A large database study in the general population in England

Transcription

1 Time-dependent markers of comorbidity severity and change are associated with increased risk of mortality in heart failure: A large database study in the general population in England Claire A. Rushton, NIHR Doctoral Fellow, Lecturer & Cardiology Nurse Lucy Riley, Sta?s?cian Peter Jones, Professor of Sta?s?cs Umesh T. Kadam, Professor of Health Services Research & Clinical Epidemiology Health Services Research Unit HS RU

2 Background Heart failure (HF) is a common chronic disease with increasing prevalence HF prognosis is poor, varies across individuals and changes over =me Comorbidity is common in HF and the presence of comorbidity is known to influence prognosis

3 Diabetes 1.34 (1.24, 1.46) COPD 1.39 (1.21, 1.60) CKD 1.52 (1.34, 1.71)

4 Background Chronic disease comorbidity changes over =me Current evidence Renal disease which incorporates change Limited to hospital sedngs and selected HF groups LiEle evidence of comorbidity severity and change on prognos=c outcomes

5 Hypothesis In the general HF popula=on, comorbidity stra=fica=on by increasing severity and longitudinal change would be associated with worse mortality.

6 Objec?ves To inves=gate associa=ons between: recent comorbidity severity and mortality risk. recent change in comorbidity severity and mortality risk.

7 Methods Study popula=on Incident cohort of HF pa=ents aged 40 years Clinical Prac=ce Research Datalink (CPRD) with first consulta=on code for heart failure Cohort entry 1 st January 2002 and 1 st March 2012.

8 Methods: Exposures Diabetes (status) Consulta=on code or or prescrip=on Severity (drugs (physiological measure measure ) ) HbA1c None Severity change <5.5% No HbA1c drug category % Oral hypoglycaemic change >1% increase % drugs Increase in drug % category >1% decrease Insulin only % Decrease in drug >9.5% category <1% change

9 Methods: Exposures CKD egfr<60 ml/min/m2 Severity (physiological measure) ) Severity change > (ref) <15 Classifica?on 12 Absolute Percentage change >15mls >25% decrease 6mls 6-25% to decrease 15mls decrease any Any % increase 0mls 0-5% to decrease 5mls decrease (ref) (ref)

10 Confounders: Methods Person and social (age, gender, depriva=on) Clinical factors (BMI, BP, Cholesterol, Haemoglobin) Lifestyle factors (alcohol, smoking) HF medica=ons (ACEi, ARB, B- blocker, diure=c). Outcomes: All- cause mortality

11 Analysis Nested case- control study Risk set sampling Cases matched to 4 controls on follow- up and calendar =me

12 Case Risk Cohort set control sampling study study First HF consultation code Case cohort sampling Physiological severity change measured over 1 year Most recent measure within 3-years 6- mnths CPRD HF COHORT Most recent measure within 3-years 4-month time window Control sample mirrors the distribu=on 4-month of time window 1- year exposure!me in the source popula=on Drug severity change measured over 1 year Odds ra=o will be an unbiased es=ma=on of the incidence rate ra=o Cumula?ve sampling Allows the comparison of exposure across cases and controls in real =me Risk set sampling

13 Analysis Imputa=on of missing confounders Tested con=nuous variables for linearity Inves=gated collinearity One variable selected if >50% Con=nuous variables centred at their means if quadra=c extension included

14 Analysis Condi?onal Logis?c Regression Unadjusted and adjusted for all confounders For CKD change further adjustment for baseline renal func=on

15 Results 50,114 HF pa=ents Follow- up, 0-12 years (2.57 years [ ]) Age median 78 years [IQR 71-85] 47.1% female 26,729 (53.3%) died; 106,916 controls

16 Diabetes prevalence Baseline: 10,533 (21%) Matched: 31,962 (23.9%) 90% recent HbA1c (4 months; IQR ) 82% second measure (10 months; IQR ) CKD prevalence Baseline: 20,084 (49.8%) Matched: 66,301 (55.4%) 90% recent egfr (3 months; IQR ) 79% second measure (10 months; IQR )

17 Diabetes HbA1c Severity Comorbidity effects Stratified comorbidity effects OR 95% CI Reference group: No Diabetes HbA1c categories (%) Diabetes (all) < >9.5

18 Diabetes Drug Severity 1.7 Comorbidity effects Stratified comorbidity effects OR 95% CI Reference group: No Diabetes Diabetes (all) No medica=ons Oral hypoglycaemic +/- insulin Insulin only

19 Diabetes HbA1c Severity Change 1.7 Comorbidity effect Stratified comorbidity effects OR 95% CI Diabetes (all) DM <1% change HbA1c Reference group: No Diabetes DM >1% increase HbA1c DM <1% change or >1% increase HbA1c (combined group) DM >1% decrease HbA1c

20 CKD severity 3.26 OR 95% CI < egfr ml/min/m >105

21 CKD Severity Change OR 95% CI Classification 1 (absolute change) Classification 2 2 (percentage change) change) Reference group: 5mls change Reference group: 5% change Any increase 6mls to to 15mls >15mls >15mls decrease Any Any % % increase 6-25% 6-25% decrease >25% >25% decrease decrease decrease decrease

22 Conclusions In the general popula=on of HF, diabetes and CKD comorbidi=es are associated with an increased risk of mortality. This risk is significantly stra=fied by measures of recent comorbidity severity and change. This has important implica=ons for developing new HF comorbidity interven=ons to improve outcomes in the general popula=on

23 Acknowledgements Claire Rushton is funded by a National Institute for Health Research Doctoral Research Fellowship