Antimicrobial resistance is one of the most pressing healthcare threats, and increased antibiotic usage has been associated with antimicrobial resistance at both population and individual levels. Pneumonia is one of the most common reasons why antibiotics are prescribed, and so it is potentially a major driver of resistance. However, the diagnostic criteria for pneumonia remain uncertain, and, to date, most analyses of this condition using electronic health records rely on diagnostic codes. However, diagnostic codes can be difficult to interpret, as their primary function is administrative, ensuring that hospitals get appropriately paid for the patients they have treated. Thus they have intrinsic weaknesses for epidemiological analyses.
However, there are many other types of routinely collected electronic data which could be used in conjunction with diagnostic codes – the challenge is working out exactly how we can use this kind of “big data” to identify important disease phenotypes to investigate their epidemiology. In this project, we will investigate how physiological measurements (e.g. temperature, oxygen saturation, heart rate, respiratory rate), results of routine blood tests, and features extracted from text X-ray reports using natural language processing could be used to optimise identification of pneumonia from electronic health records.
We will do this within the Oxford Biomedical Research Centre Antimicrobial Resistance and Modernising Microbiology Theme, where we have created a very large anonymised electronic research database, the Infections in Oxfordshire Research Database (IORD; https://oxfordbrc.nihr.ac.uk/research-themes-overview/antimicrobial-resistance-and-modernising-microbiology/infections-in-oxfordshire-research-database-iord/). IORD includes ~400 million patient records from 20 years and ~3.6 million individuals, and has generic research ethics approval for infection-related analyses.
Candidates should have a strong interest in statistical data analysis and pursuing a research career in clinical or non-clinical epidemiology. The student will be supervised by Sarah Walker, Professor of Medical Statistics and Epidemiology, and co-supervised by David Eyre, Robertson Fellow at the Big Data Institute and Tim Peto, Professor of Infectious Diseases. They will be based at the John Radcliffe Hospital and join a 40-strong group working to translate new molecular technologies and advances in clinical informatics into benefits for patients.
The student will be exposed to and trained in
Training will include short-courses available through the department and university and informal on-the-job training through regular supervisory meetings. The student would also have access to the courses provided by the Department for Continuing Education. The student would need to collaborate and work with a large multidisciplinary team including Infectious Disease Clinicians, Statisticians, Software Engineers, Computer Scientists Bioinformaticians and Microbiologists in the Modernising Medical Microbiology group at the John Radcliffe Hospital and at the Big Data Institute.
Project reference number: 983
|Professor (Ann) Sarah Walker||Experimental Medicine Division||Oxford University, John Radcliffe Hospital||GBRfirstname.lastname@example.org|
|Dr David Eyre||Experimental Medicine Division||Oxford University, John Radcliffe Hospital||GBRemail@example.com|
|Professor Tim Peto||Experimental Medicine Division||Oxford University, John Radcliffe Hospital||GBRfirstname.lastname@example.org|
BACKGROUND: Weekend hospital admission is associated with increased mortality, but the contributions of varying illness severity and admission time to this weekend effect remain unexplored. METHODS: We analysed unselected emergency admissions to four Oxford University National Health Service hospitals in the UK from Jan 1, 2006, to Dec 31, 2014. The primary outcome was death within 30 days of admission (in or out of hospital), analysed using Cox models measuring time from admission. The primary exposure was day of the week of admission. We adjusted for multiple confounders including demographics, comorbidities, and admission characteristics, incorporating non-linearity and interactions. Models then considered the effect of adjusting for 15 common haematology and biochemistry test results or proxies for hospital workload. FINDINGS: 257 596 individuals underwent 503 938 emergency admissions. 18 313 (4·7%) patients admitted as weekday energency admissions and 6070 (5·1%) patients admitted as weekend emergency admissions died within 30 days (p<0·0001). 9347 individuals underwent 9707 emergency admissions on public holidays. 559 (5·8%) died within 30 days (p<0·0001 vs weekday). 15 routine haematology and biochemistry test results were highly prognostic for mortality. In 271 465 (53·9%) admissions with complete data, adjustment for test results explained 33% (95% CI 21 to 70) of the excess mortality associated with emergency admission on Saturdays compared with Wednesdays, 52% (lower 95% CI 34) on Sundays, and 87% (lower 95% CI 45) on public holidays after adjustment for standard patient characteristics. Excess mortality was predominantly restricted to admissions between 1100 h and 1500 h (p=0·04). No hospital workload measure was independently associated with mortality (all p values >0·06). INTERPRETATION: Adjustment for routine test results substantially reduced excess mortality associated with emergency admission at weekends and public holidays. Adjustment for patient-level factors not available in our study might further reduce the residual excess mortality, particularly as this clustered around midday at weekends. Hospital workload was not associated with mortality. Together, these findings suggest that the weekend effect arises from patient-level differences at admission rather than reduced hospital staffing or services. FUNDING: NIHR Oxford Biomedical Research Centre. Hide abstract
BACKGROUND: Community-acquired pneumonia (CAP) is a major cause of mortality and morbidity in many countries but few recent large-scale studies have examined trends in its incidence. METHODS: Incidence of CAP leading to hospitalisation in one UK region (Oxfordshire) was calculated over calendar time using routinely collected diagnostic codes, and modelled using piecewise-linear Poisson regression. Further models considered other related diagnoses, typical administrative outcomes, and blood and microbiology test results at admission to determine whether CAP trends could be explained by changes in case-mix, coding practices or admission procedures. RESULTS: CAP increased by 4.2%/year (95% CI 3.6 to 4.8) from 1998 to 2008, and subsequently much faster at 8.8%/year (95% CI 7.8 to 9.7) from 2009 to 2014. Pneumonia-related conditions also increased significantly over this period. Length of stay and 30-day mortality decreased slightly in later years, but the proportions with abnormal neutrophils, urea and C reactive protein (CRP) did not change (p>0.2). The proportion with severely abnormal CRP (>100 mg/L) decreased slightly in later years. Trends were similar in all age groups. Streptococcus pneumoniae was the most common causative organism found; however other organisms, particularly Enterobacteriaceae, increased in incidence over the study period (p<0.001). CONCLUSIONS: Hospitalisations for CAP have been increasing rapidly in Oxfordshire, particularly since 2008. There is little evidence that this is due only to changes in pneumonia coding, an ageing population or patients with substantially less severe disease being admitted more frequently. Healthcare planning to address potential further increases in admissions and consequent antibiotic prescribing should be a priority. Hide abstract