Nomogram for predicting 6-month BMIZ changes in children with Crohn’s disease: a single-center retrospective study

Abstract

Background

Perianal disease (PD) is an aggressive phenotype in children with Crohn’s disease (CD), but its measurable impact on short-term nutritional recovery remains understudied. This study aimed to evaluate the independent effect of PD on the 6-month Body Mass Index Z-score (BMIZ) trajectory and to develop a practical nomogram for early risk stratification in pediatric CD.

Methods

A single-center retrospective cohort of 51 pediatric CD patients (29 with PD and 22 without PD) with at least 6 months of follow-up was enrolled from a tertiary hospital in Fujian, southern China, between July 2017 and March 2025. BMIZ trajectories over time were estimated using linear mixed models (LMM). Predictors of 6-month ΔBMIZ were screened using univariate regression (P < 0.20), followed by LASSO regression (tenfold cross-validation) to construct a nomogram. Model performance was internally validated using 1000 bootstrap resamples (R², MAE, RMSE) and calibration plots. Decision curve analysis (DCA) was performed to evaluate clinical utility for predicting inadequate nutritional recovery (ΔBMIZ ≤ 0.40).

Results

LMM analysis showed that PD status significantly reduced the slope of BMIZ recovery after adjusting for age, sex, and biologic therapy (β = -0.085, 95% CI -0.147 to -0.023, P = 0.010). LASSO regression identified six core predictors: perianal disease, positive fecal occult blood test (FOBT), biologic therapy, erythrocyte sedimentation rate (ESR), hemoglobin (Hb), and 25-hydroxyvitamin D3 (25(OH)D3). The nomogram demonstrated satisfactory predictive performance (R² = 0.51, MAE = 0.40, RMSE = 0.49), with calibration plots showing good agreement between predicted and observed ΔBMIZ. DCA confirmed a positive net benefit for predicting inadequate nutritional recovery (ΔBMIZ ≤ 0.40) across a threshold probability range of 5% to 85%.

Conclusion

Perianal disease was independently associated with suboptimal short-term nutritional recovery in children with CD. The proposed nomogram, integrating phenotypic, inflammatory, and nutritional markers, may help identify high-risk patients at diagnosis or early after diagnosis and support individualized nutritional monitoring and therapeutic planning in southern Chinese pediatric CD patients.

Introduction

Crohn’s disease (CD) in children is a chronic immune-mediated inflammatory disorder whose incidence has risen substantially worldwide, including in East Asia [1, 2]. Unlike adult-onset CD, pediatric CD occurs during a critical period of growth and development, making children particularly vulnerable to nutritional impairment, delayed puberty, and growth failure [3, 4]. Previous studies report that 15–40% of children with CD present with growth impairment at diagnosis [5]. Delayed nutritional restoration and persistent growth retardation may lead to long-term inflammatory burden, irreversible growth deficits, reduced bone mineral density, and metabolic disorders that impact lifelong health and quality of life [4–6]. Consequently, the STRIDE II consensus explicitly designates “restoration of normal growth” as a top therapeutic goal in pediatric inflammatory bowel disease (IBD) [7].

Perianal disease (PD) is considered an aggressive phenotype of pediatric CD [8]. It has been associated with extensive bowel involvement, greater inflammatory burden, a more complicated disease course, and earlier escalation to biologic therapy. Children with PD also remain at increased risk for fistula-related complications, abscesses, and surgery [3, 8, 9]. However, although PD is widely regarded as a marker of severe disease, its specific impact on short-term nutritional recovery during the early post-diagnostic period has not been well characterized, particularly in East Asian pediatric populations.

Body mass index Z-score (BMIZ) is a practical and widely used indicator of nutritional status in children with CD. Improvement in BMIZ after diagnosis may reflect early recovery from inflammation-driven catabolism and inadequate intake [10, 11]. However, it remains unclear whether short-term BMIZ recovery differs according to disease phenotype, especially between children with and without PD receiving contemporary treatment. In addition, no clinically practical tool is currently available to integrate phenotype, inflammatory burden, and nutritional markers to predict early nutritional recovery at the individual level.

