P-003 - Severity risk assessment of opioid-induced respiratory depression or overdose in patients with cancer

Abstract Title: Severity risk assessment of opioid-induced respiratory depression or overdose in patients with cancer


Background: In 2021, 80,411 individuals in the United States succumbed to overdoses involving opioids, constituting 75% of all drug overdose fatalities that year, with the majority attributed to the emergence of new synthetic opioids. These overdoses typically result from opioid-induced respiratory depression (OIRD), initiated by agonist binding of mu-receptors in the brainstem respiratory networks. The Risk Index for Overdose or Serious Opioid-induced Respiratory Depression (RIOSORD) was initially developed within the Veterans Health Administration (VHA) population and subsequently refined and validated using claims data from a commercially insured health plan (CIP). The CIP-RIOSORD, comprising a 16-item questionnaire, categorizes patients into risk classes based on their total score, offering a probability estimate for an OIRD event within the subsequent 6 months. Patients are classified into CIP-RIOSORD risk classes ranging from 1 to 7, with higher values indicating increased OIRD/overdose risk. The strengths of the CIP-RIOSORD are in providing a patient-specific, evidence-based, quantitative prediction of future OIRD events, potentially informing preventive interventions and adjustments to opioid prescribing practices in high-risk populations. However, owing to the limited number of published studies utilizing the CIP-RIOSORD across diverse patient populations, its clinical utility remains relatively low.

Purpose/Objectives: A review of previously published clinical prediction models for opioid-induced respiratory depression (OIRD) or overdose concluded that external validation of such models was essential yet infrequently undertaken. The robustness of clinical prediction models is strengthened by application across diverse patient populations and data sources. Recognizing the lack of studies examining the CIP-RIOSORD in patients with cancer, we sought to address this gap. The study purpose was to evaluate the CIP-RIOSORD in patients with cancer receiving care at an academic outpatient palliative care consultation service. The study objectives were to establish baseline CIP-RIOSORD scores and identify the predominant predictive factors associated with OIRD or overdose events in this patient cohort. We hypothesized that our sample population would exhibit higher CIP-RIOSORD scores and variable predictive factors compared to the original patient data utilized for CIP-RIOSORD model development.


Methods: This was an IRB-exempt retrospective study collecting data from July to October 2023. Patients, 18 years of age or older, diagnosed with cancer, were referred by oncologists for various reasons with symptom management being the most common. To address potential confounding factors for patients already seen by our palliative care team whereby drug therapy was initiated or changed, data were divided. Group 1 comprised new consultations prior to drug therapy interventions by our palliative care team. Group 2 included the first documented follow-up consultation during the data collection period. Group 2 patients were on drug therapies that were changed by our palliative care team.



Select CIP-RIOSORD predictive factors were collected from electronic health records (EPIC) from the preceding 6 months of the palliative care visit. Patients were stratified into a CIP-RIOSORD risk class based on the total score. Sheikh et al. further classified CIP-RIOSORD risk classes as low (risk classes 1 and 2), medium (risk classes 3 and 4) and high (risk classes 5, 6, and 7). Statistical analysis included Wilcoxon-Mann Whitney, Chi-squared, and Fisher's exact tests. Data were managed in Excel™ and were analyzed by SAS version 9.3.


Results: There were 234 palliative care consultations from July 1, 2023 to October 31, 2023. One-hundred and thirty-four of the 234 consultations met the inclusion criteria. There were 63 (47%) new consultations (Group 1) and 71 (53%) first documented follow-up consultations (Group 2). Caucasian (n = 65, 49%) women (n = 85, 63%) with gastrointestinal cancer (n = 43, 32%) were the most common demographic characteristics. Biological sex, self-reported ethnicity, self-reported race, and presence of metastatic disease were similar between groups (p > 0.05).



For Group 1, the most common CIP-RIOSORD predictive factors were antidepressant (n = 28, 44%), benzodiazepine (n = 12, 19%), and long-acting opioid use (n = 12, 19%). For Group 2 (n = 71), the most common CIP-RIOSORD predictive factors were long-acting opioid use (n = 35, 49%), antidepressant use (n = 27, 38%), and daily oral morphine equivalent (OME) > 100 (n = 27, 38%). Long-acting opioid use and daily OME > 100 were statistically significant between groups.



The overall CIP-RIOSORD score for Group 1 patients was 14.8 ± 15.5 (mean ± SD), representing a risk class 4. From the Sheikh et al risk class criteria, there were 21 (low risk), 25 (medium), and 17 (high) patients in Group 1. The overall CIP-RIOSORD score for Group 2 patients was 20.4 ± 16.5, representing a risk class 5. There was a statistically significant difference in overall CIP-RIOSORD scores between groups (p < 0.05). From the Sheikh et al risk class criteria, there were 15 (low risk), 24 (medium), and 32 (high) patients in Group 2.

Conclusions/Implications for future research and/or clinical care: This study evaluated the CIP-RIOSORD in patients with cancer receiving outpatient palliative care. The evaluation of the CIP-RIOSORD in a different patient population is needed for external validation before integration into clinical practice. Group 1, comprising new consultations, exhibited a baseline CIP-RIOSORD score indicating a 15% probability of a serious opioid-related event within six months with the majority of patients classified as moderate risk. In contrast, the majority of patients in Group 2 were classified as high risk, with a 30% probability of a serious opioid-related event within six months. The most common predictive factors between groups were antidepressant and benzodiazepine use, with long-acting opioid use and daily OME > 100 being significantly different between groups. These results provide evidence supporting the use of the CIP-RIOSORD in a clinical setting as a risk assessment tool for cancer patient populations who require opioids.


References: 1. Centers for Disease Control and Prevention, National Center for Health Statistics. Drug overdose death rates. (Accessed 2024 May 6) [Available from: https://nida.nih.gov/research-topics/trends-statistics/overdose-death-rates).

2. Zedler B, Xie L, Wang L, Joyce A, Vick C, Brigham J, et al. Development of a Risk Index for Serious Prescription Opioid-Induced Respiratory Depression or Overdose in Veterans' Health Administration Patients. Pain Med. 2015;16(8):1566-79.

3. Zedler BK, Saunders WB, Joyce AR, Vick CC, Murrelle EL. Validation of a Screening Risk Index for Serious Prescription Opioid-Induced Respiratory Depression or Overdose in a US Commercial Health Plan Claims Database. Pain Med. 2018;19(1):68-78.

4. Tseregounis IE, Henry SG. Assessing opioid overdose risk: a review of clinical prediction models utilizing patient-level data. Transl Res. 2021;234:74-87.

5. Sheikh S, Fernandez R, Smotherman C, Brailsford J, Langaee T, Velasquez E, et al. A pilot study to identify pharmacogenomic and clinical risk factors associated with opioid related falls and adverse effects in older adults. Clin Transl Sci. 2023:16(11):2331-44.