The impact of changes in coding on mortality reports using the example of sepsis.

OBJECTIVES: NHS Digital issued new guidance on sepsis coding in April 2017 which was further modified in April 2018. During these timeframes some centres reported increased sepsis associated mortality, whilst others reported reduced mortality, in some cases coincident with specific quality improvement programmes. We hypothesised that changes in reported mortality could not be separated from changes in coding practice. METHODS: Hospital Episode Statistics from the Admitted Patient Care dataset for NHS hospitals in England, from April 2016 to March 2020 were analysed. Admissions of adults with sepsis: an International Classification of Diseases 10 (ICD-10) code associated with the Agency for Healthcare Research and Quality Clinical Classifications Software class 'Septicaemia (except in labour)', were assessed. Patient comorbidities were defined by other ICD-10 codes recorded within the admission episode. RESULTS: 1,081,565 hospital episodes with a coded diagnosis of sepsis were studied. After April 2017 there was a significant increase in admission episodes with sepsis coded as the primary reason for admission. There were significant changes in the case-mix of patients with a primary diagnosis of sepsis after April 2017. An analysis of case-mix, hospital and year treated as random effects, defined a small reduction in sepsis associated mortality across England following the first change in coding guidance. No centre specific improvement in outcome could be separated from these random-effects. CONCLUSION: Changes in sepsis coding practice altered case-mix and case selection, in ways that varied between centres. This was associated with changes in centre-specific sepsis associated mortality, over time. According to the direction of change these may be interpreted either as requiring local investigation for cause or as supporting coincident changes in clinical practice. A whole system analysis showed that centre specific changes in mortality cannot be separated from system-wide changes. Caution is therefore required when interpreting sepsis outcomes in England, particularly when using single centre studies to inform or support guidance or policy.

Ethnicity and risk of death in patients hospitalised for COVID-19 infection in the UK: an observational cohort study in an urban catchment area.

BACKGROUND: Studies suggest that certain black and Asian minority ethnic groups experience poorer outcomes from COVID-19, but these studies have not provided insight into potential reasons for this. We hypothesised that outcomes would be poorer for those of South Asian ethnicity hospitalised from a confirmed SARS-CoV-2 infection, once confounding factors, health-seeking behaviours and community demographics were considered, and that this might reflect a more aggressive disease course in these patients. METHODS: Patients with confirmed SARS-CoV-2 infection requiring admission to University Hospitals Birmingham NHS Foundation Trust (UHB) in Birmingham, UK between 10 March 2020 and 17 April 2020 were included. Standardised admission ratio (SAR) and standardised mortality ratio (SMR) were calculated using observed COVID-19 admissions/deaths and 2011 census data. Adjusted HR for mortality was estimated using Cox proportional hazard model adjusting and propensity score matching. RESULTS: All patients admitted to UHB with COVID-19 during the study period were included (2217 in total). 58% were male, 69.5% were white and the majority (80.2%) had comorbidities. 18.5% were of South Asian ethnicity, and these patients were more likely to be younger and have no comorbidities, but twice the prevalence of diabetes than white patients. SAR and SMR suggested more admissions and deaths in South Asian patients than would be predicted and they were more likely to present with severe disease despite no delay in presentation since symptom onset. South Asian ethnicity was associated with an increased risk of death, both by Cox regression (HR 1.4, 95% CI 1.2 to 1.8), after adjusting for age, sex, deprivation and comorbidities, and by propensity score matching, matching for the same factors but categorising ethnicity into South Asian or not (HR 1.3, 95% CI 1.0 to 1.6). CONCLUSIONS: Those of South Asian ethnicity appear at risk of worse COVID-19 outcomes. Further studies need to establish the underlying mechanistic pathways.

Consolidated quality improvements following benchmarking with cardiothoracic surgery registries-a systematic review.

The influence of registries in medicine is large. However, there has been no systematic assessment conducted to quantify the impact of benchmarking with registries focused on cardiothoracic surgery. Numerous publications conclude that registry participation leads to improvement of outcomes for patients. A large number of registries provide evidence sub-structured by statistics that show decreases in morbidity and mortality in the participants' clinical units. Many authors praise the benchmarking method making use of databases of registries as having a positive effect on outcome of care. However, studies proving the direct causal relation between the use of cardiothoracic surgery-oriented registries and improvement of clinical in-hospital outcomes are extremely scarce. We aimed to analyse the causal relation between the use of cardiothoracic surgery-oriented registries and improvement of clinical outcomes. In a systematic literature review, publications demonstrating the use of registry data to obtain consolidated quality improvements were selected. After analysis of 2990 scientific publications, 6 studies filled the inclusion criteria. The selected studies acknowledged that benchmarking of data against registries was used for a focused and methodologically organized improvement in cardiothoracic departments. In conjunction with the impact of the applied methods on healthcare, their results demonstrate quantifiable enhanced local outcomes over time.