Therefore, in this retrospective longitudinal cohort study, we investigated the independent effect of PD on BMIZ recovery trajectories during the first 6 months after diagnosis in southern Chinese children with CD. We also developed and internally validated a nomogram to predict 6-month ΔBMIZ using routinely available clinical and laboratory variables, with the aim of supporting early risk stratification and individualized nutritional management.

Methods

Ethics approval and consent

This retrospective study was approved by the Ethics Committee of The First Affiliated Hospital of Fujian Medical University (Approval No. MRCTA, ECFAH FMU [2019] 219). Written informed consent was obtained from the parents or caregivers. This study was conducted in accordance with the ethical principles of the 1964 Declaration of Helsinki and its later amendments. Clinical trial number: not applicable.

Study design and patients

This single-center retrospective cohort study included 51 pediatric patients newly diagnosed with Crohn’s disease admitted to the Pediatric Department of the First Affiliated Hospital of Fujian Medical University between July 2017 and March 2025. The diagnoses of all participants met the consensus criteria for inflammatory bowel disease (IBD) in children established by the European Society for Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) [12].

Based on comprehensive perianal evaluation, including digital rectal examination and pelvic magnetic resonance imaging, patients were classified into a PD group and a non-PD group. PD was defined by the presence of perianal fistulas, abscesses, or other perianal lesions consistent with Crohn’s disease.

Data collection

Baseline variables were defined as data collected at or near the time of diagnosis. Data recorded included age at diagnosis, sex, mode of delivery, feeding pattern, family history of inflammatory bowel disease (IBD), time from symptom onset to diagnosis, disease phenotype according to the Paris classification, extraintestinal manifestations, baseline disease activity indices, baseline anthropometric parameters, and baseline laboratory measurements. A history of abdominal or perianal surgery referred to surgery performed before or at the time of enrollment work-up. In contrast, treatment variables such as exclusive enteral nutrition (EEN) and biologic therapy were analyzed as treatment exposure during the first 6 months after diagnosis rather than as baseline characteristics.

Assessment of disease activity and mucosal healing

Clinical disease activity was assessed using the Pediatric Crohn’s disease activity index (PCDAI). Clinical remission was defined as a score < 10 points, and a score above 10 points was classified as active disease [13]. Endoscopic disease severity was assessed using the simple endoscopic score of Crohn’s disease (SES-CD), with mucosal healing (MH) defined when the SES-CD score was < 3 points [14].

Definition of physical growth and nutritional indicators

Baseline and 6-month BMIZ and height-for-age Z-score (HAZ) were measured by trained pediatric nurses using standardized equipment. Both indices were calculated according to the World Health Organization reference standards for children and adolescents aged 5–19 years [15]. The primary nutritional outcome, ΔBMIZ, was defined as BMIZ at 6-month follow-up minus BMIZ at baseline and was used to reflect the magnitude of short-term nutritional recovery after diagnosis.

Definition of biologic therapy

Biologic therapy consisted of infliximab (IFX) administration according to a standardized induction and maintenance regimen. The induction phase consisted of intravenous infusions of 5 mg/kg at weeks 0, 2, and 6, and the maintenance phase consisted of infusions of 5–10 mg/kg every 8 weeks. Participants had to have been on treatment for at least six consecutive months to be included in the study.

Laboratory measurements

Blood and fecal samples were collected from all participants at baseline and at follow-up examination after 6 months. The following parameters were systematically analyzed:

• Inflammatory markers: white blood cell count (WBC), C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) and platelet count (PLT).

• Nutritional and metabolic parameters: Hemoglobin (Hb), hematocrit (HCT), serum albumin, 25(OH)D3, folic acid, and vitamin B12.

