Alfianingtyas, Marcellia Jihatul (2026) Analisis Competing Risk Durasi Rawat Inap Pasien BPJS Penyakit Tuberkulosis Menggunakan Model Cause-Specific Hazard dan Fine-Gray Subdistribution Hazard. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5006221012-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5006221012-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Tuberkulosis (TB) merupakan salah satu penyakit menular yang masih menjadi masalah kesehatan di Indonesia. Pasien TB yang menjalani rawat inap memiliki beberapa kemungkinan status akhir perawatan, yaitu pulang hidup, meninggal, dan pulang paksa, sehingga terdapat kejadian yang saling bersaing (competing risk) dalam proses analisis durasi rawat inap. Penelitian ini bertujuan untuk menganalisis durasi rawat inap pasien TB peserta BPJS Kesehatan di Pulau Jawa tahun 2024 dengan mempertimbangkan kejadian pesaing serta mengidentifikasi faktor yang memengaruhi status akhir perawatan. Data yang digunakan merupakan data pasien TB rawat inap sebanyak 2.097 pasien. Hasil penelitian menunjukkan bahwa sebagian besar pasien memiliki status pulang hidup sebanyak 1.948 pasien (92,89%) dengan median lama rawat inap selama 3 hari. Pada pemodelan awal ditemukan pelanggaran asumsi proportional hazard pada beberapa kovariat, sehingga digunakan model Extended pada pendekatan Cause-Specific Hazard dan Fine–Gray Subdistribution Hazard. Berdasarkan nilai Akaike Information Criterion (AIC), model Extended pada kedua pendekatan memberikan hasil terbaik, dengan status pulang meninggal menunjukkan kecocokan model yang lebih baik dibandingkan status pulang hidup. Pada model Extended Cause-Specific Hazard, faktor yang berpengaruh signifikan terhadap laju kejadian pulang meninggal adalah umur, tingkat keparahan, dan status kepesertaan BPJS. Sedangkan, pada model Extended Fine–Gray Subdistribution Hazard, faktor yang berpengaruh signifikan terhadap probabilitas kumulatif kejadian pulang meninggal adalah umur dan status kepesertaan BPJS. Hasil kurva prediksi Cumulative Incidence Function (CIF) menunjukkan bahwa pasien usia lanjut dan pasien dengan tingkat keparahan lebih tinggi memiliki probabilitas kumulatif kematian yang lebih besar selama masa rawat inap. Berdasarkan hasil penelitian, pendekatan competing risk lebih sesuai dibandingkan model Cox Proportional Hazard karena mampu mmpertimbangkan adanya kejadian pesaing dalam analisis durasi rawat inap pasien TB. Selain itu, model Cause-Specific Hazard dan Fine–Gray Subdistribution Hazard bersifat saling melengkapi, di mana Cause-Specific Hazard menunjukkan laju kejadian pada individu yang masih berisiko, sedangkan Fine–Gray Subdistribution Hazard mengestimasi probabilitas kumulatif kejadian pada tingkat populasi.
=====================================================================================================================================
Tuberculosis (TB) is one of the infectious diseases that remains a major public health problem in Indonesia. TB patients undergoing hospitalization may experience several possible final outcomes, namely discharge alive, death, and discharge against medical advice, resulting in competing risks in the analysis of hospital length of stay. This study aims to analyze the length of hospital stay of TB patients covered by the National Health Insurance (BPJS Kesehatan) in Java Island in 2024 by considering competing events and identifying factors influencing the final discharge status. The data consist of 2,097 hospitalized TB patients. The results show that most patients were discharged alive, accounting for 1,948 patients (92.89%), with a median length of stay of 3 days. Initial modeling indicated violations of the proportional hazard assumption for several covariates; therefore, Extended models were applied in both the Cause-Specific Hazard and Fine–Gray Subdistribution Hazard approaches. Based on the Akaike Information Criterion (AIC), the Extended models provided the best fit, with the death outcome showing better model adequacy compared to the discharge alive outcome. In the Extended Cause-Specific Hazard model, significant factors affecting the hazard of death were age, disease severity, and BPJS membership status. Meanwhile, in the Extended Fine–Gray Subdistribution Hazard model, age and BPJS membership status were significant factors influencing the cumulative incidence of death. The Cumulative Incidence Function (CIF) curves indicate that elderly patients and those with more severe conditions have a higher cumulative probability of death during hospitalization. Overall, the competing risk approach is more appropriate than the Cox Proportional Hazard model because it accounts for competing events in the analysis of hospital length of stay among TB patients. Furthermore, the Cause-Specific Hazard and Fine–Gray Subdistribution Hazard models are complementary, where the Cause-Specific Hazard analyzes the hazard among individuals still at risk, while the Fine–Gray Subdistribution Hazard estimates the cumulative incidence at the population level.
Actions (login required)
 |
View Item |