HPE HPE 85 – Spring 2017 - Page 21

Meeting report Issue 85 | Spring 2017 Safety Coordinator, Nationwide Children’s Hospital, Ohio, US). In July 2013 her hospital embarked on the development of the “You matter” campaign – a multidisciplinary initiative to support second victims. A four-hour training programme for peer supporters has been developed. This includes teaching of basic peer support skills by clinical psychologists. Training programmes always start with people describing their own experiences of incidents – “Some people have shared things that they have not talked about for 20 years”, said Dr Merandi. More than 500 peer supporters have now been trained and more than 400 encounter forms have been completed, the majority of which come from the emergency room (ER) and intensive care unit (ICU). Trained peer supporters wear green ‘You matter’ badges, she added. One unexpected finding was that interpreters often became second victims; one had explained that it could be extremely traumatic to deliver bad news in a foreign language to a patient or relative. Speakers agreed that awareness of the phenomenon of second victimhood and the existence of a non-punitive culture were important ingredients for setting up support schemes for second victims. Informatics and precision medicine ‘Big data’ can be harnessed to guide and inform precision medicine, according to Russ Altman (Professor of Bioengineering, Genetics, Medicine and Biomedical Data Science, Stanford University, California, US). He described three landmark projects in this field. Pharmacogenetics A patient’s genetic status is “knowable in advance” and there is now sufficient evidence to make pharmacogenomics a useful tool in routine practice. The advent of rapid, cheaper, complete genome sequencing combined with user-friendly clinical guidelines, such as the pharmacogenomics implementation consortium (CPIC) guidelines, makes this possible, said Professor Altman. CPIC guidelines for specific drugs have been designed to make the research literature accessible and usable for practitioners rather than researchers. The CPIC guidelines have been developed from the research-oriented pharmacogenomics knowledge base (PharmGKB) database that was produced in Professor Altman’s laboratory. A key feature of the database is pathway diagrams that show how every drug listed is metabolised. CPIC guidelines are now available for warfarin, tricyclic antidepressants, codeine and many others. However, physicians do not have the time or the knowledge to use genetic information (about medicines) routinely – ideally pharmacists should be selecting and prescribing the most suitable drugs and doses for individual patients, he said. “Almost everyone has something that makes genome sequencing worthwhile,” said Professor Altman. For example, a colleague had turned out to be heterozygous for a null mutation in CYP2C19 – an enzyme that is critical for metabolism of proton pump inhibitors, source of information about hitherto unreported drug interactions. FDA releases all adverse reaction reports, but these reports are generally considered to be “very noisy data” and difficult to interpret, said Professor Altman. One researcher analysed reports for patterns of events that predicted the likelihood of a drug altering blood glucose. In this way he was able to identify drugs that are known to alter blood glucose as a side effect – and this served as validation for the method. He then examined pairs of drugs, neither of which appeared to affect blood glucose alone, but when taken together, gave the same signal as known glucose-altering drugs. Many combinations were too rare “Second victims have been defined as those who suffer emotionally when the care that they provide leads to patient harm” anti-epileptics, clopidogrel and citalopram. When, in future, drug treatment is required, this information could be useful in selecting the most appropriate agents, he suggested. Unexpected drug interactions Existing databases contain a wealth of information that can be searched using appropriate techniques. In one such project, the 20 million abstracts in PubMed were searched to identify the 170,000 abstracts that contained a sentence that included a drug, a gene and an effect. Researchers argued that if a gene is involved in the metabolism of a number of drugs, then drug–drug interactions might occur as a result of competition for pathways. This hypothesis was tested by sifting the data to find drug combinations that were known to interact. Next, other combinations were explored. One of the predicted drug–drug interactions was metoprolol and dextromethorphan, both of which are metabolised by CYP2D6. The discovery of a published case study of a woman given the two drugs in hospital who suffered from severe side effects from both drugs confirmed the existence of the interaction, said Professor Altman. Using records and FDA databases to discover drug interactions The database of adverse reactions reported to the Food and Drug Administration (FDA) could be another to be of interest but the combination of pravastatin and paroxetine seemed likely to occur in practice. It was estimated that there could be 0.5–1.0 million Americans taking both drugs. In order to demonstrate an effect, it was necessary find patients who were taking one of these drugs, had had a glucose measurement and then were given the other drug and had a further glucose measurement within a 40-day time frame. Only 11 patients could be found at Stanford but colleagues at Vanderbilt and Harvard supplied more than 100 additional cases. The pooled analysis showed an average increase in blood glucose levels of 16mg/dl. “This is not a class effect – it is specific to the pravastatin-paroxetine combination”, emphasised Professor Altman. The initial study excluded diabetics because it was reasoned they would have detected and corrected the rise in glucose as part of their routine monitoring. However, when diabetics were analysed, the average rise in blood glucose on the combination was 60mg/dl – “crazily high”, said Professor Altman. The effect was confirmed experimentally in a mouse model – the mice were fed on butter and Sprite. In a further development of this work, the researcher examined Internet search logs (Bing) to see whether patients were experiencing symptoms and searching for information. He found that patients searching for information on the two drugs and phrases such as “peeing a lot” caused a ten-fold ‘bump’ above the hospitalpharmacyeurope.com 19