EUROPEAN SOCIETY OF CARDIOLOGY CONGRESS 2010 FEV1 predicts length of stay and in-hospital mortality in patients undergoing cardiac surgery Nicholas L Mills, David A McAllister, Sarah Wild, John D MacLay, Andrew Robson, David E Newby, William MacNee, John A Innes, Vipin Zamvar Dr Nicholas L Mills British Heart Foundation Intermediate Clinical Research Fellow University/BHF Centre for Cardiovascular Science The University and Royal Infirmary of Edinburgh No conflicts of interest
Changes in the patient population undergoing cardiac surgery Department of Cardiothoracic Surgery, the Edinburgh Heart Centre 2010
Number of coronary artery bypass operations and percutaneous coronary interventions per year, 1991 to 2007, United Kingdom www.heartstats.org
Observed in-hospital mortality after cardiac surgery among all patients CABG/MVR CABG/AVR CABG In-hospital morbidity and mortality in 67,764 patients undergoing cardiac surgery at 22 centers in the US National Cardiovascular Network. Alexander, K. P. et al. J Am Coll Cardiol 2000;35:731-738
Logistic EuroSCORE EuroSCORE Calculator app (by Edward Bender) Nashef SAM et al Eur J Cardiothorac Surg 1999;16:9-13.
Logistic EuroSCORE Roques F et al Eur Heart J 2003;24:1-2.
Forced Expiratory Volume in one second (FEV 1 ) as a measure of physiological reserve Survival among isolated CABG surgery patients with or without COPD FEV 1 is a robust and repeatable measure of pulmonary physiology It is well established that patients with chronic obstructive pulmonary disease (COPD) have poorer outcomes following cardiac surgery Leavitt, B. J. et al. Circulation 2006;114:I-430-I-434
FEV 1 as a measure of physiological reserve All-cause mortality risk by relative FEV1 for lifelong non-smokers and heavy current smokers Relationship between FEV1, smoking status and odds ratio for cardiovascular mortality Current smoker Ex-smoker Never smoker FEV 1 is also an independent predictor of both all cause and cardiovascular mortality in the general population Hole et al BMJ. 1996;313:711-715; Young et al. Eur Respir J. 2007;30:616-22
Hypothesis FEV 1 will predict in-hospital mortality and length of stay in patients undergoing cardiac surgery The association between FEV 1 and outcome will be independent of cigarette smoking and pulmonary disease
Study design and methodology Retrospective single-centre cohort study January 1 st 2001 to December 31 st 2007 All patients (aged >40 years) who underwent coronary artery bypass grafting (CABG) and/or valve repair or replacement surgery Patients who underwent emergency surgery were excluded Patient characteristics, details of cardiac surgery and clinical outcomes were recorded in an electronic cardiac surgical database (TOMCAT Clinical Systems, Philips) by trained staff at the time of admission 1 the logistic EuroSCORE was calculated at the time of surgery 1 In an external Scottish Quality Improvement Programme audit in 2008 these data were found to be 98.5% accurate.
Spirometry and pulmonary disease Spirometry was performed using a standard wedge bellows spirometer (Model S, Vitalograph, UK) according to British Thoracic Society standards by trained clinical physiologists Airflow obstruction was defined a priori as an FEV 1 /FVC ratio below 0.7 and FEV 1 percent predicted less than 80% 1 Chronic pulmonary disease defined as the long term use of bronchodilators or steroids for lung disease 2 1 National Institute for Clinical Excellence (2004) 2 EuroSCORE (1999)
Statistical analysis Primary outcome in-patient mortality Secondary outcome continuous in-patient hospital stay Statistical models: - The odds ratio for in-hospital mortality and proportional change in in-patient hospital stay were estimated by quintile of FEV1, and per one standard deviation decrement in FEV1, using logistic regression and linear regression respectively Comparison of FEV 1 with EuroSCORE Discrimination was compared using area under the curve and the Net Reclassification Index (Delong et al 1988,Pencina et al, 2008) Calibration was compared using the le Cessie-van Houwelingen normal test statistic (Hosmer et al, 1987)
Study population 2,241 patients met eligibility criteria (age >40 years, elective or urgent CABG or valve surgery) 93% of patients (n= 2,082) had spirometry performed prior to surgery as part of routine assessment Majority had spirometry performed within 30 days of surgery with 97% performed within the same year
Frequency Study population 200 150 100 50 0 0 1,000 2,000 3,000 4,000 5,000 6,000 FEV1, ml Airflow obstruction on spirometry (FEV 1 /FVC <0.7 and FEV 1 <80% predicted) was present in 318 (15%) patients Clinical diagnosis of COPD in 28 (1.3%), asthma in 24 (1.2%) and chronic pulmonary disease in 20 (1.0%) patients
Study population
In-hospital mortality Odds ratio for in-hospital mortality for a given FEV 1 compared to the odds ratio for the mean FEV 1 after adjusting for age, sex, height, BMI, type of operation, smoking status, COPD, chronic pulmonary disease, pre-operative angina, and pre-operative dyspnoea P=0.28 for for non-linearity
In-hospital mortality by quintile of FEV 1 Q1 Q2 Q3 Q4 Q5 FEV 1 litres 1.31 1.95 2.41 2.89 3.57 Number of patients 416 417 416 417 416 Mortality, n (%) 36 (8.7%) 16 (3.8%) 12 (2.9%) 3 (0.7%) 4 (1.0%) Odds Ratio 1 5.23 2.76 2.28 0.69 1 In-hospital mortality is 9-times higher in the lowest quintile compared to the highest quintile of FEV 1 (P<0.001) Adjusted odds ratios are 5-times higher (P<0.001) In-hospital death was 2.58-fold higher per standard deviation decrement (800mL) in FEV 1 1 adjusted for age, sex, height, body mass index, type of operation, smoking status, COPD, chronic pulmonary disease, pre-operative angina, pre-operative dyspnoea.
