Cohort follow-up using computer linkage with routinely collected data

The validation of a computerised record linkage system for matching members of a defined cohort with routinely collected national data sources is reported for the first time. The two national sources relate to mortality and inpatient data and provide contrasting characteristics in their method of collection. The linkage system produces a group of possible matches based on identifying information restricted to surname, initials, sex and date of birth which, with a clerically assisted scrutiny, gives levels of sensitivity of 66% for the mortality and 81% for the inpatient data. Specificity can be increased to 100% if conventional follow-up methods are used for the limited set of matches classified as probable by the clerical scrutiny.     

Identifying multidrug resistant tuberculosis transmission hotspots using routinely collected data

In most countries with large drug resistant tuberculosis epidemics, only those cases that are at highest risk of having MDRTB receive a drug sensitivity test (DST) at the time of diagnosis. Because of this prioritized testing, identification of MDRTB transmission hotspots in communities where TB cases do not receive DST is challenging, as any observed aggregation of MDRTB may reflect systematic differences in how testing is distributed in communities. We introduce a new disease mapping method, which estimates this missing information through probability-weighted locations, to identify geographic areas of increased risk of MDRTB transmission. We apply this method to routinely collected data from two districts in Lima, Peru over three consecutive years. This method identifies an area in the eastern part of Lima where previously untreated cases have increased risk of MDRTB. This may indicate an area of increased transmission of drug resistant disease, a finding that may otherwise have been missed by routine analysis of programmatic data. The risk of MDR among retreatment cases is also highest in these probable transmission hotspots, though a high level of MDR among retreatment cases is present throughout the study area. Identifying potential multidrug resistant tuberculosis (MDRTB) transmission hotspots may allow for targeted investigation and deployment of resources.     

Validity of routinely collected hospital admissions data on diabetes

Data from the clinical records of patients known to have diabetes and admitted to hospital in North West London, Cambridge or Newcastle were compared with data on the same admissions taken from the system responsible for recording information on all acute hospital admissions (Hospital Activity Analysis). In 89 out of 751 admissions (12%), either the sex, date of birth or marital status of the patient was incorrectly recorded. The diagnosis of diabetes was omitted in 32 (10%) of 315 admissions in which diabetes or a complication of diabetes was regarded as the principal cause of admission and in 12 (23%) out of 53 in which the principal cause was closely related to diabetes. These included cases in which the diagnosis had not been stated (by the clinician) on the discharge summary (the source document for Hospital Activity Analysis) as well as instances in which the coding clerk had failed to record the diagnosis. The validity of information collected on hospital admissions is dependent on the presentation of data by the clinician to the coding clerk. There may be a lack of awareness of the importance of the clinical discharge summary as a source document for such systems.     

The descriptive epidemiology of rheumatoid arthritis in Catalonia: a retrospective study using routinely collected data

Clinical Rheumatology - Information on the epidemiology of rheumatoid arthritis (RA) in Southern Europe is scarce. We estimated the age- and gender-adjusted incidence and prevalence of RA in...     

Estimation of myocardial infarction mortality from routinely collected data in Western Australia

The accuracy of routinely collected mortality data for ischemic heart disease (IHD) as indicators of death from acute myocardial infarction (AMI) was assessed in ages 25-64 years, according to the WHO criteria defined in 1983. Cases were identified from computer records (linked for individuals) of all death certificates and hospital discharges in Western Australia between 1971 and 1982. Where the official cause was IHD about 90% of deaths fulfilled the WHO criteria for definite or possible AMI. Up to 10% of fatal cases of definite or possible AMI were coded to other causes in the official death statistics, however it appeared that variations in this figure with changes in coding practices could cause appreciable bias in the estimation of secular trends in IHD mortality. This problem could largely be overcome by reviewing fatal events where the death certificate was coded to one of a limited number of other ICD rubrics.     

