Faculty Bibliography

BACKGROUND: Recording concerns about child maltreatment, including minor concerns, is recommended by the General Medical Council (GMC) and National Institute for Health and Clinical Excellence (NICE) but there is evidence of substantial under-recording. AIM: To determine whether a simple coding strategy improved recording of maltreatment-related concerns in electronic primary care records. DESIGN AND SETTING: Clinical audit of rates of maltreatment-related coding before January 2010-December 2011 and after January-December 2012 implementation of a simple coding strategy in 11 English family practices. The strategy included encouraging general practitioners to use, always and as a minimum, the Read code 'Child is cause for concern'. A total of 25,106 children aged 0-18 years were registered with these practices. We also undertook a qualitative service evaluation to investigate barriers to recording. METHOD: Outcomes were recording of 1) any maltreatment-related codes, 2) child protection proceedings and 3) child was a cause for concern. RESULTS: We found increased recording of any maltreatment-related code (rate ratio 1.4; 95% CI 1.1-1.6), child protection procedures (RR 1.4; 95% CI 1.1-1.6) and cause for concern (RR 2.5; 95% CI 1.8-3.4) after implementation of the coding strategy. Clinicians cited the simplicity of the coding strategy as the most important factor assisting implementation. CONCLUSION: This simple coding strategy improved clinician's recording of maltreatment-related concerns in a small sample of practices with some 'buy-in'. Further research should investigate how recording can best support the doctor-patient relationship. HOW THIS FITS IN: Recording concerns about child maltreatment, including minor concerns, is recommended by the General Medical Council (GMC) and National Institute for Health and Clinical Excellence (NICE), but there is evidence of substantial under-recording. We describe a simple clinical coding strategy that helped general practitioners to improve recording of maltreatment-related concerns. These improvements could improve case finding of children at risk and information sharing.

BACKGROUND: Autosomal dominant polycystic kidney disease (ADPKD) causes progressive renal damage and is a leading cause of end-stage renal failure. With emerging therapies it is important to devise a method for early detection. We aimed to identify factors from routine clinical data which can be used to distinguish people with a high likelihood of having ADPKD in a primary health care setting. METHOD: A cross-sectional study was undertaken using data from the Quality Intervention in Chronic Kidney Disease trial extracted from 127 primary care practices in England. The health records of 255 people with ADPKD were compared to the general population. Logistic regression was used to identify clinical features which distinguish ADPKD. These clinical features were used to stratify individual risk using a risk score tool. RESULTS: Renal impairment, proteinuria, haematuria, a diastolic blood pressure over 90 mmHg and multiple antihypertensive medications were more common in ADPKD than the general population and were used to build a regression model (area under the receiver operating characteristic curve; 0.79). Age, gender, haemoglobin and urinary tract infections were not associated with ADPKD. A risk score (range -3 to +10) of >/=0 gave a sensitivity of 70.2% and specificity 74.9% of for detection. CONCLUSIONS: Stratification of ADPKD likelihood from routine data may be possible. This approach could be a valuable component of future screening programs although further longitudinal analyses are needed.

We report trends in type 2 diabetes mellitus and obesity in adults residing in the Arabian Gulf States. Among the Saudi population, the prevalence of diabetes increased from 10.6% in 1989 to 32.1% in 2009. Prevalence of the disease increased faster among Saudi men than women, with growth rates of 0.8% and 0.6% per year, respectively.

BACKGROUND:NHS hospitals collect a wealth of administrative data covering accident and emergency (A&E) department attendances, inpatient and day case activity, and outpatient appointments. Such data are increasingly being used to compare units and services, but adjusting for risk is difficult. OBJECTIVES:To derive robust risk-adjustment models for various patient groups, including those admitted for heart failure (HF), acute myocardial infarction, colorectal and orthopaedic surgery, and outcomes adjusting for available patient factors such as comorbidity, using England’s Hospital Episode Statistics (HES) data. To assess if more sophisticated statistical methods based on machine learning such as artificial neural networks (ANNs) outperform traditional logistic regression (LR) for risk prediction. To update and assess for the NHS the Charlson index for comorbidity. To assess the usefulness of outpatient data for these models. MAIN OUTCOME MEASURES:Mortality, readmission, return to theatre, outpatient non-attendance. For HF patients we considered various readmission measures such as diagnosis-specific and total within a year. METHODS:We systematically reviewed studies comparing two or more comorbidity indices. Logistic regression, ANNs, support vector machines and random forests were compared for mortality and readmission. Models were assessed using discrimination and calibration statistics. Competing risks proportional hazards regression and various count models were used for future admissions and bed-days. RESULTS: = 0.73 for death following stroke. Calibration was often good for procedure groups but poorer for diagnosis groups, with overprediction of low risk a common cause. The machine learning methods we investigated offered little beyond LR for their greater complexity and implementation difficulties. For HF, some patient-level predictors differed by primary diagnosis of readmission but not by length of follow-up. Prior non-attendance at outpatient appointments was a useful, strong predictor of readmission. Hospital-level readmission rates for HF did not correlate with readmission rates for non-HF; hospital performance on national audit process measures largely correlated only with HF readmission rates. CONCLUSIONS:Many practical risk-prediction or casemix adjustment models can be generated from HES data using LR, though an extra step is often required for accurate calibration. Including outpatient data in readmission models is useful. The three machine learning methods we assessed added little with these data. Readmission rates for HF patients should be divided by diagnosis on readmission when used for quality improvement. FUTURE WORK:As HES data continue to develop and improve in scope and accuracy, they can be used more, for instance A&E records. The return to theatre metric appears promising and could be extended to other index procedures and specialties. While our data did not warrant the testing of a larger number of machine learning methods, databases augmented with physiological and pathology information, for example, might benefit from methods such as boosted trees. Finally, one could apply the HF readmissions analysis to other chronic conditions. FUNDING:The National Institute for Health Research Health Services and Delivery Research programme.

