Outcome assessment of heart failure patients in Hull and East Yorkshire under telehealth care5th July 2016

John StamfordChandra KambhampatiSteffen PauwsAndrew L Clark

Content

Change the way you think about Hull | 5 July 2016 | 2

• iCase EPSRC Project Overview [EP/L505468/1]

• Introduce Home Telehealth Monitoring (HTM)– What is HTM?– Literature

• Longitudinal Dataset– Hull Lifelab– Matching and extracting patients

• Evaluation– Results

Project Overview

Change the way you think about Hull | 5 July 2016 | 3

• Project:– Predictive Algorithms for Telehealth Service

Improvement and Evaluation (PATSIE)

• Overall project scope…1. Better models of mortality and hospitalization

risk facilitating patient selection for telehealth2. Accurate impact analysis of telehealth service

delivery with respect to outcomes, quality of care and cost-effectiveness

The Impact of Heart Failure

Change the way you think about Hull | 5 July 2016 | 4

Impact• 2006/2007 England and Wales

• 250,000 hospital deaths and discharges • 65,000 of them being a first time diagnosis (Cleland 2011)

• £563 million per year in the UK (Cleland 2011)• 1 million inpatient bed days per year (NICE 2012)

• 5 million people in the USA (Soran 2008)• 10 million in Europe (Giamouzis 2012)

• Readmission• 30% within 3 months (Zhang 2013)• 50% within 6 months (Woodend 2008, Giamouzis 2012)

What is Home Telehealth Monitoring?

Change the way you think about Hull | 5 July 2016 | 5

Patient

Measurements CriteriaResting heart rate < 50 beats/min

Or> 80 beats/min

Systolic blood pressure < 90 mm HgOr> 140 mm Hg

Weight Change Change > 2kgCleland (2005), Dendale (2014)

How does HTM perform (literature)?

Change the way you think about Hull | 5 July 2016 | 6

(Inglis 2011)

Our Work

Change the way you think about Hull | 5 July 2016 | 7

• Aims– Analyse the effectiveness of HTM for patients with CHF

• Hull-Lifelab (Clark et al 2014)

– 6,300 patients

• 380 HTM Patients– 129 useable

• The problem– Matching the HTM patients with similar patients

TablesBaselineNo. Records

TotalNo. Records

No. Variables

Blood/Laboratory 5,802 18,412 75

Medication 6,287 14,342 131

Echocardiogram 6,021 13,314 78

Examination 6,003 14,155 52

QoL 4,488 10,130 181

History 6,254 - 70

Hospitalisation - 27,667 48

Mortality - 6,287 112

Baseline (Before Matching)

Change the way you think about Hull | 5 July 2016 | 8

Variable All Normal Care HTM P-valueNumber of Patients 3214 3085 129 NAAge 72.2 (11.3) 72.4 (11.2) 67.6 (11.4) < 0.001Female 40% 40.70% 20.90% < 0.001ACE 55.40% 54.60% 78.30% < 0.001Betablocker 58% 57.30% 79.20% < 0.001Digoxin 15.50% 15% 30.20% < 0.001Diuretic 68% 67% 97.20% < 0.001Calcium Channel Blocker 2.90% 3% 0% < 0.001Furosemide EquivDailyDose (mg) 36.3 (43.4) 34.9 (42.3) 78.1 (52.5) < 0.001Warfarin 23.60% 23.10% 36.80% 0.002BP Systolic (mmHg) 139.1 (25) 139.5 (25) 126.2 (23.5) < 0.001BP Diastolic (mmHg) 79 (14) 79.1 (13.9) 74.7 (14) 0.002NYHA Exam >= 3 32% 31.70% 40.40% 0.077NTproBNP (ngL) 2018.1 (3941.2) 1977.3 (3844.2) 3115 (5913.7) 0.077Creatinine (umolL) 106.3 (55.2) 105.7 (54.8) 125.5 (63.5) 0.003Urea (mmolL) 7.9 (5.1) 7.8 (4.9) 9.8 (7.8) 0.015Sodium (mmolL) 137.9 (3.2) 138 (3.2) 136.7 (3.6) < 0.001Myocardial Infarction 12.40% 11.80% 29.90% < 0.001Coronary Artery Bypass Graft 6.30% 6% 16.90% < 0.001

Problem

Change the way you think about Hull | 5 July 2016 | 9

?

