As shown in Table 1, children with hyperphosphatemia at and during admission had lower ages and weights, higher severity scores and longer lengths of stay in the PICU. Fifty percent of the deceased patients developed hyperphosphatemia during their admission compared to 20.8% of the survivors (p = 0.041).

Hyperphosphatemia occurred more frequently in patients treated with corticosteroids (33.3% vs. 19.8%; p = 0.030), vasoactive drugs (36.5% vs. 10.5%; p < 0.001), furosemide (39.2% vs. 6.4%; p < 0.001), acetazolamide (50% vs. 22%; p = 0.017), vitamin D (47.6% vs. 21.2%; p = 0.006) and iron (35% vs. 20.2%; p = 0.043).

Multivariate analysis revealed that hyperphosphatemia at admission was associated with younger age, surgical admission and a higher Pediatric Multiple Organ Dysfunction Score (P-MODS). Hyperphosphatemia during admission was associated with younger age, surgical admission, noninvasive mechanical ventilation (NIMV), artificial nutrition, renal replacement therapy (RRT), and a longer duration of admission, as has been described in adults7 (Table 2).

In patients undergoing RRT, phosphorus levels decrease if dialysis fluids are not supplemented with phosphate,8 but our patients received phosphate supplements in their RRT fluids. Perhaps because of this, they more frequently presented hyperphosphatemia.6

Hyperphosphatemia was more frequent in patients who required mechanical ventilation, which coincides with the observations of other studies. No pathophysiological reason has been found to explain this association. Patients with hyperphosphatemia had higher severity scores, suggesting that hyperphosphatemia could be a marker of severity in critically ill children.6,7,9

Hypophosphatemia occurred more frequently than hyperphosphatemia, which aligns with data found in the literature.6 The main causes of hypophosphatemia are intracellular redistribution due to respiratory alkalosis, decreased intake, or hemodilution when phosphate-free fluids are used.3

Hypophosphatemia was more prevalent among medical patients, those subjected to NIMV and PN, and those who died upon admission to the PICU. Patients presenting hypophosphatemia at and during admission were older and had a greater body weight (Table 1).

Some studies have found an association between hypophosphatemia and the duration of mechanical ventilation.1,6 The univariate study showed an association between hypophosphatemia upon admission and NIMV; however, this association was not confirmed in the multivariate study. We found no association between hypophosphatemia and PN, corticosteroids, furosemide or vasoactive drugs, as previously reported by other authors.1,10

The univariate analysis revealed an association between hypophosphatemia at admission and mortality. A total of 72.5% of the deceased patients had hypophosphatemia at admission compared to 19.4% of the survivors; p = 0.003. However, this association was not maintained in the multivariate analysis.

In the multivariate analysis, hypophosphatemia at admission was associated with older age, medical causes of admission and lower P-MODS scores. Hypophosphatemia during admission was only associated with older age (Table 2).

The main limitations of our study are its retrospective design, which precludes the establishment of causal relationships, and the fact that phosphate levels were not determined for some patients, which may cause the incidence of the alterations to be overestimated.

In conclusion, phosphate disturbances are frequent in children admitted to the PICU. Hyperphosphatemia was more prevalent in surgical patients, smaller patients, children subjected to NIMV, patients receiving artificial nutrition and patients with RRT. It was also associated with a longer duration of admission. Hypophosphatemia upon admission and hyperphosphatemia during admission were more prevalent in deceased patients, though a causal relationship could not be established.

All authors have made substantial contributions in each of the following areas:

Nuria del Amo-Carramiñana, Paula Lasarte-Merino, Celia Pascual-Alonso and Rafael González-Cortés: acquisition, analysis and interpretation of the data, drafting of the article, and final approval of the manuscript.

Jesús López-Herce: conception and design of the study, acquisition, analysis and interpretation of the data, drafting of the article, critical review of the intellectual content, and the final approval of the manuscript.

The authors declare that they have not used artificial intelligence for any activities related to the study or in the writing of the manuscript.

The authors have not received any funding to conduct this study.