Clinical controls and model calibration as keys to complete blood count-based studies of pediatric fistula-in-ano

Abstract

This commentary appraises a single-center evaluation of complete blood count (CBC) markers for pediatric fistula-in-ano. We commend the pragmatic focus of this study on low-cost tests and the dual assessment of discrimination and clinical utility. To strengthen robustness and generalizability, two priorities are emphasized: First, the use of clinical controls, children with perianal symptoms but without pediatric fistula-in-ano, to avoid spectrum bias; and second rigorous calibration and validation (k-fold or temporal splits, bootstrap optimism correction, calibration intercept/slope, and Brier score), with decision curve analyses derived from validated predictions. Given rapid hematologic maturation, age should be modeled nonlinearly and CBC values should be expressed as age-specific z-scores/percentiles. Additional refinements include clear timing/indication of blood draws, appropriate correlations (Spearman or age-adjusted partials), and odds ratios per interquartile range with multicollinearity checks (variance inflation factor). These steps can translate promising CBC signals into reliable, clinically actionable evidence.

Key Words: Pediatric fistula-in-ano; Complete blood count; Neutrophil-to-lymphocyte ratio; Spectrum bias; Age standardization; Receiver operating characteristic; Decision curve analysis

Core Tip: This study evaluated widely accessible complete blood count markers in a large pediatric fistula-in-ano cohort and reported promising discrimination and decision-making ability. We highlight design and analytic refinements, clinical control selection, rigorous age handling, transparent model calibration, and validated decision curve analyses that can convert these encouraging signals into robust, generalizable evidence.

TO THE EDITOR

We read with genuine interest the authors’ investigation of complete blood count (CBC) parameters and derived indices, such as the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, monocyte-to-lymphocyte ratio, systemic inflammatory response index, and systemic immune-inflammation index, for the diagnosis and clinical decision-making of pediatric fistula-in-ano (PFIA) (World Journal of Gastroenterology)[1]. This work has several notable strengths: A sizable single-center cohort, reliance on low-cost and universally obtainable tests, and a dual emphasis on discrimination [receiver operating characteristic (ROC)] and clinical utility [decision curve analysis (DCA)]. These features collectively underscore the translational promise of CBC-based markers in routine pediatric surgical practice. At the same time, several design and statistical considerations, if addressed, would materially strengthen the credibility and applicability of the conclusions.

Control selection and spectrum bias

The comparison against healthy children is informative for biological contrast but diverges from the diagnostic question faced in practice: Distinguishing PFIA from symptomatic non-PFIA presentations (such as perianal dermatitis, fissure, and nonfistulizing abscess). Healthy controls can inflate separation and net-benefit estimates[2]. We encourage a complementary analysis using age-matched clinical controls, which would more closely mirror triage and referral decisions.

Age handling in an infant-heavy cohort

CBC values change rapidly across early life (for example, physiologic lymphocytosis)[3]. The apparent differences, higher lymphocyte counts and lower composite ratios among cases, may be partly age-structured. Incorporating restricted cubic splines for age, converting CBCs to age-specific z-scores/percentiles, and presenting age-stratified effect sizes (or propensity-matched analyses within ± 1-2 months) would clarify disease-specific signals beyond maturation effects.

Timing and indication of blood draws

If cases were sampled at admission and controls at routine consultation, acute intercurrent illness, antimicrobial exposure, or periprocedural states could transiently alter leukocyte profiles. Reporting fever/infection status, recent medications, and the symptom-to-phlebotomy interval and conducting sensitivity analyses excluding acute infections would mitigate indication bias.

Statistical implementation and reporting

Association methods: Mantel’s test targets correlations between distance matrices and is not well aligned with the variable-to-variable association in this context[4]. Spearman’s test (and age-adjusted partial correlations) provides a clearer link between variables and clinical interpretation.

Model specification and calibration: Reporting odds ratios per interquartile range, checking nonlinearity (splines) and multicollinearity (variance inflation factors), and providing calibration metrics (intercept, slope, Brier score with calibration plots) will enhance transparency[5].

Validation and decision curves

ROC and DCA estimated on the same dataset risk optimistic performance. We recommend k-fold cross-validation or temporal holdout prior to thresholding, bootstrap optimism-correction for the area under the curve and net benefit, and DCAs computed from validated predictions[6]. Clear figure legends stating model formulas, covariate scaling, threshold ranges, and the “treat-all/treat-none” references will aid interpretability[7]. As currently drawn, panels A-B of Figure 4 are ROC curves, whereas panels C-D are DCAs.

Formatting consistency regarding sex stratification

The manuscript states that only male children were included; therefore, figures and text should not indicate sex-stratified analyses. Specifically, the title/Legend of Figure 3 should be corrected from “according to age and sex” to “according to age”, and any remaining references to sex-stratified results should be removed for consistency with the inclusion criteria.

CONCLUSION

This study makes a valuable contribution by exploring pragmatic, widely accessible hematologic markers in a large pediatric PFIA cohort, an approach with clear clinical relevance and scalability. Implementing the above refinements would not only address potential sources of bias but also determine whether CBC-derived indices retain their promising performance in settings that reflect real-world decision-making. If so, the field will gain a low-cost, implementable tool to support early recognition and management of PFIA.