Scholarlink

Background <p>Undernutrition significantly increases the risk of severe infections and mortality in children under five, particularly in low- and middle-income countries. Pneumonia, a leading cause of childhood death, is especially dangerous in undernourished children, yet prognostic measures to identify those at highest risk are lacking.</p> Objective <p>To identify algorithms of poor prognosis in undernourished children with clinical pneumonia for early identification of children at risk for poor outcomes.</p> Methods <p>This study analyzed a subset of children enrolled in a cohort designed to identify biomarkers of bacterial pneumonia. Children aged 2–59 months with clinical pneumonia were recruited from two rural Gambian hospitals. Clinical and anthropometric data were collected at baseline, during hospitalization, and at 30-day follow-up. Nutritional status was classified using WHO definitions for stunting (height-for-age Z-scores) and wasting (weight-for-height Z-scores) as severe (Z-scores ≤ -3), moderate (-2 ≥ Z-scores &gt; -3), and mild (-1 ≥ Z-scores &gt; -2). Prognostic outcomes were classified into good and poor. Poor prognosis included death, prolonged hospital stay (≥ 7 days), post-discharge care-seeking, and difficult to feed during admission. Good prognosis was based on a hospital stay &lt; 3 days, with good outcomes within 30 days of the initial visit. Classification tree models and penalized logistic regression models (fit through elastic net) were used to identify combinations of predictors of poor prognosis (prognostic signatures).</p> Results <p>A total of 246 children with clinical pneumonia and undernutrition (wasting or stunting) were included. Children with poor prognosis presented more frequently with respiratory distress, hypoxemia, reduced capacity oforal feeding difficulty, and anemia. As expected, undernutrition was associated with adverse outcomes. The final prognostic algorithms were accurate to identify undernourished children at risk of poor prognosis: with sensitivity and specificity &gt; 80% and area under the receiver operating characteristic curve ≥ 0.80. Furthermore, we identified accurate prognostic signatures among children with both wasting and stunting.</p> Conclusion <p>Measures collected at admission in undernourished children with clinical pneumonia can identify those at risk of poor outcomes. The prognostic signatures developed in this study may inform early risk stratification and guide timely intervention, particularly in resource-limited settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prognosis of clinical pneumonia in undernourished children in rural Gambia

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月25日

查看全文
  • Yasir Shitu Isa,
  • Megan Carelus,
  • Isabelle Silber,
  • Rasheed Salaudeen,
  • Golam Sarwar,
  • Yekini Ajao Olatunji,
  • Ilias Hossain,
  • Isaac Osei,
  • Galega Lobga,
  • Banjo Adeshola,
  • Ousman Barjo,
  • Momodou M. Drammeh,
  • Umberto D’Alessandro,
  • Patricia L. Hibberd,
  • Grant A. Mackenzie,
  • Clarissa Valim

摘要

Background

Undernutrition significantly increases the risk of severe infections and mortality in children under five, particularly in low- and middle-income countries. Pneumonia, a leading cause of childhood death, is especially dangerous in undernourished children, yet prognostic measures to identify those at highest risk are lacking.

Objective

To identify algorithms of poor prognosis in undernourished children with clinical pneumonia for early identification of children at risk for poor outcomes.

Methods

This study analyzed a subset of children enrolled in a cohort designed to identify biomarkers of bacterial pneumonia. Children aged 2–59 months with clinical pneumonia were recruited from two rural Gambian hospitals. Clinical and anthropometric data were collected at baseline, during hospitalization, and at 30-day follow-up. Nutritional status was classified using WHO definitions for stunting (height-for-age Z-scores) and wasting (weight-for-height Z-scores) as severe (Z-scores ≤ -3), moderate (-2 ≥ Z-scores > -3), and mild (-1 ≥ Z-scores > -2). Prognostic outcomes were classified into good and poor. Poor prognosis included death, prolonged hospital stay (≥ 7 days), post-discharge care-seeking, and difficult to feed during admission. Good prognosis was based on a hospital stay < 3 days, with good outcomes within 30 days of the initial visit. Classification tree models and penalized logistic regression models (fit through elastic net) were used to identify combinations of predictors of poor prognosis (prognostic signatures).

Results

A total of 246 children with clinical pneumonia and undernutrition (wasting or stunting) were included. Children with poor prognosis presented more frequently with respiratory distress, hypoxemia, reduced capacity oforal feeding difficulty, and anemia. As expected, undernutrition was associated with adverse outcomes. The final prognostic algorithms were accurate to identify undernourished children at risk of poor prognosis: with sensitivity and specificity > 80% and area under the receiver operating characteristic curve ≥ 0.80. Furthermore, we identified accurate prognostic signatures among children with both wasting and stunting.

Conclusion

Measures collected at admission in undernourished children with clinical pneumonia can identify those at risk of poor outcomes. The prognostic signatures developed in this study may inform early risk stratification and guide timely intervention, particularly in resource-limited settings.