Using data to drive service improvement: false dawns and a promised land?

Increasing emphasis and expectation is being placed on the role of healthcare data in addressing the problems faced by the NHS. The ideal is to replace the current fragmented system of individual systems and registries with a universal, integrated data system that provides frontline staff with what they need while also allowing monitoring of services, intelligent population-based commissioning and the facilitation of quality improvement (QI) and research. With the recently published tender for the creation of a federated data platform (FDP) there is optimism that these aspirations are being addressed; however, concerns remain that the future use of healthcare data in the UK will not fulfil its potential if the current well-recognised shortcomings of existing systems and processes are not dealt with.

Surviving Burn Injury: Drivers of Length of Hospital Stay.

With a reduction in the mortality rate of burn patients, length of stay (LOS) has been increasingly adopted as an outcome measure. Some studies have attempted to identify factors that explain a burn patient's LOS. However, few have investigated the association between LOS and a patient's mental and socioeconomic status. There is anecdotal evidence for links between these factors; uncovering these will aid in better addressing the specific physical and emotional needs of burn patients and facilitate the planning of scarce hospital resources. Here, we employ machine learning (clustering) and statistical models (regression) to investigate whether segmentation by socioeconomic/mental status can improve the performance and interpretability of an upstream predictive model, relative to a unitary model. Although we found no significant difference in the unitary model's performance and the segment-specific models, the interpretation of the segment-specific models reveals a reduced impact of burn severity in LOS prediction with increasing adverse socioeconomic and mental status. Furthermore, the socioeconomic segments' models highlight an increased influence of living circumstances and source of injury on LOS. These findings suggest that in addition to ensuring that patients' physical needs are met, management of their mental status is crucial for delivering an effective care plan.

Burn service costing using a mixed model methodology.

INTRODUCTION: The escalating cost of modern healthcare is threatening the fundamental "free at the point of delivery" principle of the UK National Health Services. A new remuneration system using a fixed tariff for pre-assigned diagnostic groups caters poorly for the heterogeneity of burn injuries. This study was to develop a system for Patient Level Costing (PLC), the first steps of which were to determine the true cost of burn care at service level. METHODS: Detailed interrogation was conducted of the cost of care in our Burns & Plastic Surgery Department. Costs were determined through the amalgamation of two fundamental methodologies: (1) Top-Down Costing (from detailed budgetary analysis for the hospital) and (2) Bottom-Up Costing (detailed itemised costing of staff, equipment, drugs, consumables & maintenance). These costs were categorised & using various apportionment tools, traced to specific care areas. RESULTS: We demonstrated that the accuracy of costs derived by host organisations cannot be relied upon (our Burn Service was 62% more expensive than estimated by our host organisation), which therefore questions the accuracy of most published work on burn care costing based upon these assumptions. Using our costing model, an analysis was made of the cost of running the Department with zero activity but "open & ready for business". Costs such as drugs and consumables were thus removed. This demonstrated that despite no clinical activity, costs still remained at 90% of full occupancy cost and are thus fixed costs. CONCLUSIONS: We hope application of this new system of Patient Level Costing to burn care will avoid the threatened viability of burn services imposed by changes in remuneration, although it will inevitably be an iterative process. A fuller understanding of the true cost of healthcare, facilitates service development and planning, both at a local and national level.

Extracorporeal membrane oxygenation in severe respiratory failure resulting from burns and smoke inhalation injury.

Extracorporeal membrane oxygenation (ECMO) is one of the most frequent forms of extracorporeal life support (ECLS) and can be used as rescue therapy in patients with severe respiratory failure resulting from burns and/or smoke inhalation injury. Experience and literature on this treatment option is still very limited, consequently results are varied. We report a retrospective analysis of our experience with veno-venous (VV) ECMO in burn patients. All five patients, three male and two female (age: 28-37 years) had flame type burns and smoke inhalation injury. Their Murray scores ranged between 3.25 and 3.75, and their revised Baux scores between 62 and 102. The mean pre-ECMO conventional ventilation time was 7.4days (3-13). The mean ECMO duration was 18days (8-35). Three patients were cannulated with dual lumen, two with separate cannulae. One oxygenator had to be changed due to technical issues and two patients needed two parallel oxygenators. Four patients had renal replacement therapy. All patients needed vasoconstrictor support, antibiotics and packed red blood cells (5-62 units). Three had steroid treatment. All five patients were successfully weaned from ECMO. One patient died later from multi-organ failure in the ICU, the other four patients survived. VV-ECMO is a useful rescue intervention in patients with burns related severe respiratory failure. Patients in our institution benefit from having both burns and ECMO centres with major expertise in the field under one roof. The results from this small cohort are encouraging, although more cases are needed to draw more robust conclusions.

