# Stage 2 — First-draft manuscript (literature-grounded) _Every Results bullet is hung off a Stage-1 ledger finding. Every Discussion paragraph carries its own per-finding literature evidence pack._ ## Background (first-draft) Lower respiratory tract infection (LRTI) is a leading cause of paediatric morbidity worldwide, and biochemical disturbances such as hypocalcaemia have been described variably across settings. Ionised calcium is the physiologically active fraction, but total calcium is more widely available; the two are commonly used interchangeably in paediatric practice without rigorous within-cohort comparison. This single-centre cross-sectional audit on a synthetic teaching dataset (n = 100, ages 1-5 y) was designed around 7 pre-specified findings: cohort prevalence by each definition, total-ionised concordance, continuous agreement, the reference-range comparison, and a pre-specified logistic-regression test of age as a predictor of hypocalcaemia. Each finding below is paired with the literature evidence used to interpret it. ## Results (first-draft, by ranked finding) - **#1 — Primary descriptive outcome (high)**: Hypocalcemia by total calcium: 25.0% (95% CI 17.5–34.3%, n=25/100). Roughly one in four hospitalised paediatric LRTI patients in this cohort meets the laboratory definition of hypocalcaemia. This is the headline cohort prevalence and is reported with a Wilson 95 % confidence interval to retain nominal coverage at the boundary of the parameter space. - **#2 — Primary descriptive outcome (high)**: Hypocalcemia by ionised calcium: 26.0% (95% CI 18.4–35.4%, n=26/100). Roughly one in four hospitalised paediatric LRTI patients in this cohort meets the laboratory definition of hypocalcaemia. This is the headline cohort prevalence and is reported with a Wilson 95 % confidence interval to retain nominal coverage at the boundary of the parameter space. - **#3 — Diagnostic concordance (high)**: Total–ionised hypocalcemia concordance: Cohen's κ = 0.9737 (almost perfect; observed agreement 0.99). Cohen's kappa near 1 (with observed agreement = 0.99) means total and ionised calcium classify the same patients as hypocalcaemic in this cohort — the two definitions are operationally interchangeable for case-finding here. - **#4 — Method comparison (high)**: Total and ionised calcium correlate Pearson r = 0.7693 (p = 0.0); R² = 0.5918. Pearson r = 0.77 with R² ≈ 0.59 indicates a strong but imperfect linear relationship between total and ionised calcium on the continuous scale — total calcium explains about 59 % of the variance in ionised calcium. - **#5 — Reference-range comparison (moderate)**: Ionised calcium Wilcoxon signed-rank differs from pediatric reference midpoint (p = 0.0, q-FDR = 0.0); 26.0% below reference, 0.0% above. The ionised calcium distribution differs significantly from the centre of the published paediatric reference interval after BH-FDR correction — 26 % of cohort values fall below the lower reference limit. - **#6 — Predictive modelling — null result (null)**: the age-only total-calcium model: CV-AUC = 0.472 (in-sample 0.5304, optimism 0.0584). Logistic regression of hypocalcaemia on age within the 1–5 y window did not discriminate (CV-AUC < 0.65 on 5-fold stratified cross-validation). This null result is reported in compliance with the pre-specified deployment threshold. - **#7 — Predictive modelling — null result (null)**: the age-only ionised-calcium model: CV-AUC = 0.5327 (in-sample 0.5372, optimism 0.0044). Logistic regression of hypocalcaemia on age within the 1–5 y window did not discriminate (CV-AUC < 0.65 on 5-fold stratified cross-validation). This null result is reported in compliance with the pre-specified deployment threshold. ## Discussion (first-draft, by ranked finding) ### Finding #1: Primary descriptive outcome **Cohort result.** Hypocalcemia by total calcium: 25.0% (95% CI 17.5–34.3%, n=25/100). **Interpretation.** Roughly one in four hospitalised paediatric LRTI patients in this cohort meets the laboratory definition of hypocalcaemia. This is the headline cohort prevalence and is reported with a Wilson 95 % confidence interval to retain nominal coverage at the boundary of the parameter space. **Reader implication.** Treat as the principal descriptive estimate when comparing this cohort to published prevalence figures. **Literature context.** Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:34151783], [PMID:35265997], [PMID:35000906], [PMID:36773844], [PMID:33378060], [PMID:23273676] Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:35288108], [PMID:37702687], [PMID:36524863], [PMID:35934870], [PMID:34719383], [PMID:40785088], [PMID:39431864], [PMID:34008378] ### Finding #2: Primary descriptive outcome **Cohort result.** Hypocalcemia by ionised calcium: 26.0% (95% CI 18.4–35.4%, n=26/100). **Interpretation.** Roughly one in four hospitalised paediatric LRTI patients in this cohort meets the laboratory definition of hypocalcaemia. This is the headline cohort prevalence and is reported with a Wilson 95 % confidence interval to retain nominal coverage at the boundary of the parameter space. **Reader implication.** Treat as the principal descriptive estimate when comparing this cohort to published prevalence figures. **Literature context.** Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:34151783], [PMID:35265997], [PMID:35000906], [PMID:36773844], [PMID:33378060], [PMID:23273676] Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:35288108], [PMID:37702687], [PMID:36524863], [PMID:35934870], [PMID:34719383], [PMID:40785088], [PMID:39431864], [PMID:34008378] ### Finding #3: Diagnostic concordance **Cohort result.** Total–ionised hypocalcemia concordance: Cohen's κ = 0.9737 (almost perfect; observed agreement 0.99). **Interpretation.