Dynamic prediction of risk of liver‐related outcomes in chronic hepatitis C using routinely collected data

Accuracy of risk assessments for clinical outcomes in patients with chronic liver disease has been limited given the nonlinear nature of disease progression. Longitudinal prediction models may more accurately capture this dynamic risk. The aim of this study was to construct accurate models of short‐ and long‐term risk of disease progression in patients with chronic hepatitis C by incorporating longitudinal clinical data. Data from the Hepatitis C Antiviral Long‐term Treatment Against Cirrhosis trial were analysed (n = 533 training cohort; n = 517 validation cohort). Outcomes included a composite liver outcome (liver‐related death, decompensation, hepatocellular carcinoma (HCC) or liver transplant), decompensation, HCC and overall mortality. Longitudinal models were constructed for risk of outcomes at 1, 3 and 5 years and compared with models using data at baseline only or baseline and a single follow‐up time point. A total of 25.1% of patients in the training and 20.8% in the validation cohort had an outcome during a median follow‐up of 6.5 years (range 0.5–9.2). The most important predictors were as follows: albumin, aspartate aminotransferase/alanine aminotransferase ratio, bilirubin, alpha‐fetoprotein and platelets. Longitudinal models outperformed baseline models with higher true‐positive rates and negative predictive values. The areas under the receiver‐operating characteristic curve for the composite longitudinal model were 0.89 (0.80–0.96), 0.83 (0.76–0.88) and 0.81 (0.75–0.87) for 1‐, 3‐, and 5‐year risk prediction, respectively. Model performance was retained for decompensation and overall mortality but not HCC. Longitudinal prediction models provide accurate risk assessments and identify patients in need of intensive monitoring and care.     

Monitoring the hepatitis C epidemic in England and evaluating intervention scale‐up using routinely collected data

In England, 160 000 individuals were estimated to be chronically infected with hepatitis C virus (HCV) in 2005 and the burden of severe HCV‐related liver disease has increased steadily for the past 15 years. Direct‐acting antiviral treatments can clear infection in most patients, motivating HCV elimination targets. However, the current burden of HCV is unknown and new methods are required to monitor progress. We employed a Bayesian back‐calculation approach, combining data on severe HCV‐related liver disease and disease progression, to reconstruct historical HCV incidence and estimate current prevalence in England. We explicitly modelled infections occurring in people who inject drugs, the key risk group, allowing information on the size of this population and surveillance data on HCV prevalence to inform recent incidence. We estimated that there were 143 000 chronic infections in 2015 (95% credible interval 123 000‐161 000), with 34% and 54% in those with recent and past injecting drug use, respectively. Following the planned scale‐up of new treatments, chronic infections were predicted to fall to 113 400 (94 900‐132 400) by the end of 2018 and to 89 500 (71 300‐108 600) by the end of 2020. Numbers developing severe HCV‐related liver disease were predicted to fall by at least 24% from 2015 to 2020. Thus, we describe a coherent framework to monitor progress using routinely collected data, which can be extended to incorporate additional data sources. Planned treatment scale‐up is likely to achieve 2020 WHO targets for HCV morbidity, but substantial efforts will be required to ensure that HCV testing and patient engagement are sufficiently high.     

Accuracy of routinely collected comorbidity data in patients undergoing colectomy: a retrospective study

Objectives

This paper aimed to determine the baseline accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of routinely collected comorbidity data in patients undergoing any types of colectomy.

Methods

All patients aged >18 who underwent right hemicolectomy, left hemicolectomy, sigmoid colectomy, subtotal colectomy, or total colectomy between 1 January 2015 and 1 November 2016 were identified. The following comorbidities were considered: hypertension, ischemic heart disease (IHD), diabetes, asthma, chronic obstructive pulmonary disease (COPD), cerebrovascular disease (CVD), chronic kidney disease (CKD), and hypercholesterolemia. The comorbidity data from clinical notes were compared with corresponding data in hospital episode statistics (HES) database in order to calculate accuracy, sensitivity, specificity, PPV, and NPV of HES codes for comorbidities. In order to assess the agreement between clinical notes and HES data, we also calculated Cohen’s kappa index value as a more robust measure of agreement.

Results

Overall, 267 patients comprising 2136 comorbidity codes were included. Overall, HES codes for comorbidities in patients undergoing colectomy had substandard accuracy 94% (kappa 0.542), sensitivity (39%), and NPV (89%). The HES codes were 100% specific with PPV of 100%. The results were consistent when individual comorbidities were analyzed separately.

Conclusions

Our results demonstrated that HES comorbidity codes in patients undergoing colectomy are specific with good positive predictive value; however, they have substandard accuracy, sensitivity, and negative predictive value. Better documentation of comorbidities in admission clerking proforma may help to improve the quality of source documents for coders, which in turn may improve the accuracy of coding.

