom reporting and physicians' clinical judgment[11]. The stronger link between BUN elevation and diagnostic delay in patients without vitamin D deficiency or anemia fits “silent” CeD. Common comorbidities like vitamin D deficiency and anemia often alert clinicians to CeD, but without these markers, mild symptoms may be dismissed as functional. This “masking effect” means those with BUN elevation but normal Hb and vitamin D might get renal/cardiac workups instead of GI assessment, highlighting the need for CeD screening in unexplained metabolic issues.

Age and gender also play roles in diagnostic delays. Older patients, particularly those aged 45-60, often present with atypical or vague symptoms, making diagnosis more challenging. Our study also showed that the incidence of delayed diagnosis was higher in patients aged 45-60 years, and patients over 60 years of age showed fewer delays, possibly due to increased health awareness and proactive healthcare-seeking behavior. However, we observed that age lost its significance in the multivariable model, a finding that may be attributed to the following mechanisms: First, BUN retained its unique independent predictive value after adjustment, which diminished the contribution of weaker correlates—including age and white blood cell count—as key drivers of diagnostic delay. Age, in this context, may represent a secondary phenomenon or downstream effect of underlying metabolic dysregulation rather than a primary driver of delayed diagnosis. Second, the nominal significance of age in univariate analysis (OR = 1.02, P = 0.063) reflects limited discriminative power, as the marginal association may not translate to clinical relevance. This, combined with reduced statistical power due to the moderate sample size and potential collinearity between age and BUN, age-related changes in renal function could influence both age itself and BUN levels, ultimately led to its loss of significance in the multivariable framework. Gender differences were also evident, with female patients experiencing longer diagnostic delays. Women often report non-GI symptoms, such as anemia or bone mineral density issues, which may divert attention from CeD. Subgroup analysis revealed that elevated BUN levels were strongly associated with delayed diagnosis in women, consistent with findings from studies conducted in other countries[12].

Psychological factors further contribute to diagnostic delays. Persistent symptoms can lead to depression or anxiety, discouraging patients from seeking timely medical care. Mild or episodic symptoms may prompt patients to pursue symptomatic treatment rather than definitive diagnoses, delaying further investigations[13]. Conversely, frequent healthcare visits may create mistrust in physicians' recommendations, leading to delays in essential diagnostic tests. Our findings showed that patients with frequent hospitalizations and elevated BUN levels faced higher risks of delayed diagnoses, often due to misdirected attention toward nephrology-related conditions.

Socioeconomic status and healthcare-seeking behavior are critical determinants of diagnostic delays. Patients with lower socioeconomic status are more likely to present with atypical symptoms, such as fatigue or anemia, rather than classic CeD symptoms[14]. Financial constraints, lack of health insurance, limited education, and restricted access to healthcare information exacerbate these delays. Low-income patients may be reluctant to seek medical care because of concerns about healthcare costs or even delay in seeking care after symptoms develop, which significantly increases diagnostic delay. Our subgroup analysis indicated that patients with lower education levels or unemployment experienced greater diagnostic delays, particularly when BUN levels were elevated.

From a healthcare system perspective, CeD remains underrecognized in certain regions, leading to misinterpretation of symptoms, especially in atypical cases. Physicians may fail to consider CeD or delay serological testing and biopsies. Effective communication between patients and physicians is also critical. Patients may inadequately describe their symptoms, while physicians may neglect detailed history-taking, resulting in diagnostic oversight[15]. The physician-related delay in this study (20.57 months) was shorter than delays reported in some countries, where diagnosis can take 6 to 14 years[16-19]. In contrast, our data are similar to those reported in a Danish study[20].

A key finding of this study is the association between elevated BUN levels and diagnostic delays. Dose-response and two-stage analyses revealed that higher BUN levels were significantly correlated with longer delays and increased diagnostic risk. Mediation analysis identified anemia and folate deficiency as contributing factors to this association, emphasizing the role of nutritional deficiencies in CeD pathophysiology[21].

Elevated BUN levels are multifactorial in origin, commonly linked to impaired renal function, hypovolemia, or excessive protein intake. Critically, these etiological factors may collectively disrupt folate absorption and metabolism. Folate deficiency not only predisposes individuals to megaloblastic anemia but also manifests as a constellation of non-specific clinical features—including profound fatigue and myasthenia—that overlap significantly with the manifestations of CeD. Reduced enterocyte turnover prolongs the exposure of the intestinal lumen to undigested nutrients, increasing urea production from bacterial fermentation and elevating BUN levels[22]. This pathophysiological convergence creates diagnostic ambiguity, potentially delaying recognition of underlying CeD in comorbid presentations. Patients with CeD may face an increased risk of kidney disease. Elevated BUN levels may exert a diagnostic masking effect on CeD, predisposing clinicians to prioritize investigations for renal or metabolic etiologies during initial evaluations, while potentially overlooking underlying CeD. This diagnostic overshadowing effect, compounded by symptom overlap, frequently delays CeD recognition, particularly in patients with concomitant BUN elevation and hematologic abnormalities. Anemia may elevate BUN indirectly: Reduced erythropoietin production in chronic disease states can impair renal function, slowing urea clearance[23]. Therefore, for patients with elevated BUN and anemia, especially those with related symptoms or high-risk factors, screening for CeD should be considered to ensure early diagnosis and timely intervention. This not only helps improve the nutritional status of patients, but also prevents long-term complications related to CeD. Future studies should further explore the interactions of BUN, folic acid, and anemia in the diagnosis of CeD to optimize the diagnostic process and improve diagnostic efficiency.

This study, while representing the largest single-center CeD cohort in China and leveraging our institution's role as the Xinjiang Celiac Disease Alliance hub (receiving all regional suspected CeD referrals), has some key limitations. First, sample size constraints persist. Though 166 cases is the largest Marsh ≥ II cohort in China, this remains modest compared to international multicenter studies, potentially limiting statistical power to detect weak but clinically meaningful associations—particularly in subgroup analyses with smaller strata. Second, selection bias affects generalizability. Our cohort comprises consecutive referred patients confirmed by Marsh ≥ II pathology, ensuring diagnostic accuracy but excluding two critical groups: (1) “Potential CeD” patients (seropositive but Marsh < II, possibly with longer delays due to lower clinical vigilance); and (2) Mild/asymptomatic CeD cases, which are under-referred. Thus, findings may not fully reflect the full CeD spectrum. Third, focus on Marsh ≥ II-confirmed CeD enhances result robustness by minimizing serological false positives but narrows scope. Excluding “potential CeD” individuals limits inference to advanced/symptomatic disease, potentially missing distinct delay patterns in earlier stages. Fourth, instability in subgroup analyses arises from small sample sizes in certain strata. Small samples are prone to random variation, risking over/underestimation of effect sizes; these findings are exploratory and require validation. Finally, the retrospective design introduces information bias. Data rely on historical records, which may be incomplete or subject to recall bias. Subtle/intermittent symptoms may be under-documented.

CONCLUSION

Diagnostic delays in CeD stem from a complex interplay of non-specific symptoms, patient behavior, healthcare system challenges, and comorbidities. Elevated BUN levels are independent predictors of diagnostic delay in CeD, with heterogeneity observed across gender, disease subtype, ethnicity, and pathological type. Clinicians should prioritize high-risk populations with BUN ≥ 4.3 mmol/L, particularly female patients with non-classical CeD and Kazak individuals, to reduce diagnostic delay.