Mortality risk prediction in burn injury: Comparison of logistic regression with machine learning approaches.

INTRODUCTION: Predicting mortality from burn injury has traditionally employed logistic regression models. Alternative machine learning methods have been introduced in some areas of clinical prediction as the necessary software and computational facilities have become accessible. Here we compare logistic regression and machine learning predictions of mortality from burn. METHODS: An established logistic mortality model was compared to machine learning methods (artificial neural network, support vector machine, random forests and naïve Bayes) using a population-based (England & Wales) case-cohort registry. Predictive evaluation used: area under the receiver operating characteristic curve; sensitivity; specificity; positive predictive value and Youden's index. RESULTS: All methods had comparable discriminatory abilities, similar sensitivities, specificities and positive predictive values. Although some machine learning methods performed marginally better than logistic regression the differences were seldom statistically significant and clinically insubstantial. Random forests were marginally better for high positive predictive value and reasonable sensitivity. Neural networks yielded slightly better prediction overall. Logistic regression gives an optimal mix of performance and interpretability. DISCUSSION: The established logistic regression model of burn mortality performs well against more complex alternatives. Clinical prediction with a small set of strong, stable, independent predictors is unlikely to gain much from machine learning outside specialist research contexts.

A review of the international Burn Injury Database (iBID) for England and Wales: descriptive analysis of burn injuries 2003-2011.

OBJECTIVE: To describe, for the first time, distribution (by geography, age, sex) and time trends in burn injury in England and Wales over the period that the international Burn Injury Database (iBID) has been in place. SETTING: Data from the iBID for the years 2003-2011 were used for a retrospective descriptive observational study of specialised services workload and admissions in England and Wales. PARTICIPANTS: All patients who have been visited or admitted to the burn injury specialised health service of England and Wales during the time period 2003-2011. Data cleaning was performed omitting patients with incomplete records (missingness never exceeded 5%). OUTCOME MEASURES: Workload, admissions, mortality, length of stay (LOS), geographical distribution, sex differences, age differences, total burn surface area, mechanism of Injury. RESULTS: During 2003-2011, 81,181 patients attended the specialised burn service for assessment and admission in England and Wales. Of these, 57,801 were admitted to the services. Males accounted for 63% of the total workload in specialised burn injury services, and females for 37%. The median (IQR) burn surface area was 1.5% (3.5%). The most frequent reason for burn injury was scald (38%). The median (IQR) age for all the referred workload for both genders was 21 (40). The overall mortality of the admitted patients was 1.51% and the median (IQR) LOS was 1 (5) days. CONCLUSIONS: Mortality from burn injuries in England and Wales is decreasing in line with western world trends. There is an observed increase in admissions to burn services but that could be explained in various ways. These results are vital for service development and planning, as well as the development and monitoring of prevention strategies and for healthcare commissioning.

Geographical analysis of socioeconomic factors in risk of domestic burn injury in London 2007-2013.

PURPOSE: This study aims to explore the geographical distribution of burn injuries in Greater London and the association of socioeconomic factors in areas at risk. METHODS: Data on burn injury cases classified as occurring in patients' own homes in Greater London and admitted to a specialised burns service for ≥1 day during a 7-year period were obtained from the International Burn Injury Database (iBID). Age- and gender-adjusted standardised incidence ratios (SIRs) were calculated for each Lower Layer Super Output Area (LSOA) in Greater London. Bayesian methods were used to calculate relative risks as best estimates of spatially-smoothed SIRs. RESULTS: Of a total of 2911 admissions to specialised burns services in Greater London in the study period, 2100 (72.1%) cases occurred in patients' own homes. Percentage of ethnic minorities (p=0.005), Income Deprivation Affecting Children Index (p<0.001), Health Deprivation and Disability Score (p=0.031), percentage of families with 3 or more children (p=0.004) and Barriers to Housing and Services Score (p=0.001) remained independently associated with the relative risk of paediatric domestic burn injury in a multivariate linear regression model. Percentage of ethnic minorities (p<0.001), Health Deprivation and Disability Score (p<0.001) and Barriers to Housing and Services Score (p=0.036) remained independently associated with the relative risk of adult domestic burn injury in a multivariate linear regression model. CONCLUSIONS: Socioeconomic factors are associated with an increased risk of burn injury in Greater London, but may be more important in children than adults. The specific factors identified are ethnicity, poor general health, household structure, housing issues and income deprivation affecting children.