     

Allied health in residential aged care: Using routinely collected data to improve funding opportunities

Consumers and providers have long been advocating for increased access to and delivery of allied health services in Australian residential aged care (RAC). There is significant evidence that allied health interventions are effective; however, there is limited evidence on the benefit of routine day-to-day allied health service delivery in RAC. This information is critical to effectively inform funders and policy advisors of the necessity of allied health in RAC. To improve arguments for future funding opportunities, providers, facilities and consumers need to partner together to use routinely collected, yet disparate, data, in electronic health and billing records, to improve data collection practices and evidence generation on allied health in aged care.     

Predicting diabetic foot ulceration using routinely collected data in a foot clinic. What level of prognostic accuracy can be achieved?

This study aimed to investigate the efficacy of using routinely collected clinical data in predicting the risk of diabetic foot ulcer (DFU). The first objective was to develop a prognostic model based on the most important risk factors objectively selected from a set of 39 clinical measures. The second objective was to compare the prediction accuracy of the developed model against that of a model based on only the 3 risk factors that were suggested in the systematic review and meta-analyses study (PODUS). In a cohort study, a set of 12 continuous and 27 categorical data from patients (n = 203 M/F:99/104) who attended a specialised diabetic foot clinic were collected at baseline. These patients were then followed-up for 24 months during which 24 (M/F:17/7) patients had DFU. Multivariate logistic regression was used to develop a prognostic model using the identified risk factors that achieved p < 0.2 based on univariate logistic regression. The final prognostic model included 4 risk factors (Adjusted-OR [95% CI]; p) in total. Impaired sensation (116.082 [12.06–1117.287]; p = 0.000) and presence of callus (6.257 [1.312–29.836]; p = 0.021) were significant (p < 0.05), while having dry skin (5.497 [0.866–34.89]; p = 0.071) and Onychomycosis (6.386 [0.856–47.670]; p = 0.071) that stayed in the model were not significant. The accuracy of the model with these 4 risk factors was 92.3%, where sensitivity and specificity were 78.9%, and 94.0% respectively. The 78.9% sensitivity of our prognostic 4-risk factor model was superior to the 50% sensitivity that was achieved when the three risk factors proposed by PODUS were used. Also our proposed model based on the above 4 risk factors showed to predict the DFU with higher overall prognostic accuracy. These findings have implications for developing prognostic models and clinical prediction rules in specific patient populations to more accurately predict DFU.     

A knowledge audit of the managers of primary care organizations: top priority is how to use routinely collected clinical data for quality improvement

Technology has provided improved access to the rapidly expanding evidence base and to computerized clinical data recorded as part of routine care. A knowledge audit identifies from within this mass of information the knowledge requirements of a professional group or organization, enabling implementation of an appropriately tailored knowledge-management strategy. The objective of the study is to describe perceived knowledge gaps and recommend an appropriate knowledge-management strategy for primary care. The sample comprised 18 senior managers of Primary Care Trusts: the Chairman, Chief Executive Officer, or Research and Development Lead. A series of interviews were recorded verbatim, transcribed and analysed. Knowledge requirements were broad, suggesting that a broadly based knowledge-management strategy is needed in primary care. The biggest gap in current knowledge identified is how to perform needs assessment and quality improvement using aggregated routinely collected, general practice computer data.     

Comparison of national mortality data with routinely collected health-care data to measure the occurrence of liver cirrhosis: an observational study

