Keberulangan (recurrence) insiden merupakan indikator bahwa proses pembelajaran organisasi belum sepenuhnya berhasil memutus rantai kegagalan. Penelitian ini bertujuan menganalisis hubungan antara kualitas tindakan perbaikan berbasis Hierarki Pengendalian Risiko dengan keberulangan insiden, serta peran karakteristik insiden sebagai variabel kontrol, di PT XYZ periode 2023–2025. Penelitian kuantitatif dengan desain korelasional retrospektif ini menganalisis 124 laporan investigasi insiden. Faktor kritis diklasifikasikan melalui dua skema, yaitu taksonomi internal perusahaan dan taksonomi Human Factors Analysis and Classification System (HFACS), yang ditampilkan berdampingan. Analisis dilakukan secara univariat, bivariat (Fisher's Exact Test dan Pearson Chi-Square), dan multivariat (regresi logistik ganda). Validasi reliabilitas skoring tindakan perbaikan diuji melalui inter-rater agreement (weighted Kappa = 0,722; substantial). Hasil menunjukkan tingkat keberulangan tinggi pada kedua skema (65,3% pada skema HFACS; 72,6% pada skema internal). Faktor Kritis merupakan prediktor paling dominan (OR = 7,02; p < 0,001), diikuti Klasifikasi Insiden (OR = 8,91; p = 0,003), sedangkan Kualitas Tindakan Perbaikan tidak signifikan (p = 0,556) secara konsisten pada kedua skema. Penerapan taksonomi HFACS memperinci kategori Human Error yang dominan (80,6%) menjadi Violation (38,7%) dan Error (41,1%), mengungkap keragaman jenis kegagalan yang bermanfaat bagi penyempurnaan investigasi. Penelitian menyimpulkan bahwa efektivitas pencegahan keberulangan lebih ditentukan oleh ketepatan diagnosis akar masalah pada tahap investigasi daripada level teknis tindakan perbaikan semata.

Incident recurrence indicates that organizational learning has not fully succeeded in breaking the chain of failure. This study aims to analyze the relationship between the quality of corrective actions based on the Hierarchy of Controls and incident recurrence, along with the role of incident characteristics as control variables, at PT XYZ during 2023–2025. This quantitative study with a retrospective correlational design analyzed 124 incident investigation reports. Critical factors were classified using two schemes presented side by side: the company's internal taxonomy and the Human Factors Analysis and Classification System (HFACS). Analyses were conducted at univariate, bivariate (Fisher's Exact Test and Pearson Chi-Square), and multivariate (multiple logistic regression) levels. The reliability of corrective action scoring was validated through inter-rater agreement (weighted Kappa = 0.722; substantial). Results show a high recurrence rate under both schemes (65.3% under HFACS; 72.6% under the internal scheme). Critical Factor was the most dominant predictor (OR = 7.02; p < 0.001), followed by Incident Classification (OR = 8.91; p = 0.003), while Corrective Action Quality was not significant (p = 0.556), consistently across both schemes. Applying the HFACS taxonomy refined the dominant Human Error category (80.6%) into Violation (38.7%) and Error (41.1%), revealing a diversity of failure types useful for improving investigation. The study concludes that the effectiveness of recurrence prevention is determined more by the accuracy of root-cause diagnosis during investigation than by the technical level of corrective actions alone.