A model of British in-hospital mortality among burns patients.

INTRODUCTION: Mortality among patients treated in hospital for burn is routinely examined, but none of the many models in use in the UK was developed using nationwide data. The aim of this research was to develop a prediction model using national data, representative of the British population. METHODS: Data were gathered from the international Burns Injury Database (iBID) and included 66,611 patients from England and Wales from 2003 to 2011. Core variables were selected following systematic review of the literature, expert consultation and then supplemented with variables selected through logistic regression. Discrimination and calibration of the model were assessed using the area under the receiver operating characteristic curve and the Hosmer-Lemeshow χ2 test respectively. RESULTS: Overall mortality for the years of the study in England and Wales was 1.27%. Mortality was predicted by age (and quadratic term) total burn surface area, presence of inhalation injury, presence of existing disorders and category of injury. The model gave a discrimination area under the curve of 0.97 in both internal and external validation. The calibration of the model gave a Hosmer-Lemeshow χ2 of 11.9 (p=0.3). CONCLUSION: We have reported a strongly predictive and theoretically well-founded model of in-patient mortality using nine years of data from all burn care services in England and Wales. We recommend this model for British burn service development and for international consideration of the variables to use in developing similar models with other data sources.

Accuracy of commercial reporting systems to monitor quality of care in burns.

INTRODUCTION: Capse Healthcare Knowledge Systems (CHKS) is a global commercial organisation that operates health benchmarking programmes in the UK and internationally. In absence of a specialty-specific quality monitoring programme for burn services, CHKS has been producing comparative quality data for burn services for a number of years. The major quality indicator reported by CHKS is mortality as a Risk Adjusted Mortality Index (RAMI). The accuracy of RAMI is unknown in comparison to published burn-specific mortality prediction models. METHODS: A retrospective study design was used to collect data for patients admitted to the Adult Burn Service at University Hospital South Manchester (UHSM) between January 2006 and December 2010. Data was collected from two sources, CHKS and Manchester Burn Injury Database (MBID). The demographic and injury characteristics of survivors and non-survivors were compared and Receiver Operator Curve (ROC), equivalence and non-inferiority analyses were used to assess accuracy of RAMI in comparison to Abbreviated Burn Severity Index (ABSI), Belgian Outcome of Burn Injury (BOBI) score, Baux score (Baux) and McGwin score (McGwin). RESULTS: The accuracy of RAMI to discriminate between survivors and non-survivors (area under curve=0.79, 95% CI 0.50-0.81) was significantly inferior to that of ABSI, BOBI, Baux and McGwin scores. Equivalence and non-inferiority testing of ROC curves also showed RAMI score to be inferior to ABSI, BOBI, Baux and McGwin scores at 5% significance level. CONCLUSION: CHKS RAMI provides an inaccurate and inferior monitoring of mortality as a quality indicator in burn patients compared to burn specific mortality prediction models. This study raises concerns about the ability of commercially reported systems to accurately monitor quality indicators of relevance to burn care.

Predicting length of stay in thermal burns: a systematic review of prognostic factors.