In-hospital mortality by quintile of FEV1
Odds ratio of in-hospital mortality per standard deviation decrement in FEV 1 n Odds ratio (95% CI) P value All patients 2,082 2.11 (1.48-3.08) <0.001 Elective surgery 959 2.18 (1.06-4.52) 0.03 Urgent surgery 1,123 2.39 (1.46-4.00) <0.001 CABG only 1,330 1.63 (0.8-3.28) 0.17 Valve or valve + CABG 752 2.52 (1.54-4.21) <0.001 No COPD or FEV 1 <0.70 1,544 2.13 (1.05-4.38) 0.04 0 1 2 3 4 5 Odds ratio (95% CI) 1 adjusted for age, sex, height, body mass index, type of operation, smoking status, COPD, chronic pulmonary disease, pre-operative angina, pre-operative dyspnoea.
Length of stay by quintile of FEV 1 Q1 Q2 Q3 Q4 Q5 FEV 1 litres 1.31 1.95 2.41 2.89 3.57 Number of patients 416 417 416 417 416 Length of stay, median days (IQR) 10 (7-15) 8 (7-13) 7 (6-10) 7 (6-9) 7 (6-8) Relative length of stay 1 1.35 1.30 1.16 1.05 1 Median hospital stay was 3 days longer amongst patients in the lowest quintile for FEV1 compared to patients in the highest quintile (P<0.001) 1 adjusted for age, sex, height, weight, type of operation, urgency of procedure, deprivation score, smoking status, recent myocardial infarction, extra-cardiac arteriopathy, diabetes, hypertension, stroke, atrial fibrillation, left ventricular function, asthma, COPD, bronchodilator use, preoperative renal failure, and pre-operative angina and dyspnoea
Length of stay in hospital Relative length of stay for a given FEV 1 compared to the length of stay for the mean FEV 1 adjusted for age, sex, height, BMI, type of operation, urgency of procedure, deprivation score, smoking status, extracardiac arteriopathy, diabetes, hypertension, left ventricular function, COPD, chronic pulmonary disease.. P=0.11 for for non-linearity
Receiver operating characteristic curves for EuroSCORE and FEV 1 FEV1 predicted in-hospital mortality almost as well as the EuroSCORE with area under the receiver-operating-characteristic curves (AUC) of 0.74 for FEV 1 as a single variable and 0.78 for the EuroSCORE (no significant difference, P=0.15)
Receiver operating characteristic curves for EuroSCORE and FEV 1 Addition of FEV1 to the EuroSCORE increased the AUC compared to EuroSCORE alone (AUC of 0.80 versus 0.78), but the difference was not statistically significant (p=0.26)
Discrimination using EuroSCORE and FEV 1 Net Reclassification Index 36% of patients (95% CI 12-59%, P=0.003) were more appropriately classified when FEV 1 was combined with the EuroSCORE compared to the EuroSCORE alone Calibration Model calibration was also better with FEV 1 and EuroSCORE combined (z=0.14; P=0.89) than with the EuroSCORE alone (z=1.27, P=0.21).
Limitations Single-centre retrospective study No separate calibration dataset No long term outcomes No intermediate outcomes to explore the mechanism of this association (e.g. post-operative pneumonia)
Conclusions FEV 1 is an excellent independent predictor of in-hospital mortality and length of stay for patients undergoing cardiac surgery This was true across all patients and not restricted to patients with physician diagnosed pulmonary disease or airflow obstruction Addition of FEV 1 to the EuroSCORE improved calibration and discrimination for in-hospital mortality Spirometry is a widely available physiological measure that is already performed in the majority of patients undergoing cardiac surgery Future risk prediction models should include physiological variables such as FEV 1
Acknowledgements British Heart Foundation Intermediate Clinical Research Fellowship (FS/10/024/28266) University of Edinburgh David A McAllister John D MacLay Sarah A Wild David E Newby Alastair J Innes William W MacNee Western General Hospital Andrew Robson Alastair J Innes Edinburgh Heart Centre Vipin Zamvar Jackie Howlett