LAUSR

We thank Dr Kargar and colleagues for their thoughtful comments and suggestions with regard to our recent article assessing the rate and characteristics of adverse drug reaction (ADR)-related hospitalizations in elderly Australians [1, 2]. As outlined in their letter, there are a number of challenges involved in the assessment of ADRs associated with hospitalization, particularly with respect to assessing causality and the contribution made by the ADR to the admission itself. We agree that our study methodology had both strengths and limitations. We would contend, however, that our methodology was designed to minimise these limitations. Patients with dementia were included when family members or caregivers were present during the study period and accurate medication histories could be ascertained through interviewing the family members/caregivers, along with a detailed assessment of medical records. However, time restraints and the use of convenience sampling did not allow us to recruit all patients admitted to the participating hospitals, including some patients who were deaf and blind, were unable to give informed consent to participate, or whose family members/caregivers were unavailable to be interviewed. We agree that the Naranjo algorithm assigns a score of 2, and therefore a status of ‘possible’, to ADRs for which there is a positive response to the question, ‘‘Did the adverse events appear after the suspected drug was given?’’ [3]. Some of the suspected ADRs (doubtful cases) observed in our study mimicked the patient’s disease symptoms. Withholding or stopping the suspected drug did not improve the suspected ADR, and so these cases were scored 0 or –1 to this question. Additionally, in these doubtful cases, the answers to other questions in the Naranjo algorithm were ‘‘don’t know’’ or ‘‘no’’. It is widely acknowledged that the Naranjo algorithm has limitations; however, we have adhered to its originally developed purpose—to guide consensus between clinical reviewers in estimating the probability that an event following the administration of a therapeutic dose of drug was an ADR [4]. The classification of doubtful cases observed in our study was based on clinical judgment and consensus between two experienced clinical pharmacists. These scoring rules resulted in 17 doubtful cases that were not classified as ADR-related admissions [5]. We did not assess the inter-rater agreement between the clinical pharmacists; rather, any initial differences were discussed and a consensus decision was always reached. Only ADRs that potentially caused or contributed to acute medical admissions were considered for our study. In some cases, there were multiple causes of admission; that is, the same patient may have had both ADR-related and non-drug-related causes for their admission, hence the use of the word ‘contribution’. The cases where ‘other causes’ This reply refers to the article available at https://doi.org/10.1007/ s40264-017-0612-4.