Background
Adverse drug reactions (ADRs) are an important risk to patient safety in a range of health care settings, including the interface between primary and secondary care. Although, the previous systematic reviews have examined the prevalence of ADRs causing hospital admissions, limited attention has been paid to the influence of different ADR definitions, detection methods, and population groups on the reported prevalence rates.[1]

Method
Studies were identified through electronic searches (CINAHL, EMBASE, and MEDLINE) from their respective inception dates of the databases to November 2005. This was supplemented by searching the reference lists of all retrieved studies and review articles. Studies were included if they were prospective observational studies, and used the World Health Organization (WHO) ADR definition.[2] There were no language restrictions. Two authors reviewed the studies independently. A random-effect meta-analysis was used to determine the overall pooled prevalence across the studies with 95% CIs. Sensitivity analysis examined the influence of patient age groups and the methods of ADR detection on the prevalence rates. All statistics were performed using the RevMan version 4.2 statistical software package.

Results
In all, 24 studies were identified, including covering 66 961 patients admitted to hospital. Of these, 2 080 admissions were associated with ADRs. The overall pooled prevalence rate across the 24 studies was 5% (95% CIs, 4% to 6%). The prevalence of ADR admissions in children, adults, and elderly populations were 5% (95% CI 4% to 6%), 6% (95% CI 5% to 8%), and 11% (95% CI 8% to 13%), respectively. ADR rates also varied depending on the methods of ADR detection employed in the different studies. By stratifying the method of ADR detection, the prevalence of ADR admissions determined by the combination methods of medical record review and spontaneous reporting or medical record review and interview were generally higher than that detected by medical record review alone. Cardiovascular drugs were most often the cause of ADRs in adults and the elderly, whilst respiratory drugs were most often associated with ADRs in children.

Conclusions
Approximately 5% of all hospital admissions are associated with ADRs. Higher rates were found in specific age group, particularly elderly patients who are likely to be receiving multiple medications to manage long term illness. Likewise, the ADR definition and method of detection can also explain much of the variation in the reported rates between different studies. The findings from this study can be used to target initiatives designed to improve patient safety at the interface between primary and secondary care.