Historique
Article de Périodique
Validity of International Classification of Diseases codes in identifying illicit drug use target conditions using medical record data as a reference standard: A systematic review (2020)
Auteur(s) :
MCGREW, K. M. ;
HOMCO, J. B. ;
GARWE, T. ;
DAO, H. D. ;
WILLIAMS, M. B. ;
DREVETS, D. A. ;
JAFARZADEH, S. R. ;
ZHAO, Y. D. ;
CARABIN, H.
Année
2020
Page(s) :
107825
Sous-type de document :
Revue de la littérature / Literature review
Langue(s) :
Anglais
Domaine :
Drogues illicites / Illicit drugs
Discipline :
SAN (Santé publique / Public health)
Thésaurus géographique
INTERNATIONAL
Thésaurus mots-clés
CLASSIFICATION INTERNATIONALE DES MALADIES
;
CLASSIFICATION
;
PRODUIT ILLICITE
;
VALIDITE
;
MORBIDITE
;
PATHOLOGIE
;
BASE DE DONNEES
;
ABUS
;
DEPENDANCE
Résumé :
BACKGROUND: The twenty-first century opioid crisis has spurred interest in using International Classification of Diseases (ICD) code algorithms to identify patients using illicit drugs from administrative healthcare data. We conducted a systematic review of studies that validated ICD code algorithms for illicit drug use against a reference standard of medical record data.
METHODS: Systematic searches of MEDLINE, EMBASE, PsycINFO, and Web of Science were conducted for studies published between 1980 and 2018 in English, French, Italian, or Spanish. We included validation studies of ICD-9 or ICD-10 code algorithms for an illicit drug use target condition (e.g., illicit drug use, abuse, or dependence (UAD), illicit drug use-related complications) given the sensitivity or specificity was reported or could be calculated. Bias was assessed with the Quality Assessment of Diagnostic Accuracy Studies Version 2 (QUADAS-2) tool.
RESULTS: Six of the 1210 articles identified met the inclusion criteria. For validation studies of broad UAD (n = 4), the specificity was nearly perfect, but the sensitivity ranged from 47% to 83%, with higher sensitivities tending to occur in higher prevalence populations. For validation studies of injection drug use (IDU)-associated infective endocarditis (n = 2), sensitivity and specificity were poor due to the lack of an ICD code for IDU. For all six studies, the risk of bias for the QUADAS-2 "reference standard" and "flow/timing domains" was scored as "unclear" due to insufficient reporting.
CONCLUSIONS: Few studies have validated ICD code algorithms for illicit drug use target conditions, and available evidence is challenging to interpret due to inadequate reporting.
PROSPERO Registration: CRD42019118401.
Highlights:
Few validation studies of ICD code algorithms to identify illicit drug use exist.
No two studies have validated the same ICD code algorithm for illicit drug use.
Algorithms demonstrated superior sensitivity in groups at higher risk of drug use.
Inadequate reporting limits conclusions that can be drawn from existing studies.
METHODS: Systematic searches of MEDLINE, EMBASE, PsycINFO, and Web of Science were conducted for studies published between 1980 and 2018 in English, French, Italian, or Spanish. We included validation studies of ICD-9 or ICD-10 code algorithms for an illicit drug use target condition (e.g., illicit drug use, abuse, or dependence (UAD), illicit drug use-related complications) given the sensitivity or specificity was reported or could be calculated. Bias was assessed with the Quality Assessment of Diagnostic Accuracy Studies Version 2 (QUADAS-2) tool.
RESULTS: Six of the 1210 articles identified met the inclusion criteria. For validation studies of broad UAD (n = 4), the specificity was nearly perfect, but the sensitivity ranged from 47% to 83%, with higher sensitivities tending to occur in higher prevalence populations. For validation studies of injection drug use (IDU)-associated infective endocarditis (n = 2), sensitivity and specificity were poor due to the lack of an ICD code for IDU. For all six studies, the risk of bias for the QUADAS-2 "reference standard" and "flow/timing domains" was scored as "unclear" due to insufficient reporting.
CONCLUSIONS: Few studies have validated ICD code algorithms for illicit drug use target conditions, and available evidence is challenging to interpret due to inadequate reporting.
PROSPERO Registration: CRD42019118401.
Highlights:
Few validation studies of ICD code algorithms to identify illicit drug use exist.
No two studies have validated the same ICD code algorithm for illicit drug use.
Algorithms demonstrated superior sensitivity in groups at higher risk of drug use.
Inadequate reporting limits conclusions that can be drawn from existing studies.
Affiliation :
Department of Biostatistics & Epidemiology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma, USA