Tesis

Model Prediksi Perawatan-kembali 30 hari pada Pasien Usia Lanjut yang Dirawat di Ruang Rawat Inap RS Cipto Mangunkusumo = Prediction model for 30-day unplanned readmission in hospitalized elderly patient at Cipto Mangunkusumo Hospital.

Latar belakang: Perawatan-kembali 30 hari merupakan salah satu parameter penting yang berhubungan dengan biaya kesehatan tinggi dan outcome yang buruk, namun hal ini berpotensi dicegah. Usia lanjut merupakan kelompok yang rentan mengalami perawatan dengan karakteristik khusus yang dapat dinilai dengan pengkajian paripurna pasien geriatri (P3G). Beberapa penelitian menunjukkan komponen P3G merupakan faktor prognostik perawatan-kembali pada pasien usia lanjut sehingga dapat digunakan sebagai model prediksi perawatan-kembali 30 hari pada populasi ini. Belum ada penelitian prospektif yang khusus menilai komponen P3G sebagai model prediksi perawatan-kembali 30 hari. Tujuan: Mengembangkan model prediksi perawatan-kembali 30 hari pada pasien usia lanjut yang dirawat di bangsal medik RS Cipto Mangunkusumo. Metode: Penelitian adalah studi kohort prospektif pada 263 subjek usia 60 tahun yang diikuti hingga 30 hari pasca rawat. Data demografis dan komponen P3G dikumpulkan melalui wawancara dan rekam medik saat perawatan. Analisis kesintasan secara bivariat dan multivariat berjenjang dilakukan untuk mendapatkan hazard ratio. Dikembangkan suatu model prediksi dan persamaan fungsi hazard untuk memprediksi risiko perawatan-kembali 30 hari pasca rawat. Komponen P3G yang diukur adalah skor FRAIL (fatigue, resistance, ambulance, illness, loss of weight), Geriatric Depression Scale-15 (GDS-15), Mini nutrition Assessment short form (MNA-SF), Activity Daily Living (ADL)-Barthel index, Cumulative illness rating scale-geriatric (CIRS-G), Zarits-4 item screening test, uji Mini Cog, dan polifarmasi. Hasil: Status nutrisi dan status depresi berhubungan secara signifikan dengan perawatan-kembali 30 hari dengan HR 2,368 (IK95%: 1,412–3,972, p=0,001) dan HR 1,627 (IK95%: 1,080–2,450, p=0,02), berurutan. Model prediksi menggunakan dua komponen tersebut memiliki AUC 0,663, Hosmer Lemeshow Goodness-of fit test 0,48, p < 0,005. Probabilitas perawatan kembali 30 hari pada subjek dengan gangguan nutrisi dan depresi menggunakan persamaan fungsi Hazard adalah 79%. Simpulan: Status nutrisi dan status depresi memiliki hubungan signifikan dengan perawatan-kembali 30 hari. Model prediksi perawatan-kembali 30 hari yang menggunakan komponen ini memiliki tingkat diskriminasi tidak terlalu baik dengan performa yang baik, namun dapat dihitung menggunakan suatu persamaan cox proportional Hazard.
Kata kunci: Perawatan-kembali 30 hari, status nutrisi, status depresi, pengkajian paripurna pasien geriatri (P3G), skor prediksi, analisis kesintasan.


Background: 30-day unplanned readmission is one of the most important factors associated with high costs and poor outcomes for hospitalized elderly patients. This population are vulnerable for hospital admission due to aging-related characteristics which can be assessed by comprehensive geriatrics assessment (CGA). Several studies have shown that CGA components were related to 30-day readmissions in elderly patients, on the contrary, only few studies consider these components as predictive score. Objective: To develop a prediction model for 30 days unplanned readmission in elderly patients who are treated in medical ward of Cipto Mangunkusumo Hospital. Methods: A prospective observational study followed 312 subjects aged 60 years old from admission to 30 days after discharge. Demographic data and CGA components were compeleted through interviews and medical records. Bivariate followed by stepwise multivariate survival analysis was used. Then, a prediction score and a hazard functional equation were developed to predict the risk of 30 days unplanned readmission. The CGA components measured were FRAIL score (fatigue, resistance, ambulance, illness, loss of weight), Geriatric Depression Scale15 (GDS-15), Mini nutrition Assessment short form (MNA-SF), Activity Daily Living (ADL) - Barthel index, Cumulative illness rating scale-geriatric (CIRS-G), Zarits-4 item screening test, Mini Cog test, and polypharmacy. Results: Nutritional and depression status were significantly related to 30-day unplanned readmission with HR 2,368 (CI95%: 1,412–3,972, p=0,001) and HR 1,627 (CI95%: 1,080–2,450, p=0,02), respectively. Prediction model using these two components had AUC 0,663, Hosmer Lemeshow Goodness-of-fit test 0,48, p < 0.005. Probability for readmission in a patient with nutritional and depression problem on the 30 th days after discharge using functional hazard equation was 79%. Conclusion: Nutritional and depression status have significant relationship with 30-day unplanned readmision. The prediction model had moderate level of discrimination but good calibration. Also, a cox proportional hazard equation can be calculated as an alternative.
Key words: 30-day unplanned readmision, nutritional status, depression status, comprehensive geriatric assessment, predictive score, survival analysis