• Fecal biomarkers: fecal calprotectin (FC) and fecal occult blood test (FOBT) results (recorded as binary data: positive or negative).

Statistical analysis

All statistical analyses were conducted using R software (version 4.4.1; R Foundation for Statistical Computing, Vienna, Austria) with the glmnet, rms, and ggplot2 packages. A two-sided P < 0.05 was considered statistically significant. Normality was assessed using the Shapiro–Wilk test. Continuous variables are presented as mean ± standard deviation or median (interquartile range), and categorical variables are presented as n (%). Between-group differences were assessed using the Student’s t-test, Mann–Whitney U-test, Pearson’s χ² test, or Fisher’s exact test, as appropriate.

BMIZ trajectories at baseline and 6-month follow-up were estimated using a LMM with a random intercept, adjusted for age, sex, and biologic therapy during follow-up. Predictors of 6-month ΔBMIZ were identified using a two-step procedure: univariate linear regression (P < 0.20) followed by LASSO regression with tenfold cross-validation and selection of λ1se. Given the limited sample size, univariate screening was used to reduce the number of candidate variables before penalized regression. A multivariable linear regression model was then used to construct the nomogram. Model performance was evaluated using R², MAE, and RMSE. Internal validation was conducted using 1000 bootstrap resamples with calibration plots. Decision curve analysis was conducted to assess the net clinical benefit of the nomogram for predicting inadequate nutritional recovery, defined in the present study as ΔBMIZ ≤ 0.40. This threshold was selected pragmatically based on routine clinical evaluation at our hospital. Previous studies have suggested that improvement or normalization of BMIZ is a clinically meaningful indicator of nutritional recovery in pediatric Crohn’s disease [10].

Results

Figure 1 shows the participant recruitment process.

Fig. 1.

The flowchart of participants’ inclusion

Baseline characteristics of the study population

This study included 51 children with CD followed for ≥ 6 months, including 29 (56.9%) with PD and 22 (43.1%) without PD. The two groups did not differ significantly in demographic characteristics, including age at diagnosis, sex, mode of delivery, feeding pattern, or family history of inflammatory bowel disease (Table 1).

Table 1.

Baseline clinical characteristics of pediatric Crohn’s disease patients with and without perianal disease at diagnosis and treatment exposure during the first 6 months

