Content area
Full Text
Chocolate has long been recognised as, and remains, a common cause of intoxication in dogs accounting for 25 per cent of acute presentations for intoxication. 1 2 Case numbers in Europe and the UK have been reviewed, mostly based on reports to poisons centres. 3-6 Chocolate toxicity results from the methylxanthine theobromine present in cocoa bean products, causing gastrointestinal (eg, vomiting), cardiovascular (eg, tachycardia) and central nervous (eg, agitation and seizure) signs. 7 Chocolate intoxication is mostly seen in dogs 6 8 and theobromine dose calculations based on the source of chocolate are well documented 7 9 and available online. 10 11 The current study reviews cases of chocolate exposure presented to a large sentinel network of UK veterinary practices between 2012 and 2017.
Electronic health records were collected from 229 UK veterinary practices (500 premises) by the Small Animal Veterinary Surveillance Network (SAVSNET). 12 Records included the consultation time, species, breed, sex and clinical free text (narrative) in which inadvertent personal identifying data contained in narratives had been redacted using deidentification software (Newman). Narratives were screened using a regular expression 13 to identify the presence of the word 'chocolate', including a range of misspellings and contractions (eg, 'choc', 'choclat'). Cases were tagged for study inclusion if on reading they matched a definition of potential chocolate exposure whereby ingestion triggered either specific treatment or a plan of monitoring for clinical signs of theobromine toxicity.
Data were wrangled using the Pandas library 14 in Python V.3.4 15 and analysed using Stata V.12. 16 Five unmatched controls were randomly selected for each case using animals not identified by the initial screening for chocolate. To analyse risk periods, a categorical variable was generated to indicate whether a consultation occurred from one week before to two weeks after Christmas, Easter, Valentine's Day or Halloween. Univariable analysis of the effect of sex-neutering status, age category (0 to <4 years, 4 to <8 years, 8 years and over) and risk period was undertaken. Variables that showed a significant effect were included in a multivariable logistic regression model. The effect of breed was analysed using Fisher's exact tests for each breed separately (to accommodate small numbers of...