OBJECTIVES: To investigate the effect of providing patients online access to their electronic health record (EHR) and linked transactional services on the provision, quality and safety of healthcare. The objectives are also to identify and understand: barriers and facilitators for providing online access to their records and services for primary care workers; and their association with organisational/IT system issues. SETTING: Primary care. PARTICIPANTS: A total of 143 studies were included. 17 were experimental in design and subject to risk of bias assessment, which is reported in a separate paper. Detailed inclusion and exclusion criteria have also been published elsewhere in the protocol. PRIMARY AND SECONDARY OUTCOME MEASURES: Our primary outcome measure was change in quality or safety as a result of implementation or utilisation of online records/transactional services. RESULTS: No studies reported changes in health outcomes; though eight detected medication errors and seven reported improved uptake of preventative care. Professional concerns over privacy were reported in 14 studies. 18 studies reported concern over potential increased workload; with some showing an increase workload in email or online messaging; telephone contact remaining unchanged, and face-to face contact staying the same or falling. Owing to heterogeneity in reporting overall workload change was hard to predict. 10 studies reported how online access offered convenience, primarily for more advantaged patients, who were largely highly satisfied with the process when clinician responses were prompt. CONCLUSIONS: Patient online access and services offer increased convenience and satisfaction. However, professionals were concerned about impact on workload and risk to privacy. Studies correcting medication errors may improve patient safety. There may need to be a redesign of the business process to engage health professionals in online access and of the EHR to make it friendlier and provide equity of access to a wider group of patients. A1 SYSTEMATIC REVIEW REGISTRATION NUMBER: PROSPERO CRD42012003091.

BACKGROUND: Generally benefits and risks of vaccines can be determined from studies carried out as part of regulatory compliance, followed by surveillance of routine data; however there are some rarer and more long term events that require new methods. Big data generated by increasingly affordable personalised computing, and from pervasive computing devices is rapidly growing and low cost, high volume, cloud computing makes the processing of these data inexpensive. OBJECTIVE: To describe how big data and related analytical methods might be applied to assess the benefits and risks of vaccines. METHOD: We reviewed the literature on the use of big data to improve health, applied to generic vaccine use cases, that illustrate benefits and risks of vaccination. We defined a use case as the interaction between a user and an information system to achieve a goal. We used flu vaccination and pre-school childhood immunisation as exemplars. RESULTS: We reviewed three big data use cases relevant to assessing vaccine benefits and risks: (i) Big data processing using crowdsourcing, distributed big data processing, and predictive analytics, (ii) Data integration from heterogeneous big data sources, e.g. the increasing range of devices in the "internet of things", and (iii) Real-time monitoring for the direct monitoring of epidemics as well as vaccine effects via social media and other data sources. CONCLUSIONS: Big data raises new ethical dilemmas, though its analysis methods can bring complementary real-time capabilities for monitoring epidemics and assessing vaccine benefit-risk balance.

OBJECTIVE: To identify risk factors for falls and generate two screening tools: an opportunistic tool for use in consultation to flag at risk patients and a systematic database screening tool for comprehensive falls assessment of the practice population. STUDY DESIGN AND SETTING: This multicenter cohort study was part of the quality improvement in chronic kidney disease trial. Routine data for participants aged 65 years and above were collected from 127 general practice (GP) databases across the UK, including sociodemographic, physical, diagnostic, pharmaceutical, lifestyle factors, and records of falls or fractures over 5 years. Multilevel logistic regression analyses were performed to identify predictors. The strongest predictors were used to generate a decision tree and risk score. RESULTS: Of the 135,433 individuals included, 10,766 (8%) experienced a fall or fracture during follow-up. Age, female sex, previous fall, nocturia, anti-depressant use, and urinary incontinence were the strongest predictors from our risk profile (area under the receiver operating characteristics curve = 0.72). Medication for hypertension did not increase the falls risk. Females aged over 75 years and subjects with a previous fall were the highest risk groups from the decision tree. The risk profile was converted into a risk score (range -7 to 56). Using a cut-off of >/=9, sensitivity was 68%, and specificity was 60%. CONCLUSION: Our study developed opportunistic and systematic tools to predict falls without additional mobility assessments.