Variables	Non-perianal disease (n = 22)	Perianal disease (n = 29)	P value
Demographic characteristics
Age at diagnosis, median (IQR), years	11.0 (10.0,13.0)	12.0 (10.0,13.0)	0.658
Male sex, n (%)	11 (50.0)	18 (62.1)	0.389
Cesarean delivery, n (%)	8 (36.4)	9 (31.0)	0.689
Breastfeeding history, n (%)	19 (86.4)	24 (82.8)	0.726
Time from symptom onset to diagnosis, median (IQR), days	231 (89,807)	306 (70,767)	0.805
Positive family history, n (%)	1 (4.5)	5 (17.2)	0.163
Disease phenotype and history
Previous abdominal or perianal surgery, n (%)	3 (13.6)	15 (51.7)	0.005
History of antibiotic exposure, n (%)	15 (68.2)	25 (86.2)	0.121
Paris classification (disease location)
L1 (terminal ileum), n (%)	6 (27.3)	2 (6.9)	0.048
L2 (colon), n (%)	2 (9.1)	0 (0.0)	0.098
L3 (ileocolon), n (%)	14 (63.6)	26 (89.7)	0.025
L4a (upper GI, proximal to ligament of Treitz), n (%)	3 (13.6)	3 (10.3)	0.718
L4b (upper GI, distal to ligament of Treitz), n (%)	9 (40.9)	14 (48.3)	0.601
Treatment strategies
Exclusive enteral nutrition (EEN) alone, n (%)	9 (40.9)	4 (13.8)	0.028
EEN + infliximab combination therapy, n (%)	13 (59.1)	25 (86.2)	0.028
Disease activity and anthropometric indices
PCDAI score, median (IQR)	21.3 (12.4,42.3)	30.0 (14.0,41.3)	0.299
SES-CD score, median (IQR)	25.5 (5.0,34.3)	18.0 (8.5,26.5)	0.419
HAZ (height-for-age Z-score), mean ± SD	− 0.59 ± 1.00	− 0.70 ± 1.00	0.722
BMIZ (BMI Z-score), mean ± SD	− 1.23 ± 1.00	− 0.98 ± 1.00	0.383
Laboratory parameters
White blood cell count (10⁹/L), mean ± SD	8.04 ± 3.15	7.97 ± 2.33	0.922
Hemoglobin (g/L), median (IQR)	115 (94,128)	114 (95,122)	0.524
Platelet count (10⁹/L), mean ± SD	464 ± 157	414 ± 132	0.220
C-reactive protein (mg/L), median (IQR)	21.2 (6.9,70.2)	18.2 (6.1,56.3)	0.768
Erythrocyte sedimentation rate (mm/h), median (IQR)	47.5 (22.5,93.8)	65.0 (33.5,84.5)	0.621
Albumin (g/L), mean ± SD	37.7 ± 5.9	37.1 ± 4.5	0.668
Positive fecal occult blood test (FOBT), n (%)	11 (50.0)	28 (96.6)	< 0.001
Positive fecal calprotectin (FC), n (%)	20 (90.9)	17 (58.6)	0.010
25-hydroxyvitamin D (ng/mL), median (IQR)	19.6 (12.3,24.3)	14.7 (12.3,23.7)	0.902
Folate (nmol/L), median (IQR)	12.1 (8.3,18.0)	16.9 (13.9,35.4)	0.005

Data are presented as n (%), mean ± standard deviation (SD), or median (interquartile range, IQR), as appropriate. Continuous variables were compared using Student’s t test or the Mann–Whitney U test, while categorical variables were compared using the Pearson χ² test or Fisher’s exact test, as appropriate

Abbreviations: CD Crohn’s disease, PD perianal disease, PCDAI Pediatric Crohn’s Disease Activity Index, SES-CD Simple Endoscopic Score for Crohn’s Disease, HAZ height-for-age z-score, BMIZ body mass index-for-age z-score, WBC white blood cell count, CRP C-reactive protein, ESR erythrocyte sedimentation rate

At diagnosis, the PD group showed a more complex disease phenotype than the non-PD group. Specifically, ileocolonic involvement (L3) was more common in the PD group (89.7% vs. 63.6%, P = 0.025) and a history of abdominal or perianal surgery before or at enrollment work-up was also more frequent (51.7% vs. 13.6%, P = 0.005). During the first 6 months after diagnosis, combined treatment with EEN and infliximab was more common in the PD group than in the non-PD group (86.2% vs. 59.1%, P = 0.028), indicating more intensive early treatment exposure in children with PD.

Regarding baseline growth and laboratory variables, BMIZ, HAZ, and most inflammatory markers were broadly comparable between groups. However, the proportion of patients with positive FOBT was significantly higher in the PD group (96.6% vs. 50.0%, P < 0.001), whereas positive fecal calprotectin was more frequent in the non-PD group (90.9% vs. 58.6%, P = 0.010).

Longitudinal changes in BMIZ in the two groups over 6 months

LMM was used to analyze the longitudinal changes in BMIZ over 6 months after diagnosis, adjusting for baseline age, sex, and use of biologics during follow-up. A significant interaction between PD status and time was observed (β = −0.085, 95% CI − 0.147 to − 0.023, P = 0.010), indicating that PD was associated with a slower rate of BMIZ recovery (Fig. 2).

Fig. 2.

