Premium
Accuracy and validity of using medical claims data to identify episodes of hospitalizations in patients with COPD
Author(s) -
Monfared Amir Abbas Tahami,
LeLorier Jacques
Publication year - 2006
Publication title -
pharmacoepidemiology and drug safety
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.023
H-Index - 96
eISSN - 1099-1557
pISSN - 1053-8569
DOI - 10.1002/pds.1131
Subject(s) - medicine , copd , cohort , pulmonary disease , echo (communications protocol) , diagnosis code , predictive value , gold standard (test) , cohort study , database , emergency medicine , population , computer network , environmental health , computer science
Purpose In Quebec, MED‐ECHO database can be used to estimate inhospital length of stay (LOS) and number of hospitalizations (NOH) both accurately and reliably. However, access to MED‐ECHO database is time‐consuming. Quebec medical claims database (RAMQ) can be used as an alternative source to estimate these measures. Considering MED‐ECHO as the ‘gold standard,’ this study examined the validity of using RAMQ medical claims to estimate LOS and NOH. Methods We used a cohort of 3768 elderly patients with chronic obstructive pulmonary disease (COPD) between 1990 and 1996 and identified those with inhospital claims. Inhospital LOS was defined as the total number of days with inhospital claims. Various grace periods (1–15 days) between consecutive claims were considered for the estimation of LOS and NOH. RAMQ and MED‐ECHO databases were linked using unique patient identifiers. Estimates obtained from RAMQ data were compared to those from MED‐ECHO using various measures of central tendency and predictive error estimates. Results Overall, 32.7% of patients were hospitalized at least once during the study period based on RAMQ claims, as compared to 32.0% in MED‐ECHO ( p ‐value = 0.51). The best estimates [mean ( p ‐value)] were found to be those obtained when using a 7‐day grace period. RAMQ versus MED‐ECHO estimates were: 12.2 versus 13.5 days (< 0.001) for LOS and 3.6 versus 3.7 times (0.36) for NOH. Conclusions RAMQ medical claims can be used as a reliable source to estimate LOS and NOH, particularly when time and resources are restricted. RAMQ, however, should be used with caution since slight underestimations may occur. Copyright © 2005 John Wiley & Sons, Ltd.