HTM Patients

Comparison Patients

?

2. Match

Hull-Lifelab Dataset

Propensity Matching

Change the way you think about Hull | 5 July 2016 | 10

• Estimating the likelihood of receiving treatment based on covariate scores (Osborn 2005, Austin 2009)

• Logistic Regression Model

– Dependent Variable• if the patient received HTM

– Independent Variables• age, gender and weight together with laboratory

variables (sodium (mmol/L), urea (mmol/L) and amino-terminal pro-B-type natriuretic peptide (NTproBNP) (ng/L)) and medication (furosemide (mg) and betablocker use)

– p(x) difference is < 0.02

Baseline (After Matching)

Change the way you think about Hull | 5 July 2016 | 11

Variable All Normal Care HTM P-valueNumber of Patients 202 101 101 NAAge 68.3 (12.5) 68.9 (12.9) 67.8 (12.2) 0.56Female 25.70% 28.70% 22.80% 0.42ACE 74.80% 71.30% 78.20% 0.33Betablocker 80.70% 82.20% 79.20% 0.72Digoxin 26.70% 24.80% 28.70% 0.63Diuretic 96.50% 96% 97% 1.00Calcium Channel Blocker 1% 2% 0% <0.001Furosemide EquivDailyDose (mg) 76.8 (48.4) 78.8 (48.8) 74.9 (48.2) 0.56Warfarin 37.10% 38.60% 35.60% 0.77BP Systolic (mmHg) 126.9 (24.4) 127.4 (25.3) 126.4 (23.6) 0.78BP Diastolic (mmHg) 75.7 (14.1) 76.6 (14.5) 74.9 (13.7) 0.39NYHA Exam >= 3 39.60% 39.60% 39.60% 1.00NTproBNP (ngL) 3575.8 (6070.7) 4015.9 (5931.5) 3163.7 (6207.5) 0.39Creatinine (umolL) 126.1 (66.7) 126.9 (68.5) 125.2 (65) 0.87Urea (mmolL) 10 (7.4) 10.3 (6.9) 9.8 (8) 0.64Sodium (mmolL) 136.9 (3.8) 137 (3.9) 136.8 (3.7) 0.76Myocardial Infarction 15.60% 5.20% 27.10% < 0.001Coronary Artery Bypass Graft 11.60% 9.10% 14.30% 0.47

Results

Change the way you think about Hull | 5 July 2016 | 12

Survival Analysis

Change the way you think about Hull | 5 July 2016 | 13

Alive Dead TotalHTM 93 8 101Normal Care 81 20 101Total 174 28 202

One Year Mortality (p = 0.025)

• The normal care group had a greater likelihood of dying within the first year (HR: 3.20, 95% CI: 1.40 – 7.28, p = 0.006).

Survival Estimates (Kaplan Meier)

Change the way you think about Hull | 5 July 2016 | 14

(Log rank test p = 0.00345)

Cox Proportional Hazard Ratio

Hazard Ratio 3.2

95% CI 1.404– 7.28

P-value 0.006

Change the way you think about Hull | 5 July 2016 | 15

First to Event Analysis(Composite Death or Hospitalisation)

(Log rank test p = 0.11)

Cox Proportional Hazard Ratio

Hazard Ratio 0.74

95% CI 0.51 – 1.07

P-value 0.11

Number of Hospitalisation

Change the way you think about Hull | 5 July 2016 | 16

Days Alive and Out of Hospital (DAOH)

Change the way you think about Hull | 5 July 2016 | 17

• Repeat Event Analysis

• Each patient has a max DAOH of 365 days

• HTM Group had 3273 more days (alive and out of hospital)

• Patients receiving HTM had an average of 32.4 more DAOH than the normal care group (95% CI: 10.1 – 54.7 days, p = 0.005).