Association between baseline perianal disease status and 6-month change in BMI z-score (ΔBMIZ). Scatter points represent individual patients, and the solid lines indicate fitted values from linear regression models. Patients without perianal disease exhibited a significantly greater improvement in BMIZ over 6 months compared with those with perianal disease. Regression coefficients and corresponding P values were derived from linear models adjusted for baseline age, sex, and biologic therapy

The non-PD group showed a greater increase in BMIZ during follow-up than the PD group (1.09 ± 0.70 vs. 0.58 ± 0.64 over 6 months). Except for BMIZ, no significant between-group differences were observed in the longitudinal changes of HAZ, PCDAI, SES-CD, or other nutrition-related indicators (Table 2).

Table 2.

Comparison of longitudinal changes in clinical and nutritional indicators over 6 months between pediatric Crohn’s disease patients with and without perianal disease

Outcome	Group × time interaction, β (SE)	t value	95% CI	P value	No. of observations/participants
HAZ	0.011(0.007)	1.525	− 0.003 to 0.026	0.134	102/51
BMIZ	−0.085(0.031)	−2.685	−0.147 to −0.023	0.010	102/51
PCDAI	−0.301(0.701)	−0.430	−1.692 to 1.089	0.669	102/51
SES-CD	0.241(0.481)	0.502	−0.713 to 1.195	0.618	102/51
Albumin	−0.351(0.355)	−0.990	−1.055 to 0.353	0.327	102/51
25(OH)D3	0.482(0.326)	1.482	−0.164 to 1.129	0.145	102/51
Folate	−0.669(0.656)	−1.019	−1.972 to 0.645	0.313	102/51
VitaminB12	21.131(11.991)	1.762	−2.675 to 44.937	0.084	102/51

Data are presented as estimated coefficients for the group × time interaction term derived from linear mixed-effects models, reflecting differences in the rate of change over time between groups. All models included a random intercept to account for within-subject correlations and were adjusted for baseline age, sex, and biologic therapy during follow-up

Abbreviations: BMIZ body mass index-for-age z-score, HAZ height-for-age z-score, PCDAI Pediatric Crohn’s Disease Activity Index, SES-CD Simple Endoscopic Score for Crohn’s Disease, 25(OH)D 25-hydroxyvitamin D, CI confidence interval

Univariate analysis of ΔBMIZ

Univariate linear regression analysis was performed on 17 baseline variables, using the 6-month change in BMIZ as the outcome (Fig. 3). The results showed a significant positive correlation between FOBT positivity and ΔBMIZ (β = 0.66, 95% CI 0.29 to 1.02, P < 0.001), while perianal lesions were significantly negatively associated with ΔBMIZ (β = −0.51, 95% CI −0.89 to −0.13, P = 0.010). Furthermore, ESR, hemoglobin, 25(OH)D3, and biologic therapy showed potential predictive value. The positive association between FOBT and ΔBMIZ may suggest that some markers of active but reversible inflammatory burden identify children with greater potential for nutritional rebound after treatment.

Fig. 3.

Univariate associations between baseline clinical variables and 6-month change in BMI z-score (ΔBMIZ). Forest plot displays standardized regression coefficients (β) with 95% confidence intervals obtained from univariate linear regression analyses. Continuous variables were standardized to z-scores to allow direct comparison of effect sizes, while categorical variables were analyzed using clinically relevant reference categories. Variables with P < 0.20 were considered eligible for subsequent LASSO regression

LASSO regression screening of core predictors of ΔBMIZ

The variables screened in the univariate analysis were incorporated into the LASSO regression model. Using tenfold cross-validation and the λ1se criterion, six predictors were retained: PD, FOBT, biologic therapy, ESR, hemoglobin, and 25(OH)D3 (Fig. 4).

Fig. 4.

