Macarthur C, Pless I B
Department of Community Health Sciences, University of Calgary, Alberta, Canada.
Am J Epidemiol. 1999 Mar 15;149(6):586-92. doi: 10.1093/oxfordjournals.aje.a009856.
The sensitivity, positive predictive value, and representativeness of the Canadian Hospitals Injury Reporting and Prevention Program (CHIRPP) were assessed. Sensitivity was estimated at four centers in June through August 1992, by matching independently identified injuries with those in the CHIRPP database. The positive predictive value was determined by reviewing all "injuries" in the database (at Montreal Children's Hospital) that could not be matched. Representativeness was assessed by comparing missed with captured injuries (at Montreal Children's Hospital) on demographic, social, and clinical factors. Sensitivity ranged from 30% to 91%, and the positive predictive value was 99.9% (i.e., the frequency of false-positive capture was negligible). The representativeness study compared 277 missed injuries with 2,746 captured injuries. The groups were similar on age, sex, socioeconomic status, delay before presentation, month, and day of presentation. Injuries resulting in admissions, poisonings, and those presenting overnight were, however, more likely to be missed. The adjusted odds ratio of being missed by CHIRPP for admitted injuries (compared with those treated and released) was 13.07 (95% confidence interval 7.82-21.82); for poisonings (compared with all other injuries), it was 9.91 (95% confidence interval 5.39-18.20); and for injuries presenting overnight (compared with those presenting during the day or evening), it was 4.11 (95% confidence interval 3.11-5.44). These injuries were probably missed because of inadequate education of participants in the system. The authors conclude that CHIRPP data are of relatively high quality and may be used, with caution, for research and public health policy.
对加拿大医院伤害报告与预防项目（CHIRPP）的敏感性、阳性预测值和代表性进行了评估。1992年6月至8月期间，通过将独立确认的伤害与CHIRPP数据库中的伤害进行匹配，在四个中心对敏感性进行了估计。通过审查数据库（蒙特利尔儿童医院）中所有无法匹配的“伤害”来确定阳性预测值。通过比较（蒙特利尔儿童医院）未被记录与已记录伤害在人口统计学、社会和临床因素方面的情况来评估代表性。敏感性范围为30%至91%，阳性预测值为99.9%（即假阳性记录的频率可忽略不计）。代表性研究比较了277例未被记录的伤害与2746例已记录的伤害。两组在年龄、性别、社会经济地位、就诊前延迟时间、就诊月份和日期方面相似。然而，导致住院的伤害、中毒以及夜间就诊的伤害更有可能被遗漏。CHIRPP遗漏住院伤害（与治疗后出院的伤害相比）的调整比值比为13.07（95%置信区间7.82 - 21.82）；中毒（与所有其他伤害相比）的调整比值比为9.91（95%置信区间5.39 - 18.20）；夜间就诊伤害（与白天或晚上就诊的伤害相比）的调整比值比为4.11（95%置信区间3.11 - 5.44）。这些伤害可能由于系统参与者教育不足而被遗漏。作者得出结论，CHIRPP数据质量相对较高，可谨慎用于研究和公共卫生政策。
