Potential for erroneous interpretation of poisoning outcomes due to changes in National Poison Data System reporting

Abstract

Background/Objective. In 2006, the annual report of poison centers in the United States changed the method of reporting profiles for generic substance categories from all exposures to single-substance exposures only. The objective of this study is to describe the potential impact of this reporting change on longitudinal analysis of outcomes. Methods. Generic substance categories with data available for all years of the study were manually extracted from Table 22 of the National Poison Data System (NPDS) annual reports for 2002–2007. For each generic substance category, the following data were extracted for each of the 6 years: total number of substance mentions (2002–2005) or single-substance exposures (2006–2007), reason (unintentional or intentional), pediatric exposures (children age <6 years), and outcomes of major effect and death. Data were compared using descriptive analysis (Wilcoxon signed-rank test) and negative binomial regression. Results. There were 65 generic substance categories (30 drug categories and 35 nondrug categories) that had data in all study years. For drug categories the average annual number of reported deaths by substance category decreased by 80.8%, from 2,229 in year 2002–2005 to 428 after the 2006 reporting change (p < 0.0001). The average annual number of reported major outcomes by substance category dropped by 76.0% (p < 0.0001). The impact on nondrug categories was similar: the annual average number of deaths and major effects by substance category decreased by about 50% from 394 and 4,639 per year during 2002–2005 to 198 deaths (p < 0.0001) and 2,357 major effects (p ≤ 0.0001) during 2006–2007. After controlling for potential covariates, multivariate regression showed that there were significant decreases in average rates of reported deaths (61.7 and 35.9%) and major effects (36.3 and 11.2%) for drug categories and nondrug categories, respectively (p < 0.01 for all). Conclusions. Overall rates of major outcomes and deaths reported to poison control centers from 2002 to 2007 have remained constant. The new method of describing demographic data in Table 22 results in outcomes that are different from those reported in previous NPDS annual reports. Comparing NPDS generic substance outcome data before and after the reporting change in 2006 will yield inaccurate results if the change in reporting methodology is not taken into account.

Keywords:

• Poison control centers
• Toxicology
• Reproducibility of results
• Fatality
• Epidemiology