BackgroundIncidence of liver cirrhosis is increasing faster than the four most common diagnosed cancers (breast, colon, lung, prostate) in the UK. Mortality rates are commonly used to estimate the occurrence of cirrhosis because no other monitoring system is available for cirrhosis, but this measure has methodological limitations; there is a time lag between diagnosis and mortality, and not everyone with cirrhosis dies directly from the disease. This study compared estimates of liver cirrhosis incidence in England with mortality rates based on similar definitions of disease, over the same period.MethodsThis population-based cohort study used data from the Clinical Practice Research Datalink (CPRD) and linked English Hospital Episode Statistics (HES) to establish the number of new cases of cirrhosis between 1998 and 2009, on the basis of diagnostic codes for cirrhosis, oesophageal varices, and portal hypertension. Similar codes were used to identify deaths due to liver cirrhosis in the Office for National Statistics death registry over the same period. Mortality rates were calculated by dividing the number of deaths due to cirrhosis by mid-year population. Incidence rates were determined by dividing the number of cases by the total number of person-years from the CPRD-HES population at risk. Poisson regression was used to establish rate ratios adjusted for age and sex. Ethics approval was obtained where necessary.Findings5118 adults with an incident diagnosis of cirrhosis were identified; crude incidence increased from 24·6 to 38·4 per 100 000 person-years from 1998 to 2009. This increase fitted a continuous model with an average yearly incidence rate ratio of 1·04 (95% CI 1·03–1·05), corresponding to a 50·6% increase over the 12 years. By contrast, mortality rates for cirrhosis were 6·0 and 5·9 per 100 000 person-years in 1998 and 2009, respectively, representing a decrease of 2·5% over time.InterpretationMortality rates underestimated liver cirrhosis incidence by at least four times between 1998 and 2009. Alternative sources of routinely collected data could be used to more accurately monitor the trends in this disease. The strength of this study is the comparison of two measures of disease occurrence over the same time period; the limitation is the validity of cause of death.FundingUniversity of Nottingham and Nottingham University Hospitals NHS Trust. The funder had no role in the study.     

Social factors influencing hospital utilisation by tuberculosis patients in the Russian Federation: analysis of routinely collected data.

OBJECTIVE: To establish whether admissions, discharges and hospital utilisation for tuberculosis (TB) in Russia are independent of sex, age, disability and employment status. STUDY POPULATION AND METHODS: Analysis of hospital admissions, discharges and in-patient utilisation using routinely collected data in Samara Region of the Russian Federation. RESULTS: Male, unemployed and disabled adults were significantly more likely to be hospitalised (P < 0.001). The unemployed and pensioners were more likely to have multiple admissions. Unemployed adults were more likely to have longer average lengths of stay per admission (P < 0.001), with a cumulative length of stay for unemployed and disabled adults significantly greater than for employed adults and adults with no disability. Interruption of hospital care was significantly more frequent in male, disabled and unemployed patients (P < 0.001). CONCLUSIONS: Socio-economic factors influence hospital admission patterns and the length of stay for patients when hospitalised, as the providers of TB services attempt to mitigate the lack of social care provision for patients. For the WHO DOTS strategy to be effectively implemented and sustained in the Russian Federation health system, social sector linkage issues need to be addressed.     

Use of routinely collected past surveillance data in identifying and mapping high-risk areas in a malaria endemic area of Sri Lanka

Stratification of malaria endemic areas on eco-epidemiological criteria is an important step in planning and implementing malaria control programs. The uses of stratification of malaria endemic areas lead to better targeting of control measures such as residual insecticide spraying in countries where unstable malaria transmission occur. In this study, two methods that can be used for stratification of malaria endemic areas in Sri Lanka using routinely collected surveillance data over a period of 9 years are described. In the first method, the median Annual Parasite Incidence (API) was used as the criterion to classify an area as at risk for malaria while in the second method, the API and the Falciparum Rate (FR) were used as the criteria. Risk maps were produced by plotting the results of the analyses on maps generated by EPIMAP. The potential uses of risk maps are discussed.     

Opportunities for earlier diagnosis of type 1 diabetes in children: A case-control study using routinely collected primary care records

Background

The epidemiology of type 1 diabetes mellitus (T1DM) suggests diagnostic delays may contribute to children developing diabetic ketoacidosis at diagnosis. We sought to quantify opportunities for earlier diagnosis of T1DM in primary care.

Use of routinely collected Hospital Episode Statistics to describe alcohol-related frequent attenders to hospitals

Background

Excess alcohol consumption is a growing public health problem, causing 5·3% of deaths worldwide in those aged under 60 years. In the UK, alcohol use costs the National Health Service (NHS) £3·5 billion annually, 80% of which is used for hospital-based care. Alcohol-related hospital admissions have doubled in the past 8 years in England. With no universal way of identifying hospital admissions of alcohol-related frequent attenders (ARFAs), the true burden on the NHS is unknown including use of accident and emergency services, costs of inpatient admissions, and long-term health and social care. We aimed to identify characteristics of ARFAs to investigate whether they differ from other hospital service users and to better understand the complexity of this group.