Predictor selection for ΔBMIZ using LASSO regression. A Coefficient profiles of candidate predictors generated by LASSO regression. Each line represents the coefficient trajectory of one candidate predictor as the regularization parameter λ changes. Six predictors were retained in the final model: perianal disease (PD), positive fecal occult blood test (FOBT), biologic therapy, erythrocyte sedimentation rate (ESR), hemoglobin (Hb), and 25-hydroxyvitamin D3 [25(OH)D3]. B 10-fold cross-validation plot for selection of the optimal λ. The two dashed vertical lines indicate the values of λ corresponding to the minimum mean cross-validated error and the 1-standard-error criterion, respectively

Multivariate model and nomogram construction

A multivariate linear regression model was constructed based on the LASSO screening results (Table 3). Perianal disease status was negatively associated with ΔBMIZ (β = −0.682, P < 0.001), while FOBT positivity (β = 0.434, P = 0.015) and biologic therapy (β = 0.443, P = 0.028) were independently associated with BMIZ improvement.

Table 3.

Multivariable linear regression model for predicting 6-month changes in BMI z-score (ΔBMIZ)

Predictor	β coefficient	SE	95% CI	P value
Perianal disease (Yes vs No)	−0.682	0.158	−1.001 to −0.363	< 0.001
FOBT (Positive vs Negative)	0.434	0.170	0.090 to 0.778	0.015
Biologic therapy (Yes vs No)	0.443	0.195	0.049 to 0.837	0.028
Vitamin D (25(OH)D3, ng/mL)	−0.009	0.011	−0.031 to 0.014	0.444
ESR (mm/h)	0.003	0.002	−0.002 to 0.008	0.232
Hemoglobin (g/L)	−0.004	0.005	−0.013 to 0.006	0.447

Regression coefficients were estimated using pooled results from multiple imputation datasets. Variables were selected using least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation

Abbreviations: BMIZ body mass index-for-age z-score, FOBT fecal occult blood test, ESR erythrocyte sedimentation rate, CI confidence interval

Based on this, a nutritional prognostic nomogram was constructed (Fig. 5) to predict individual 6-month ΔBMIZ. Perianal disease, FOBT, and biologic therapy had relatively high weights in the model.

Fig. 5.

Nomogram for predicting 6-month change in BMI z-score (ΔBMIZ) in pediatric Crohn’s disease. The nomogram integrates six core predictors identified by LASSO regression. For each predictor, points are assigned by projecting vertically to the “Points” axis. The sum of all points corresponds to the predicted ΔBMIZ at 6 months. All continuous variables are presented in their original clinical units to facilitate clinical application

Performance and calibration of the nomogram model

The nomogram model has a R² of 0.51, a MAE of 0.40, and an RMSE of 0.49 on the apparent data. After 1000 bootstrap internal validations, the calibration curves show good agreement between the predicted and observed values, with no significant overfitting observed (Fig. 6).

Fig. 6.

Calibration curve of the ΔBMIZ prediction model. The calibration plot compares predicted and observed ΔBMIZ values at 6 months. The dashed line represents ideal calibration, the dotted line indicates apparent performance, and the solid line shows bootstrap bias-corrected calibration based on 1000 resamples. The close agreement between the bias-corrected and ideal curves indicates good calibration of the model

Decision curve analysis for the prediction of inadequate nutritional recovery (ΔBMIZ ≤ 0.40)

The decision curve analysis for predicting inadequate nutritional recovery (ΔBMIZ ≤ 0.40) showed a positive net benefit across a threshold probability range of approximately 5% to 85%, indicating that applying the nomogram within this range yields a greater net benefit than default strategies. This suggests that the proposed model possesses good clinical utility (Fig. 7).

Fig. 7.

