Predictive Modeling for Suicide-Related Outcomes and Risk Factors among Patients with Pain Conditions: A Systematic Review

Shu Huang, Motomori O. Lewis, Yuhua Bao, Prakash Adekkanattu, Lauren Adkins, Samprit Banerjee, Jiang Bian, Walid F. Gellad, Amie Goodin, Yuan Luo, Jill A. Fairless, Theresa L. Walunas, Debbie L. Wilson, Yonghui Wu, Pengfei Yin, David W. Oslin, Jyotishman Pathak, Wei‐Hsuan Lo‐CiganicJournal of Clinical Medicine 2022 · DOI

TL;DR

This systematic review identifies 107 risk factors for suicide-related outcomes in patients with pain conditions, with depression being the most prevalent. While existing suicide prediction models (SPMs) show moderate performance, they suffer from limitations like lack of validation and reliance on structured data, highlighting a critical need for more robust and clinically applicable predictive tools.

What Problem Does This Paper Address?

Suicide remains a significant public health concern, and individuals with chronic pain conditions are at a disproportionately higher risk. The interplay between pain, mental health, and suicidal ideation is complex, making it challenging to identify those most vulnerable. While suicide prediction models (SPMs) exist, there's a lack of comprehensive understanding of their effectiveness and the specific risk factors relevant to the pain population. This paper aims to bridge this gap by systematically reviewing existing SPMs and identifying risk factors within this specific patient group, ultimately seeking to improve the prediction and prevention of suicide-related outcomes.

Methods

The authors conducted a systematic review of observational studies published between January 2000 and December 2020, focusing on identifying risk factors and evaluating suicide prediction models (SPMs) for adult patients (≥18 years) with any pain condition. They employed a rigorous search strategy across multiple databases and grey literature, adhering to PRISMA guidelines. Studies were included if they identified risk or protective factors for suicide-related outcomes or developed/evaluated SPMs. Data extraction involved categorizing risk factors and their data sources (structured, unstructured, or collected), and assessing the quality of SPM studies using the modified Quality In Prognosis Studies (QUIPS) tool.

Datasets

This paper is a systematic review and does not introduce new datasets. It synthesizes data from 87 previously published studies. The data sources within these studies varied widely, including:

  • Structured data: Clinical documentation, billing activities, EMR fields, administrative claims.
  • Unstructured data: Clinical notes in EMRs requiring NLP for extraction.
  • Collected data: Surveys, questionnaires, registries.

Key Insights and Ideas

  • Depression is King (for risk): Depression and its severity emerged as the most frequently reported risk factor (33% of studies), underscoring its critical role in suicide risk among pain patients.
  • Beyond Structured Data: A significant portion of identified risk factors (approximately 20%) require data beyond readily available structured EMR fields, necessitating the use of unstructured clinical notes or dedicated data collection. This highlights a data gap for current predictive models.
  • SPMs are Promising but Flawed: While some SPMs show decent predictive capabilities (e.g., C-statistics up to 0.84), their clinical utility is hampered by low positive predictive values (PPVs), meaning many identified "high-risk" individuals may not actually be at imminent risk.
  • Validation is Crucial and Lacking: Only two of the eight SPM studies performed external validation, a critical step to ensure model generalizability and prevent overfitting. This lack of validation is a major weakness.
  • Pain-Specific Risk Factors: Beyond general mental health factors, unique risk factors for this population include migraine/headaches, fibromyalgia, opioid use/dosage, and perceived burdensomeness.

Key Results

  • 107 Risk Factors Identified: The review identified a comprehensive list of 107 suicide-related risk factors, grouped into 27 categories.
  • SPM Performance Varies: For the 8 SPM studies, reported metrics included:
  • C-statistics (n=3): 0.67–0.84
    • Overall accuracy (n=5): 0.78–0.96
    • Sensitivity (n=2): 0.65–0.91
    • Positive Predictive Values (PPV) (n=3): 0.01–0.43
  • Risk of Bias: Four of the eight SPM studies had a moderate-to-high risk of bias, indicating potential limitations in their methodology and findings.
  • Data Source Distribution: 67% of risk factors were identifiable from structured data, 19% from unstructured data, and 15% from both.

Limitations and Future Work

The authors acknowledge two primary limitations:

  1. Language Barrier: Studies not published in English were excluded, potentially missing relevant research.
  2. Limited SPM Studies: The small number of SPM studies for pain patients, coupled with varied reporting of effect estimates, prevented pooled analyses.

Future work should focus on:

  • Developing Robust SPMs: Creating more mature and validated SPMs, potentially using composite measures of suicide-related outcomes to increase event frequency.
  • Integrating Diverse Data Sources: Leveraging unstructured data (e.g., clinical notes) through NLP to capture a broader range of risk factors.
  • Comprehensive Reporting: Standardizing the reporting of prediction metrics to allow for unbiased evaluation of SPM performance.
  • Longitudinal Designs: Prioritizing longitudinal study designs to establish temporality between predictors and outcomes.
  • Including Opioid-Related Factors: Explicitly incorporating opioid use, OUD, and overdose history as risk factors in future models.
  • Clinical Utility Assessment: Investigating the practical clinical utility and cost-effectiveness of developed SPMs.
  • Public Data Sharing: Encouraging the public availability of data and methods for future research and validation.

Summary updated

Abstract

Suicide is a leading cause of death in the US. Patients with pain conditions have higher suicidal risks. In a systematic review searching observational studies from multiple sources (e.g., MEDLINE) from 1 January 2000-12 September 2020, we evaluated existing suicide prediction models' (SPMs) performance and identified risk factors and their derived data sources among patients with pain conditions. The suicide-related outcomes included suicidal ideation, suicide attempts, suicide deaths, and suicide behaviors. Among the 87 studies included (with 8 SPM studies), 107 suicide risk factors (grouped into 27 categories) were identified. The most frequently occurring risk factor category was depression and their severity (33%). Approximately 20% of the risk factor categories would require identification from data sources beyond structured data (e.g., clinical notes). For 8 SPM studies (only 2 performing validation), the reported prediction metrics/performance varied: C-statistics (<i>n</i> = 3 studies) ranged 0.67-0.84, overall accuracy(<i>n</i> = 5): 0.78-0.96, sensitivity(<i>n</i> = 2): 0.65-0.91, and positive predictive values(<i>n</i> = 3): 0.01-0.43. Using the modified Quality in Prognosis Studies tool to assess the risk of biases, four SPM studies had moderate-to-high risk of biases. This systematic review identified a comprehensive list of risk factors that may improve predicting suicidal risks for patients with pain conditions. Future studies need to examine reasons for performance variations and SPM's clinical utility.
BibTeX
@article{huang2022predictive,
  author = {Shu Huang and Motomori O. Lewis and Yuhua Bao and Prakash Adekkanattu and Lauren Adkins and Samprit Banerjee and Jiang Bian and Walid F. Gellad and Amie Goodin and Yuan Luo and Jill A. Fairless and Theresa L. Walunas and Debbie L. Wilson and Yonghui Wu and Pengfei Yin and David W. Oslin and Jyotishman Pathak and Wei‐Hsuan Lo‐Ciganic},
  title = {{Predictive Modeling for Suicide-Related Outcomes and Risk Factors among Patients with Pain Conditions: A Systematic Review}},
  journal = {Journal of Clinical Medicine},
  year = {2022},
  doi = {10.3390/jcm11164813},
}
Figure