Laporkan Masalah

Pengembangan dan Validasi Skor Prediksi Kematian pada Anak dengan Covid-19

Gladys Indika Danudibroto, Dr. dr. Rina Triasih, M. Med (Paed), Ph. D, Sp. A (K).; dr. Agung Triono, Sp. A (K)

2022 | Tesis | MAGISTER ILMU KEDOKTERAN KLINIS

Latar belakang: Identifikasi awal anak dengan COVID-19 dalam periode rawat inap merupakan suatu hal yang penting untuk mengoptimalisasi tatalaksana dan menyampaikan prognosis pasien. Tujuan penelitian: Mengembangkan dan memvalidasi skor prediksi kematian COVID-19 anak. Metode penelitian: Studi kohort retrospektif dengan subyek usia 1 bulan-18 tahun yang menjalani rawat inap di RSUP Dr. Sardjito Yogyakarta, Indonesia dengan periode Maret 2020 - September 2021. Variabel prediktor kematian dinilai saat admisi rumah sakit dan dihitung nilai likelihood ratio (LR) positif dan negatif. Variabel dengan LR positif >=2 dan negatif 0,5 akan disusun menjadi suatu skor dengan metode Spiegelhalter Knill-Jones. Area Under the Curve for Receiver Operating Characteristic (AUC-ROC) digunakan untuk menentukan nilai cut-off dari total skor. Validasi internal dilakukan dengan menilai sensitivitas, spesifisitas, likelihood ratio positif, negatif dan akurasi. Hasil penelitian: Pengembangan sistem skor melibatkan 136 subyek, 21 subyek (15,4%) memiliki luaran meninggal pada. Dari 13 variabel, enam variabel merupakan faktor prediktor kematian COVID-19 anak; tidak ada komorbiditas (LR 0,36; nilai skor -4), tidak ada takipnea (LR 0,34; nilai skor -1), ada hipoksemia (LR 3,85; nilai skor 7), tidak ada hipoksemia (LR 0,50; nilai skor -4), peningkatan kadar BUN (LR 3,08; nilai skor 7), ada MISC (LR 6,8; nilai skor 13), ada indikasi perawatan PICU (LR 2,26; nilai skor 5) dan tidak ada indikasi perawatan PICU (LR 0,29; nilai skor -7). Nilai AUC pada pengembangan skor adalah 0,91 (IK 95% 0,86 - 0,97), dengan nilai cut-off >=2,5. Delapan puluh empat subyek dilakukan validasi eksternal dengan nilai AUC 0,86 (IK95% 0,73 - 0,99), sensitivitas 78,57%, spesifisitas 85,71%, LR (+) 5,49, LR (-) 0,25 dan akurasi 84,52%. Kesimpulan: Total skor >=2,5 dapat memprediksi secara valid mortalitas diantara anak yang menjalani rawat inap dengan COVID-19. Validasi eksternal dibutuhkan untuk meningkatkan kemampuan diagnostik skor prediksi COVID-19 anak.

Background: Early identification of children who hospitalized with COVID-19 is a great importance and may aid in delivering proper treatment and optimizing use of resources. Aims: To develop and to validate the Pediatrics Covid-19 Mortality Prediction Score in hospitalized children with COVID-19. Methods: A retrospective cohort study of children (1 months to 18 years old) was conducted from March 2020 -September 2021 in Dr. Sardjito Hospital, Yogyakarta, Indonesia. All variables whom the likelihood ratio positive >=2 and negative 0.5 were developed as models of scoring by Spiegelhalter Knill-Jones method. The Area Under the Curve for Receiver Operating Characteristic (AUC- ROC) was used to assess the cut-off point of the total score. Internal validation was obtained to evaluate the sensitivity, specificity, positive-negative likelihood ratio, and accuracy of scoring. Results: The development of scoring system included 136 patients, twenty-one (15.4%) were died during their hospitalization. From 13 variables, six were predictive factors of mortality. They are comorbidity (LR- 0.36; score -4), rapid breathing (LR- 0.34; score -1), hypoxemia (LR+ 3.85; score +7 and LR- 0.50; score -4), increase of ureum level (LR+ 3.08; score +7), MISC (LR+ 6.8; score +13), and needs of intensive care (LR 0.29; score -7). The mean AUC in the development cohort was 0.91 (CI95% 0.86 - 0.97) at an optimal cut-off value of >=2.5. The validation cohort included 84 patients, fourteen (16.67%) were died. The mean AUC with cut-off value >=2.5 of validation cohort was 0.86 (CI95% 0,73 - 0,99), showing a sensitivity 78.57%, specificity 85.71%, LR+ of 5.49, LR- of 0.25, and accuracy of 84.52%. Conclusion: Score >=2.5 demonstrated a valid tool for predicting mortality among children who hospitalized with COVID-19. External validation is required to improve the predictive performance of our models scoring.

Kata Kunci : COVID-19, anak, mortalitas, prediksi, skor, children, mortality, prediction, score

  1. S2-2022-484888-abstract.pdf  
  2. S2-2022-484888-bibliography.pdf  
  3. S2-2022-484888-tableofcontent.pdf  
  4. S2-2022-484888-title.pdf