Decision curve analysis (DCA) for the prediction of inadequate nutritional recovery (ΔBMIZ ≤ 0.40). The y-axis measures the standardized net benefit. The red line represents the nomogram. The grey line represents the assumption that all patients have inadequate recovery. The black horizontal line represents the assumption that no patients have inadequate recovery. The decision curve shows that if the threshold probability is between 5 and 85%, using the nomogram to predict inadequate recovery provides a greater net benefit than treating all or treating no patients

Discussion

In this longitudinal cohort study of children with CD, we observed that patients with PD had significantly attenuated BMIZ improvement within the first six months post-diagnosis compared to those without PD. Importantly, this association remained significant after adjustment for age, sex, and biologic therapy in the LMM. These findings suggest that PD is not only a marker of aggressive disease phenotype, but also a clinically relevant indicator of impaired short-term nutritional recovery [16–18].

This distinction is important from a clinical perspective. PD is already recognized as a high-risk phenotype associated with more complicated disease behavior and earlier treatment escalation [17–19]. Our findings extend this understanding by showing that children with PD may remain vulnerable to suboptimal nutritional recovery even during the early treatment period. In practice, this means that the presence of PD may help identify children who require not only tighter inflammatory control, but also earlier and more intensive nutritional assessment, growth monitoring, and individualized supportive care during the post-diagnostic induction phase [4, 20].

Mechanistically, PD reflects significant transmural inflammation, which is a hallmark feature in East Asian pediatric CD populations [21, 22]. Chronic inflammation triggers the release of pro-inflammatory cytokines, such as TNF-α, which disrupts the growth hormone (GH)-IGF-1 axis, impairing growth and metabolism [4, 20, 23]. These processes result in delayed weight gain and impaired growth recovery. Additionally, the clinical manifestations of PD, including perianal pain, discomfort during defecation, and limited physical activity, further contribute to poor nutritional intake and energy imbalance, thereby hindering BMIZ normalization [20, 24]. Unlike some luminal inflammatory markers, which may improve rapidly with induction therapy, PD may continue to be associated with impaired nutritional recovery during the early post-diagnosis period [25].

A key finding is that several predictors, such as positive FOBT, biologic therapy, higher ESR, and lower hemoglobin, are associated with greater ΔBMIZ improvement. Although this seems counterintuitive at first, these markers may reflect active but reversible inflammatory burden, which allows for better nutritional recovery once inflammation is controlled. FOBT positivity was independently associated with greater ΔBMIZ improvement, consistent with recent studies in pediatric CD cohorts [26]. FOBT typically indicates active mucosal injury and intestinal bleeding, correlating with disease severity [27–29]. In East Asian pediatric CD patients, FOBT positivity may represent reversible inflammation rather than irreversible damage, facilitating a transition to an anabolic state after treatment [4, 30, 31]. Furthermore, biologic therapy was identified as a predictor in our model. However, since the initiation of biologics is closely linked to initial disease severity and clinical discretion, it may reflect confounding by indication rather than serving as an entirely independent prognostic factor. This nuance should be considered when applying the nomogram in clinical practice.

In the multivariable model, clinical indicators such as ESR, Hb, 25(OH)D3, folate, and vitamin B12 showed consistent clinical relevance, even if not statistically significant. These markers align with common inflammatory and nutritional profiles in Chinese pediatric CD patients [32]. ESR remains a key marker of systemic inflammation, which drives growth impairment in pediatric CD [4, 20]. Hemoglobin reflects the impact of chronic inflammation and iron metabolism, both critical in East Asian CD populations where anemia is prevalent [33]. Low vitamin D is associated with disease activity and bone health [34–36]. Folate and vitamin B12 deficiencies, linked to ileal involvement or restrictive diets, are critical for long-term neurological and hematopoietic risks [37–39].

Our results also confirm that BMIZ is more responsive than HAZ for short-term nutritional recovery, as weight-related indices react faster to caloric intake and inflammatory control, whereas linear growth shows a delay [40, 41]. Additionally, ΔBMIZ is less confounded by acute-phase responses than biochemical markers like albumin [42–44].

