VALIDITY OF E-PPGBM BASED NUTRITIONAL STATUS DATA IN DETERMINING STUNTING PREVALENCE IN NORTH TORAJA REGENCY

Abstract

Accelerating stunting reduction is challenging during planning and budgeting, which requires accurate data. The significant discrepancy between the national survey results and e-PPGBM data raises crucial questions regarding the accuracy and utility of each dataset in the context of policy formulation. This study aimed to assess the differences in stunting prevalence between E-PPGBM and gold standard measurements and to evaluate the diagnostic accuracy and agreement of height-for-age z-scores (HAZ). A cross-sectional study was conducted among 188 children aged 0–59 months, randomly selected from North Toraja Regency. Anthropometric measurements were performed by community health cadres (E-PPGBM) and trained enumerators using standardized instruments (gold standard). Data were analyzed using receiver operating characteristic (ROC) curves, Bland–Altman plots, and descriptive statistics. The AUC value for detecting stunting was 0.925 (95% CI: 0.881–0.970), with a sensitivity of 70.7 persen, specificity of 95.6 percent, PPV of 91.4 percent, and a negative predictive value (NPV) 83.1 percent. The prevalence of stunting based on E-PPGBM (30.9%) was lower than the gold standard (39.9%), indicating underdiagnosis. The mean difference in HAZ values was 0.12, with 95 percent limits of agreement ranging from –1.28 to 1.53. E-PPGBM demonstrated excellent diagnostic accuracy but limited sensitivity, with a tendency to overestimate HAZ values. Strengthening cadre training, supervision, and standardization of anthropometric tools is necessary to enhance the reliability of stunting surveillance data.

Keywords: E-PPGBM, stunting, community health cadres, measurement accuracy

ABSTRAK

Salah satu tantangan dalam percepatan penurunan stunting terletak pada tahap perencanaan dan penganggaran yang membutuhkan data dan informasi yang akurat. Namun, perbedaan signifikan antara hasil survei nasional dan data E-PPGBM menimbulkan pertanyaan krusial tentang keakuratan dan kegunaan masing-masing data dalam konteks kebijakan. Penelitian ini bertujuan untuk menganalisis perbedaan prevalensi stunting antara E-PPGBM dengan gold standard serta menilai akurasi diagnostik dan tingkat kesepakatan pengukuran HAZ. Desain penelitian yang digunakan adalah cross-sectional dengan populasi balita usia 0-59 bulan di Kabupaten Toraja utara. Sebanyak 188 balita dipilih menggunakan simple random sampling. Pengukuran panjang/tinggi badan dilakukan oleh kader posyandu (E-PPGBM) dan enumerator terlatih sebagai referensi dengan alat terstandar (gold standard). Data dianalisis menggunakan uji diagnostik (Kurva ROC), analisis Bland-Altman, dan statistik deskriptif. Hasil penelitian menunjukkan nilai AUC E-PPGBM untuk mendeteksi stunting adalah 0,925 (95% CI: 0,881–0,970), yang mengindikasikan akurasi sangat baik. Sensitivitas sebesar 70,7 persen, spesifisitas 95,6 persen, NPV 83,1 persen, dan PPV 91,4 persen. Rata-rata selisih HAZ adalah 0,12 dengan 95 persen Limits of Agreement (LoA) -1,28 hingga 1,53, menunjukkan variabilitas yang besar. Prevalensi stunting menurut E-PPGBM (30,9%) lebih rendah dibandingkan gold standard (39,9%), mengindikasikan underdiagnosis. Penelitian ini menemukan bahwa perbedaan HAZ antara E-PPGBM dengan gold standard menunjukkan kecenderungan HAZ lebih tinggi pada E-PPGBM pada semua kelompok umur. Diperlukan peningkatan kapasitas kader dan standarisasi alat untuk meningkatkan kualitas data E-PPGBM.

Kata kunci: E-PPGBM, stunting, kader posyandu, akurasi diagnostik