Identifying cardiac surgery operations in hospital episode statistics administrative database, with an OPCS-based classification of procedures, validated against clinical data.

OBJECTIVES: Administrative databases with dedicated coding systems in healthcare systems where providers are funded based on services recorded have been shown to be useful for clinical research, although their reliability is still questioned. We devised a custom classification of procedures and algorithms based on OPCS, enabling us to identify open heart surgeries from the English administrative database, Hospital Episode Statistics, with the objective of comparing the incidence of cardiac procedures in administrative and clinical databases. DESIGN: A comparative study of the incidence of cardiac procedures in administrative and clinical databases. SETTING: Data from all National Health Service Trusts in England, performing cardiac surgery. PARTICIPANTS: Patients classified as having cardiac surgery across England between 2004 and 2015, using a combination of procedure codes, age >18 and consultant specialty, where the classification was validated against internal and external benchmarks. RESULTS: We identified a total of 296 426 cardiac surgery procedures, of which majority of the procedures were coronary artery bypass grafting (CABG), aortic valve replacement (AVR), mitral repair and aortic surgery. The matching at local level was 100% for CABG and transplant, >90% for aortic valve and major aortic procedures and >80% for mitral. At national level, results were similar for CABG (IQR 98.6%-104%), AVR (IQR 105%-118%) and mitral valve replacement (IQR 86.2%-111%). CONCLUSIONS: We set up a process which can identify cardiac surgeries in England from administrative data. This will lead to the development of a risk model to predict early and late postoperative mortality, useful for risk stratification, risk prediction, benchmarking and real-time monitoring. Once appropriately adjusted, the system can be applied to other specialties, proving especially useful in those areas where clinical databases are not fully established.

Randomized comparison of the clinical outcome of single versus multiple arterial grafts: the ROMA trial-rationale and study protocol.

SUMMARY: The primary hypothesis of the ROMA trial is that in patients undergoing primary isolated non-emergent coronary artery bypass grafting, the use of 2 or more arterial grafts compared with a single arterial graft (SAG) is associated with a reduction in the composite outcome of death from any cause, any stroke, post-discharge myocardial infarction and/or repeat revascularization.The secondary hypothesis is that in these patients, the use of 2 or more arterial grafts compared with a SAG is associated with improved survival. The ROMA trial is a prospective, unblinded, randomized event-driven multicentre trial comprising at least 4300 subjects. Patients younger than 70 years with left main and/or multivessel disease will be randomized to a SAG or multiple arterial grafts to the left coronary system in a 1:1 fashion. Permuted block randomization stratified by the centre and the type of second arterial graft will be used. The primary outcome will be a composite of death from any cause, any stroke, post-discharge myocardial infarction and/or repeat revascularization. The secondary outcome will be all-cause mortality. The primary safety outcome will be a composite of death from any cause, any stroke and any myocardial infarction. In all patients, 1 internal thoracic artery will be anastomosed to the left anterior descending coronary artery. For patients randomized to the SAG group, saphenous vein grafts will be used for all non-left anterior descending target vessels. For patients randomized to the multiple arterial graft group, the main target vessel of the lateral wall will be grafted with either a radial artery or a second internal thoracic artery. Additional grafts for the multiple arterial graft group can be saphenous veins or supplemental arterial conduits. To detect a 20% relative reduction in the primary outcome, with 90% power at 5% alpha and assuming a time-to-event analysis, the sample size must include 845 events (and 3650 patients). To detect a 20% relative reduction in the secondary outcome, with 80% power at 5% alpha, the sample size must include 631 events (and 3650 patients). To be conservative, the sample size will be set at 4300 patients. The primary outcome will be tested according to the intention-to-treat principle. The primary analysis will be a Cox proportional hazards regression model, with the treatment arm included as a covariate. If non-proportional hazards are observed, alternatives to Cox proportional hazards regression will be explored.

Hospital Surgical Volumes and Mortality after Coronary Artery Bypass Grafting: Using International Comparisons to Determine a Safe Threshold.

OBJECTIVE: To estimate a safe minimum hospital volume for hospitals performing coronary artery bypass graft (CABG) surgery. DATA SOURCE: Hospital data on all publicly funded CABG in five European countries, 2007-2009 (106,149 patients). DESIGN: Hierarchical logistic regression models to estimate the relationship between hospital volume and mortality, allowing for case mix. Segmented regression analysis to estimate a threshold. FINDINGS: The 30-day in-hospital mortality rate was 3.0 percent overall, 5.2 percent (95 percent CI: 4.0-6.4) in low-volume hospitals, and 2.1 percent (95 percent CI: 1.8-2.3) in high-volume hospitals. There is a significant curvilinear relationship between volume and mortality, flatter above 415 cases per hospital per year. CONCLUSIONS: There is a clear relationship between hospital CABG volume and mortality in Europe, implying a "safe" threshold volume of 415 cases per year.