In October 2014, Public Health England (PHE) identified cases of Shiga toxin-producing Escherichia coli (STEC) serogroup O157 sharing a multiple locus variable-number tandem repeat analysis (MLVA) profile. We conducted a case-control study using multivariable logistic regression to calculate adjusted odds ratios (aOR) and 95% confidence intervals (CI) testing a range of exposures. Cases were defined as laboratory-confirmed STEC O157 with the implicated MLVA profile, were UK residents aged ⩾18 years with symptom onset between 25 September and 30 October 2014, and had no history of travel abroad within 5 days of symptom onset. One hundred and two cases were identified. Cases were mostly female (65%; median age 49, range 2-92 years). It was the second largest outbreak seen in England, to date, and a case-control study was conducted using market research panel controls and online survey methods. These methods were instrumental in the rapid data collection and analysis necessary to allow traceback investigations for short shelf-life products. This is a new method of control recruitment and this is the first in which it was a standalone recruitment method. The case-control study suggested a strong association between consumption of a ready-to-eat food and disease (aOR 28, 95% CI 5·0-157) from one retailer. No reactive microbiological testing of food items during the outbreak was possible due to the short shelf-life of the product. Collaboration with industrial bodies is needed to ensure timely traceback exercises to identify contamination events and initiate appropriate and focused microbiological testing and implement control measures.

Improving access to tuberculosis (TB) care and ensuring early diagnosis are two major aims of the WHO End TB strategy and the Collaborative TB Strategy for England. This study describes risk factors associated with diagnostic delay among TB cases in England. We conducted a retrospective cohort study of TB cases notified to the Enhanced TB Surveillance System in England between 2012 and 2015. Diagnostic delay was defined as more than 4 months between symptom onset and treatment start date. Multivariable logistic regression was used to identify demographic and clinical factors associated with diagnostic delay. Between 2012 and 2015, 22 422 TB cases were notified in England and included in the study. A third (7612) of TB cases had a diagnostic delay of more than 4 months. Being female, aged 45 years and older, residing outside of London and having extra-pulmonary TB disease were significantly associated with a diagnostic delay in the multivariable model (aOR = 1.2, 1.2, 1.2, 1.3, 1.8, respectively). This study identifies demographic and clinical factors associated with diagnostic delay, which will inform targeted interventions to improve access to care and early diagnosis among these groups, with the ultimate aim of helping reduce transmission and improve treatment outcomes for TB cases in England.

Established methods of recruiting population controls for case–control studies to investigate gastrointestinal disease outbreaks can be time consuming, resulting in delays in identifying the source or vehicle of infection. After an initial evaluation of using online market research panel members as controls in a case–control study to investigate a Salmonella outbreak in 2013, this method was applied in four further studies in the UK between 2014 and 2016. We used data from all five studies and interviews with members of each outbreak control team and market research panel provider to review operational issues, evaluate risk of bias in this approach and consider methods to reduce confounding and bias. The investigators of each outbreak reported likely time and cost savings from using market research controls. There were systematic differences between case and control groups in some studies but no evidence that conclusions on the likely source or vehicle of infection were incorrect. Potential selection biases introduced by using this sampling frame and the low response rate are unclear. Methods that might reduce confounding and some bias should be balanced with concerns for overmatching. Further evaluation of this approach using comparisons with traditional methods and population-based exposure survey data is recommended.

No abstract