** Cohen's kappa near 1 (with observed agreement = 0.99) means total and ionised calcium classify the same patients as hypocalcaemic in this cohort — the two definitions are operationally interchangeable for case-finding here. **Reader implication.** In settings where ionised calcium is not routinely available, total calcium is a reasonable substitute for binary hypocalcaemia screening in this age band on this dataset; external validation in real cohorts is required. **Literature context.** Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:24658220] ### Finding #4: Method comparison **Cohort result.** Total and ionised calcium correlate Pearson r = 0.7693 (p = 0.0); R² = 0.5918. **Interpretation.** Pearson r = 0.77 with R² ≈ 0.59 indicates a strong but imperfect linear relationship between total and ionised calcium on the continuous scale — total calcium explains about 59 % of the variance in ionised calcium. **Reader implication.** Substituting total for ionised on the continuous scale loses ≈ 41 % of the variance. The mg/dL Bland-Altman plot quantifies the limits of agreement directly. **Literature context.** Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:24658220] ### Finding #5: Reference-range comparison **Cohort result.** Ionised calcium Wilcoxon signed-rank differs from pediatric reference midpoint (p = 0.0, q-FDR = 0.0); 26.0% below reference, 0.0% above. **Interpretation.** The ionised calcium distribution differs significantly from the centre of the published paediatric reference interval after BH-FDR correction — 26 % of cohort values fall below the lower reference limit. **Reader implication.** Confirms that the cohort skews lower on ionised calcium than the reference population; total calcium does not show the same shift. **Literature context.** Relevant cited records for this evidence pack are listed for traceability; no additional causal, prognostic, or population-transfer claim is made from them unless directly measured in the present cohort. [PMID:35288108], [PMID:37702687], [PMID:36524863], [PMID:35934870], [PMID:34719383], [PMID:40785088], [PMID:39431864], [PMID:34008378] ### Finding #6: Predictive modelling — null result **Cohort result.** the age-only total-calcium model: CV-AUC = 0.472 (in-sample 0.5304, optimism 0.0584). **Interpretation.** Logistic regression of hypocalcaemia on age within the 1–5 y window did not discriminate (CV-AUC < 0.65 on 5-fold stratified cross-validation). This null result is reported in compliance with the pre-specified deployment threshold. **Reader implication.** Age alone is not a useful predictor of hypocalcaemia in this age band; do not deploy these models for clinical decision-support. **Literature context.** No directly comparable paediatric LRTI evidence was identified in the pre-specified literature search for this finding. ### Finding #7: Predictive modelling — null result **Cohort result.** the age-only ionised-calcium model: CV-AUC = 0.5327 (in-sample 0.5372, optimism 0.0044). **Interpretation.** Logistic regression of hypocalcaemia on age within the 1–5 y window did not discriminate (CV-AUC < 0.65 on 5-fold stratified cross-validation). This null result is reported in compliance with the pre-specified deployment threshold. **Reader implication.** Age alone is not a useful predictor of hypocalcaemia in this age band; do not deploy these models for clinical decision-support. **Literature context.** No directly comparable paediatric LRTI evidence was identified in the pre-specified literature search for this finding. ## Discussion — additional graph-derived evidence (Stage 3) The following clinically-significant edges were carried by the literature graph but were not surfaced by the per-finding evidence packs. They are appended here for integration into the Discussion: - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:25916433] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:33391183] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:34636899] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:42033311] - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:42033311] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41991079] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41991079] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38146724] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38146724] - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38146724] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41991079] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38146724] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:39431864] - *Albumin.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37478022] - *Magnesium.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37478022] - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38529460] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41991079] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41991079] - *Fgf23.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37850343] - *Magnesium.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:35265997] - *Magnesium.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:34008378] - *Mortality.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:33378060] - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:38529460] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37850343] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:41968081] - *Vitamin D.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:33982091] - *Magnesium.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:34273186] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:36524863] - *Klotho.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37850343] - *Phosphate.* Flagged by the literature graph as context not measured in the present cohort; source records: [PMID:37850343]