Identifying subtypes of chronic kidney disease with machine learning: development, internal validation and prognostic validation using linked electronic health records in 350,067 individuals.

BACKGROUND: Although chronic kidney disease (CKD) is associated with high multimorbidity, polypharmacy, morbidity and mortality, existing classification systems (mild to severe, usually based on estimated glomerular filtration rate, proteinuria or urine albumin-creatinine ratio) and risk prediction models largely ignore the complexity of CKD, its risk factors and its outcomes. Improved subtype definition could improve prediction of outcomes and inform effective interventions. METHODS: We analysed individuals ≥18 years with incident and prevalent CKD (n = 350,067 and 195,422 respectively) from a population-based electronic health record resource (2006-2020; Clinical Practice Research Datalink, CPRD). We included factors (n = 264 with 2670 derived variables), e.g. demography, history, examination, blood laboratory values and medications. Using a published framework, we identified subtypes through seven unsupervised machine learning (ML) methods (K-means, Diana, HC, Fanny, PAM, Clara, Model-based) with 66 (of 2670) variables in each dataset. We evaluated subtypes for: (i) internal validity (within dataset, across methods); (ii) prognostic validity (predictive accuracy for 5-year all-cause mortality and admissions); and (iii) medications (new and existing by British National Formulary chapter). FINDINGS: After identifying five clusters across seven approaches, we labelled CKD subtypes: 1. Early-onset, 2. Late-onset, 3. Cancer, 4. Metabolic, and 5. Cardiometabolic. Internal validity: We trained a high performing model (using XGBoost) that could predict disease subtypes with 95% accuracy for incident and prevalent CKD (Sensitivity: 0.81-0.98, F1 score:0.84-0.97). Prognostic validity: 5-year all-cause mortality, hospital admissions, and incidence of new chronic diseases differed across CKD subtypes. The 5-year risk of mortality and admissions in the overall incident CKD population were highest in cardiometabolic subtype: 43.3% (42.3-42.8%) and 29.5% (29.1-30.0%), respectively, and lowest in the early-onset subtype: 5.7% (5.5-5.9%) and 18.7% (18.4-19.1%). MEDICATIONS: Across CKD subtypes, the distribution of prescription medication classes at baseline varied, with highest medication burden in cardiometabolic and metabolic subtypes, and higher burden in prevalent than incident CKD. INTERPRETATION: In the largest CKD study using ML, to-date, we identified five distinct subtypes in individuals with incident and prevalent CKD. These subtypes have relevance to study of aetiology, therapeutics and risk prediction. FUNDING: AstraZeneca UK Ltd, Health Data Research UK.

A retrospective cohort study predicting and validating impact of the COVID-19 pandemic in individuals with chronic kidney disease.

Chronic kidney disease (CKD) is associated with increased risk of baseline mortality and severe COVID-19, but analyses across CKD stages, and comorbidities are lacking. In prevalent and incident CKD, we investigated comorbidities, baseline risk, COVID-19 incidence, and predicted versus observed one-year excess death. In a national dataset (NHS Digital Trusted Research Environment [NHSD TRE]) for England encompassing 56 million individuals), we conducted a retrospective cohort study (March 2020 to March 2021) for prevalence of comorbidities by incident and prevalent CKD, SARS-CoV-2 infection and mortality. Baseline mortality risk, incidence and outcome of infection by comorbidities, controlling for age, sex and vaccination were assessed. Observed versus predicted one-year mortality at varying population infection rates and pandemic-related relative risks using our published model in pre-pandemic CKD cohorts (NHSD TRE and Clinical Practice Research Datalink [CPRD]) were compared. Among individuals with CKD (prevalent:1,934,585, incident:144,969), comorbidities were common (73.5% and 71.2% with one or more condition[s] in respective data sets, and 13.2% and 11.2% with three or more conditions, in prevalent and incident CKD), and associated with SARS-CoV-2 infection, particularly dialysis/transplantation (odds ratio 2.08, 95% confidence interval 2.04-2.13) and heart failure (1.73, 1.71-1.76), but not cancer (1.01, 1.01-1.04). One-year all-cause mortality varied by age, sex, multi-morbidity and CKD stage. Compared with 34,265 observed excess deaths, in the NHSD-TRE and CPRD databases respectively, we predicted 28,746 and 24,546 deaths (infection rates 10% and relative risks 3.0), and 23,754 and 20,283 deaths (observed infection rates 6.7% and relative risks 3.7). Thus, in this largest, national-level study, individuals with CKD have a high burden of comorbidities and multi-morbidity, and high risk of pre-pandemic and pandemic mortality. Hence, treatment of comorbidities, non-pharmaceutical measures, and vaccination are priorities for people with CKD and management of long-term conditions is important during and beyond the pandemic.

Lifetime risk of cardiovascular-renal disease in type 2 diabetes: a population-based study in 473,399 individuals.