BACKGROUND: Continued improvement in all aspects of the management of thermal injury has resulted in marked improvements in the traditionally reported outcome of mortality. This has resulted in the search for alternative parameters that can be monitored to indicate the performance of burn services. Length of stay (LOS) in hospitalised burn patients has long been considered reflective of injury-associated morbidity, cost and the quality of care, which can be monitored consistently across services. AIM: We undertook a systematic review of published literature pertaining to LOS prognostication in thermal burns to identify the relevant factors, quantify the risk associated with these factors and identify predictive prognostic models. METHODS: Electronic searches were performed on MEDLINE, CINHAL, EMBASE, Web of Science, the Cochrane collection and a general web search was performed using Google. The searches were complemented by a manual search of the contents of leading burns journals. Quality of the studies included in the review was evaluated against published standards for prognostic studies. RESULTS: Fourteen studies were included in the review after meeting the inclusion/exclusion criteria. Age and %TBSA were the strongest predictors of LOS in these studies. Other significant predictors included % full thickness burn, female gender, inhalation injury, surgery including escharotomy and the depth of burn. Nine studies reported multivariate models for predicting LOS in patients sustaining thermal injury. None of these models were validated and the goodness-of-fit statistic (R2) ranged from 0.15 to 0.75. CONCLUSION: This review has demonstrated that %TBSA and age are the best predictors of LOS in published literature. Current prognostic models do not explain a significant proportion of variation in LOS.

Burns represent a significant proportion of the total serious trauma workload in England and Wales.

Traumatic injury is the leading cause of death in the first four decades of life. However, current estimates for traumatic injury rates fail to take into account burns. The aim of this work was to estimate the contribution of burns to serious traumatic injury in England and Wales. We have determined population-based burns rates using the International Burn Injury Database (iBID, www.ibidb.org) which collects data from regional burn centres, and non-burns rate using data from the Trauma Audit and Research Network (TARN) which collects data from emergency departments (ED, www.TARN.ac.uk). Due to incomplete national coverage of TARN, non-burns rates were estimated using data from 94 EDs that contributed data to TARN. Both non-burn and burns rates were calculated nationally and for each regional burn service catchment area (n=17). Only serious injuries (≥72 h admission or death) were included. Burns rate was 4.7 and non-burns rate 82.7 per 100,000 per year nationally. Burns therefore contributed 5.4% of all serious traumatic injuries. Contribution of burns in different regional burn service catchment areas was between 1.5% and 12%. This data suggests that burns contribute significantly to the overall trauma workload, and should be carefully considered in healthcare planning and policy.

Do illness perceptions and mood predict healing time for burn wounds? a prospective, preliminary study.

UNLABELLED: Some burn wounds take longer to heal than others, but this cannot be fully explained by physical factors such as burn size and depth. Research interest has therefore focussed on the potential contribution of psychological factors, such as perception of the burn and distress, to the wound healing process. OBJECTIVES: Using the framework of Leventhal's Common-Sense Model, we investigated whether patients' perceptions of their burn wounds and distress contributed to healing time, and whether this was via the mediating role of adherence to treatment recommendations. METHOD: Seventy-two adult burn-injured outpatients completed questionnaire measures of burn perceptions (Brief Illness Perception Questionnaire), distress (Hospital Anxiety and Depression Scale), trauma symptoms (Impact of Event Scale-Revised) and appearance concerns (Derriford Appearance Scale-24). Burn characteristics, healing time and adherence data were taken from clinic notes. RESULTS: Distress, trauma symptoms and appearance concerns were positively correlated with negative burn perceptions. In regression analysis, burn perceptions added significantly to the prediction of burn healing time after age, medical factors and burn characteristics had been controlled for. Adherence measures were not significantly correlated with burns perceptions. CONCLUSIONS: Our findings suggest that patients' perceptions of their burns contribute to healing time. Further research on the mechanisms of this association is warranted.

Three-dimensional imaging of human skin and mucosa by two-photon laser scanning microscopy.

BACKGROUND: Various structural components of human skin biopsy specimens are difficult to visualize using conventional histologic approaches. METHODS: We used two-photon microscopy and advanced imaging software to render three-dimensional (3D) images of in situ nerves, blood vessels, and hair follicles labeled with various fluorescent markers. Archived frozen human skin biopsy specimens were cryosectioned up to 150 micro m in thickness and fluorescently stained with rhodamine- or fluorescein-labeled antibodies or lectins. Optical sections were collected by two-photon microscopy and the resulting data sets were analyzed in three dimensions using Voxx software. RESULTS: Reconstructed image volumes demonstrated the complex 3D morphology of nerves, blood vessels and adnexal structures in normal mucocutaneous tissue. CONCLUSION: Two-photon microscopy and Voxx rendering software allow for detailed 3D visualization of structures within human mucocutaneous biopsy specimens, as they appear in situ, and facilitate objective interpretation of variations in their morphology. These techniques may be used to investigate disorders involving cutaneous structures that are difficult to visualize by means of traditional microscopy.