DAOH %DAOHHTM 35,385.3 96%Normal Care 32,112.4 87%

Conclusion

Change the way you think about Hull | 5 July 2016 | 18

• Results show, in Hull and East Riding of Yorkshire…– HTM patients have better survival rates– Less likely to die within one year– Have more days alive and out of hospital

• Propensity matching– Reduces differences in baseline characteristics– Allows valid outcome assessment

• Repeat event analysis (DAOH) overcomes limitation of first to event analysis

• Possible future work– Understand the results– Develop models to…

• Identify which patients would be more likely to benefit from HTM• Develop models to predict hospitalisation events

References

Change the way you think about Hull | 5 July 2016 | 19

• Austin, P. C. Discussion of ‘A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003, Stat. Med., vol. 28, no. 15, pp. 1999–2011, 2009.

• Clark, A. L. Cleland, J. G. F. Goode, K Kazmi, S. The Hull-Lifelab Dataset - A longitudinal cohort study of patients diagnosed with Heart Failure, 2014.

• Cleland, J.G., et al., Noninvasive home telemonitoring for patients with heart failure at high risk of recurrent admission and death: the Trans-European Network-Home-Care Management System (TEN-HMS) study. J Am Coll Cardiol, 2005. 45(10): p. 1654-64

• Cleland, J.G., et al., The national heart failure audit for England and Wales 2008-2009. Heart, 2011. 97(11): p. 876-86

• Dendale, P., et al., Effect of a telemonitoring-facilitated collaboration between general practitioner and heart failure clinic on mortality and rehospitalization rates in severe heart failure: the TEMA-HF 1 (TElemonitoring in the MAnagement of Heart Failure) study. European Journal of Heart Failure, 2012. 14(3): p. 333-340

• Giamouzis, G. Mastrogiannis, D. Koutrakis, K. Karayannis, G. Parisis, C. Rountas, C. Adreanides, E. Dafoulas, G. E. Stafylas, P. C. Skoularigis, J. Giacomelli, S. Olivari, Z. and Triposkiadis, F. “Telemonitoring in chronic heart failure: A systematic review,” Cardiol. Res. Pract., vol. 1, 2012.

• Inglis, S. C., Clark, R. A., McAlister, F. A., Stewart, S., & Cleland, J. G. F. (2011). Which components of heart failure programmes are effective? A systematic review and meta-analysis of the outcomes of structured telephone support or telemonitoring as the primary component of chronic heart failure management in 8323 patients: Abridged Coc. European Journal of Heart Failure, 13(9), 1028–1040.

• NICE, “NICE guidance recommends new treatment for some people with chronic heart failure,” 2012 Available: http://www.nice.org.uk/news/press-and-media/nice-guidance-recommends-new-treatment-for-some-people-with-chronic-heart-failure.

• Osborn ,C. E. Statistical Applications For Health Information Management. Jones & Bartlett Learning, 2005.

• Soran, O.Z., et al., A randomized clinical trial of the clinical effects of enhanced heart failure monitoring using a computer-based telephonic monitoring system in older minorities and women. J Card Fail, 2008. 14(9): p. 711-7

• NHS (2014) - http://www.nhs.uk/conditions/Heart-failure/Pages/Introduction.aspx

• Woodend, A.K., et al., Telehome monitoring in patients with cardiac disease who are at high risk of readmission. Heart Lung, 2008. 37(1): p. 36-45.

• Zhang, J. Goode, K. M. Rigby, A. Balk, A. H. M. M. and Cleland, J. G. “Identifying patients at risk of death or hospitalisation due to worsening heart failure using decision tree analysis: Evidence from the Trans-European Network-Home-Care Management System (TEN-HMS) Study,” Int J Cardiol, vol. 163, no. 2, pp. 149–156, Feb. 2013