BACKGROUND: Cardiovascular and renal diseases (CVRD) are major causes of mortality in individuals with type 2 diabetes (T2D). Studies of lifetime risk have neither considered all CVRD together nor the relative contribution of major risk factors to combined disease burden. METHODS: In a population-based cohort study using national electronic health records, we studied 473,399 individuals with T2D in England 2007-2018. Lifetime risk of individual and combined major adverse renal cardiovascular events, MARCE (including CV death and CVRD: heart failure; chronic kidney disease; myocardial infarction; stroke or peripheral artery disease), were estimated, accounting for baseline CVRD status and competing risk of death. We calculated population attributable risk for individual CVRD components. Ideal cardiovascular health was defined by blood pressure, cholesterol, glucose, smoking, physical activity, diet, and body mass index (i.e. modifiable risk factors). RESULTS: In individuals with T2D, lifetime risk of MARCE was 80% in those free from CVRD and was 97%, 93%, 98%, 89% and 91% in individuals with heart failure, chronic kidney disease, myocardial infarction, stroke and peripheral arterial disease, respectively at baseline. Among CVRD-free individuals, lifetime risk of chronic kidney disease was highest (54%), followed by CV death (41%), heart failure (29%), stroke (20%), myocardial infarction (19%) and peripheral arterial disease (9%). In those with HF only, 75% of MARCE after index T2D can be attributed to HF after adjusting for age, gender, and comorbidities. Compared with those with > 1, < 3 and ≥3 modifiable health risk behaviours, achieving ideal cardiovascular health could reduce MARCE by approximately 41.5%, 23.6% and 17.2%, respectively, in the T2D population. CONCLUSIONS: Four out of five individuals with T2D free from CVRD, and nearly all those with history of CVRD, will develop MARCE over their lifetime. Early preventive measures in T2D patients are clinical, public health and policy priorities.

Real world outcomes in platinum sensitive relapsed ovarian, fallopian tube, or peritoneal cancer treated in routine clinical practice in the United Kingdom prior to poly-ADP ribose polymerase inhibitors.

INTRODUCTION: The introduction of poly-ADP ribose polymerase inhibitors in ovarian cancer has demonstrated significantly improved progression free survival in four randomized controlled clinical trials in patients with platinum sensitive relapsed ovarian cancer. While overall survival data remain immature, this real world evidence study sets a baseline for future evaluation of poly-ADP ribose polymerase inhibitors. METHODS: A retrospective chart review was undertaken to investigate real world survival outcomes across 13 National Health Service Trusts in England, Wales, and Scotland. Patients were included if they had platinum sensitive relapsed high grade serous ovarian cancer and had responded to secondline platinum based chemotherapy. Clinical data were collected retrospectively from electronic prescribing records and chart notes. The index date for overall survival analysis was defined as the later of (1) day 1 of the final secondline platinum based treatment or (2) date of response to secondline treatment. The primary objective was overall survival from the index date. Secondary objectives included progression free survival and overall survival by subsequent line of treatment. BRCA mutation status was collected where available. Quality of life questionnaires were not assessed within this study. RESULTS: 233 patients were identified who met the study inclusion criteria. Patient characteristics were consistent with other published data, with a median age of 61 years (range 35-85). Sensitivity analysis of the primary objective demonstrated that the earliest point poly-ADP ribose polymerase inhibitors may be initiated (following completion of secondline chemotherapy) is associated with a median overall survival of 19.8 months. Secondline median overall survival and progression free survival from the index date were 19.3±2.4 months and 7.3±1.2 months, respectively. 144 patients were treated with thirdline chemotherapy with median overall survival and progression free survival from the index date (either date of last cycle of thirdline treatment or date of response to thirdline treatment) of 8.3±2.6 and 4.4±1.8 months, respectively. CONCLUSION: Overall survival was shown to be shorter in this real world study compared with randomized clinical trials, and underlines the differences in clinical outcomes of patients in a real life setting. This baseline real world study has demonstrated poor survival outcomes in this patient group prior to availability of poly-ADP ribose polymerase inhibitors.

Linking hospital patient records for suspected or established acute coronary syndrome in a complex secondary care system: a proof-of-concept e-registry in National Health Service Scotland.

Aims: To implement secondary care electronic record linkage for patients hospitalized with suspected or known acute coronary syndrome (ACS) in a complex regional health care system and evaluate this e-Registry in terms of patterns of service delivery and 1-year outcomes. Methods and results: Existing electronic hospital records were linked to create episodes of care using (i) a patient administration system, (ii) invasive cardiovascular procedure referrals, and (iii) a catheter laboratory record. Data were extracted for admissions (1 October 2013-30 September 2014) with International Classification of Disease (ICD)-10 diagnosis of angina (I200-I209), myocardial infarction (I210-I229), other ischaemic heart disease (I240-I249) or heart failure (I50), linked to other sources to develop a secondary care ACS e-registry and analysed within a Safe Haven. Episodes of care were categorized into care pathways and evaluated in terms of patient characteristics, as well as service delivery metrics and outcomes including mortality. In all, 2327 patients had 2472 episodes of care. Diagnoses were hierarchically classified as ST-elevation myocardial infarction (STEMI) (586, 25.2%), non-ST-elevation myocardial infarction (NSTEMI) (1068, 45.9%), unspecified myocardial infarction (146, 6.3%), unstable angina (527, 22.6%) for the first hospitalization for each patient within the study period. Six care pathways were mapped. Percutaneous coronary intervention rate for STEMI was 80.2% and for NSTEMI 33.1%. Unadjusted all-cause mortality was 9.0% and 3.0% for STEMI and NSTEMI at 30 days, rising to 11.9% and 11.6% at 1 year. Analyses were validated by independent source data verification. Conclusion: The e-registry has enabled analysis of ACS hospitalizations in a complex health care system with implications for quality improvement and research.