Faculty Bibliography

BACKGROUND: 'Failure to rescue'--death after a treatable complication--is used as a nursing sensitive quality indicator in the USA. It is associated with the size of the nursing workforce relative to patient load, for example patient to nurse ratio, although assessments of nurse sensitivity have not previously considered other staff groups. This study aims to assess the potential to derive failure to rescue and a proxy measure, based on long length of stay, from English hospital administrative data. By exploring change in coding practice over time and measuring associations between failure to rescue and factors including staffing, we assess whether two measures of failure to rescue are useful nurse sensitive indicators. DESIGN: Cross sectional observational study of routinely collected administrative data. PARTICIPANTS: Discharge data from 66,100,672 surgical admissions to 146 general acute hospital trusts in England (1997-2009). RESULTS: Median percentage of surgical admissions with at least one secondary diagnosis recorded increased from 26% in 1997/1998 to 40% in 2008/2009. Regression analyses showed that mortality based failure to rescue rates were significantly associated (P<0.05) with several hospital characteristics previously associated with quality, including staffing levels. Lower rates of failure to rescue were associated with a greater number of nurses per bed and doctors per bed in a bivariate analysis. Higher total clinically qualified staffing (doctors+nurses) per bed and a higher number of doctors relative to the number of nurses were both associated with lower mortality based failure to rescue in the fully adjusted analysis (P<0.05); however, the extended stay based measure showed the opposite relationship. CONCLUSION: Failure to rescue can be derived from English administrative data and may be a valid quality indicator. This is the first study to assess the association between failure to rescue and medical staffing. The suggestion that it is particularly sensitive to nursing is not clearly supported, nor is the suggestion that the number of patients with an extended hospital stay is a good proxy.

BACKGROUND: Serum phosphate is a known risk factor for cardiovascular events and mortality in people with chronic kidney disease (CKD), however data on the association of these outcomes with serum phosphate in the general population are scarce. We investigate this relationship in people with and without CKD in a large community-based population. METHODS: Three groups from an adult cohort of the Quality Improvement in Chronic Kidney Disease (QICKD) cluster randomised trial (ISRCTN56023731) were followed over a period of 2.5 years: people with normal renal function (N = 24,184), people with CKD stages 1-2 (N = 20,356), and people with CKD stages 3-5 (N = 13,292). We used a multilevel logistic regression model to determine the association between serum phosphate, in these groups, and a composite outcome of all-cause mortality, cardiovascular events, and advanced coronary artery disease. We adjusted for known cardiovascular risk factors. FINDINGS: Higher phosphate levels were found to correlate with increased cardiovascular risk. In people with normal renal function and CKD stages 1-2, Phosphate levels between 1.25 and 1.50 mmol/l were associated with increased cardiovascular events; odds ratio (OR) 1.36 (95% CI 1.06-1.74; p = 0.016) in people with normal renal function and OR 1.40 (95% CI 1.09-1.81; p = 0.010) in people with CKD stages 1-2. Hypophosphatemia (<0.75 mmol/l) was associated with fewer cardiovascular events in people with normal renal function; OR 0.59 (95% CI 0.36-0.97; p = 0.049). In people with CKD stages 3-5, hyperphosphatemia (>1.50 mmol/l) was associated with increased cardiovascular risk; OR 2.34 (95% CI 1.64-3.32; p<0.001). Other phosphate ranges were not found to have a significant impact on cardiovascular events in people with CKD stages 3-5. CONCLUSIONS: Serum phosphate is associated with cardiovascular events in people with and without CKD. Further research is required to determine the mechanisms underlying these associations.

BACKGROUND: Most chronic diseases are managed in primary and ambulatory care. The chronic care model (CCM) suggests a wide range of community, technological, team and patient factors contribute to effective chronic disease management. Ontologies have the capability to enable formalised linkage of heterogeneous data sources as might be found across the elements of the CCM. OBJECTIVE: To describe the evidence base for using ontologies and other semantic integration methods to support chronic disease management. METHOD: We reviewed the evidence-base for the use of ontologies and other semantic integration methods within and across the elements of the CCM. We report them using a realist review describing the context in which the mechanism was applied, and any outcome measures. RESULTS: Most evidence was descriptive with an almost complete absence of empirical research and important gaps in the evidence-base. We found some use of ontologies and semantic integration methods for community support of the medical home and for care in the community. Ubiquitous information technology (IT) and other IT tools were deployed to support self-management support, use of shared registries, health behavioural models and knowledge discovery tools to improve delivery system design. Data quality issues restricted the use of clinical data; however there was an increased use of interoperable data and health system integration. CONCLUSIONS: Ontologies and semantic integration methods are emergent with limited evidence-base for their implementation. However, they have the potential to integrate the disparate community wide data sources to provide the information necessary for effective chronic disease management.

Computerised Medical Record (CMR) data are widely used for secondary purposes such as service evaluation and epidemiological research. Data are increasingly aggregated from different medical facilities with various CMR vendors over time. It is increasingly difficult to manage the large quantity of data. Experiential learning in diabetes and chronic kidney disease (CKD) suggests simplistic processing can lead to errors. To maximise analytical ability for the Quality Improvement in CKD (QICKD) trial, we developed an agile data management process. By removing the need to import and process data in a relational database, we reduced processing and analysis time. We demonstrated usage of our new agile method to rapidly develop complex queries to identify how blood pressure varied between patients included or excluded from Quality and Outcomes Frameworks (QOF) pay-for-performance (P4P) targets in UK primary care. We describe a novel specification language that allows clinicians to focus on identifying variables to extract useful information from CMRs. Data for research questions were available in &lt;1hour instead of longer times previously required through use of an SQL database.

OBJECTIVES: To model the frequency of 'last minute' cancellations of planned elective procedures in the English NHS with respect to the patient and provider factors that led to these cancellations. METHODS: A dataset of 5,288,604 elective patients spell in the English NHS from January 1st, 2007 to December 31st, 2007 was extracted from the Hospital Episode Statistics. A binary dependent variable indicating whether or not a patient had a Health Resource Group coded as S22--'Planned elective procedure not carried out'--was modeled using a probit regression estimated via maximum likelihood including patient, case and hospital level covariates. RESULTS: Longer waiting times and being admitted on a Monday were associated with a greater rate of cancelled procedures. Male patients, patients from lower socio-economic groups and older patients had higher rates of cancelled procedures. There was significant variation in cancellation rates between hospitals; Foundation Trusts and private facilities had the lowest cancellation rates. CONCLUSIONS: Further research is needed on why Foundation Trusts exhibit lower cancellation rates. Hospitals with relatively high cancellation rates should be encouraged to tackle this problem. Further evidence is needed on whether hospitals are more likely to cancel operations where the procedure tariff is lower than the S22 tariff as this creates a perverse incentive to cancel. Understanding the underlying causes of why male, older and patients from lower socio-economic groups are more likely to have their operations cancelled is important to inform the appropriate policy response. This research suggests that interventions designed to reduce cancellation rates should be targeted to high-cancellation groups.