Causes of variation in reported in-hospital CPR survival: A critical review

The objective of this study was to determine the sources of the wide variation in reported in-hospital cardiopulmonary resuscitation (CPR) survival. English-language articles published between 1972 and 1994 pertaining to in-hospital CPR survival in adults were obtained from MEDLINE, Current Contents, and relevant reference lists. Studies were required to enroll consecutive patients, separate data for prehospital arrests from in-hospital arrests, and use survival to hospital discharge as an outcome measure. Study characteristics were noted and the strength of the research methods were judged using eight methodologic standards. Overall survival to discharge among the 68 studies that met the inclusion criteria ranged from 0 to 28.9%. Studies were performed in diverse settings using diverse research methods. Only the patient population studied was associated with reported overall survival (P = 0.008). Studies performed in general medical wards, medical intensive care units, elderly populations, or cancer patients, reported lower survival rates than studies performed in hospital-wide populations. Mean survival rates for clinical subgroups also varied widely. The heterogeneity of the studies prohibited pooling subgroup data. Studies of in-hospital CPR are more notable for their differences than for their similarities. It is therefore only possible to provide rough estimates of overall survival and survival for patients with particular clinical characteristics. Differences in overall cohort disease severity, as reflected by the specific patient population studied, explain much of this variation. However, variation caused by differences in the use of DNR orders, case definition, subgroup definition and treatment cannot be underestimated. Future studies will need to address each of these sources of heterogeneity.