In this context, positive FOBT may indicate active mucosal injury rather than irreversible damage. Similarly, elevated ESR, anemia, and vitamin D deficiency reflect systemic inflammation and nutritional deficits at diagnosis. For children who achieve good inflammatory control, these baseline abnormalities may coexist with greater recovery potential. In contrast, PD represents a more persistent and aggressive disease state, with ongoing symptoms and functional compromise, which may explain why PD predicts poorer BMIZ recovery, while other markers are associated with greater improvement.

The nomogram developed in this study integrates routinely available clinical and laboratory variables and showed satisfactory internal performance. Clinically, it is intended to be applied at diagnosis or early after diagnosis, when baseline phenotype, inflammatory markers, and nutritional indicators are available. Children predicted to have poor 6-month BMIZ recovery may benefit from closer growth surveillance, earlier dietitian involvement, more intensive nutritional assessment, and timely optimization of anti-inflammatory therapy. Therefore, the model is best viewed as an adjunctive risk-stratification tool to support, rather than replace, individualized clinical judgment.

This study has several limitations. First, this study is constrained by its retrospective, single-center design. Although treatment variables were adjusted for in the multivariable analyses, residual confounding by baseline disease severity cannot be excluded variables such as PCDAI, SES-CD, disease extent, and treatment allocation may partly reflect underlying disease burden and could have influenced the observed associations. Second, our sample size was relatively small. Although penalized regression and bootstrap resampling were utilized to mitigate the risk, the model coefficients may still carry some degree of instability, and potential model optimism or overfitting cannot be entirely ruled out, despite internal validation. Third, the cohort was derived from a tertiary referral center in southern China, where the prevalence of PD was relatively high. Because East Asian pediatric CD populations may differ from Western cohorts in phenotype distribution and perianal involvement, the generalizability of the model to other settings remains uncertain. Consequently, our findings should be interpreted with caution, and external validation in larger, multicenter prospective cohorts is required before broader clinical application.

Future studies should examine whether early ΔBMIZ predicts longer-term outcomes such as final height, bone health, body composition, and sustained remission. Additional incorporation of detailed nutritional exposure data, serial inflammatory markers, and more granular measures of perianal disease severity may also improve mechanistic understanding and predictive performance.

Conclusion

In conclusion, we developed and internally validated a nomogram for predicting 6-month ΔBMIZ in children with Crohn’s disease. Perianal disease was independently associated with poor short-term nutritional recovery. This model, which integrates phenotypic, inflammatory, and nutritional markers, can aid in early risk stratification and personalized nutritional management. However, external validation in larger multicenter cohorts is needed for broader clinical application.

Abbreviations

CD

Crohn’s disease

PD

Perianal disease

BMIZ

Body mass index Z-score

LMM

Linear mixed-effects model

ΔBMIZ

The 6-month change in BMIZ

WBC

White blood cell count

CRP

C-reactive protein

PLT

Platelet count

25(OH)D3

25-Hydroxyvitamin D3

HCT

Hematocrit

ESR

Erythrocyte sedimentation rate

Hb

Hemoglobin

HAZ

Height-for-age Z-score

FC

Fecal calprotectin

FOBT

Fecal occult blood test

Authors’ contributions

Conceptualization, Y.Z. and P.W.; methodology, Y.Z. and P.W.; software, Y.Z. and P.W.; formal analysis, Y.Z. and P.W.; investigation, X.P., B.M., Y.L. and X.Z.; data curation, X.P., B.M., Y.L. and X.Z.; writing—original draft preparation, Y.Z. and P.W.; writing—review and editing, S.C.; visualization, Y.Z. and P.W.; supervision, S.C.; project administration, S.C. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by Startup Fund for Scientific Research of Fujian Medical University (Grant No. 2024QH1073).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of the First Affiliated Hospital of Fujian Medical University (Approval No. MRCTA, ECFAH FMU [2019] 219). Written informed consent was obtained from the parents or caregivers. This study was conducted in accordance with the ethical principles of the 1964 Declaration of Helsinki and its later amendments. Clinical trial number: not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.