Presentation is loading. Please wait.

Presentation is loading. Please wait.

Sonam Maghera 3 BSc, Sudhir Sundaresan 1,2 MD, P James Villeneuve 1,2 MD PhD, Andrew J Seely 1,2 MD PhD, Donna E Maziak 1 MD MSc,, Farid M Shamji 1 MD,

Similar presentations


Presentation on theme: "Sonam Maghera 3 BSc, Sudhir Sundaresan 1,2 MD, P James Villeneuve 1,2 MD PhD, Andrew J Seely 1,2 MD PhD, Donna E Maziak 1 MD MSc,, Farid M Shamji 1 MD,"— Presentation transcript:

1 Sonam Maghera 3 BSc, Sudhir Sundaresan 1,2 MD, P James Villeneuve 1,2 MD PhD, Andrew J Seely 1,2 MD PhD, Donna E Maziak 1 MD MSc,, Farid M Shamji 1 MD, and Sebastien Gilbert 1,2 MD, 1 Division of Thoracic Surgery, The Ottawa Hospital 2 The Ottawa Hospital Research Institute 3 Faculty of Medicine, University of Ottawa Ottawa, Canada Predictors of Prolonged Air Leak After Lung Resection

2 Disclosure  None

3 Background  Prolonged air leak is significant problem:  3-25% of lung resections  ↑ LOS, ↑ cost  ↑ risk of pleural complications  Patient suffering (prolonged chest tube drainage)  Predictive models:  Complex/counter-intuitive  Rely on intraoperative and/or postoperative factors

4 Objective  Develop a scoring system to identify patients at risk for prolonged air leak Hypothesis  Patients at higher risk can be identified using widely available preoperative data

5 Methods I  Single institution  Elective pulmonary resections:  Prospective model building cohort (n=225)  Retrospective model testing cohort(n=100)  Outcome:  Prolonged air leak (> 7 days) requiring hospitalization

6 Methods II  Preoperative factors selected based on literature  Categorize variables:  FEV1% < 80% or ≥ 80%  DLCO% < 80% or ≥ 80%  MRC dyspnea score = 1 or > 1  Univariate analysis:  Identify potential predictors (p < 0.2)

7 Methods III  Multivariate analysis:  Coefficients rounded to nearest ½ unit  Calculate score for each patient  Sensitivity/specificity analysis:  Identify score threshold  Best compromise between sensitivity and false-positive rate (ROC analysis)

8 Univariate Analysis Air Leak ≤ 7 Days (n=207) Air Leak > 7 Days (n=18) p Age (years)67 (59-73)65 (55-76)0.71 Male gender79 (38)11 (61)0.08 BMI28 (24-31)24 (22-27)<0.01 Smoker (Yes)145 (70)17 (94)0.03 MRC Dyspnea Score (>1)39 (19)7 (39)0.06 FEV1% (<80%)90 (44)11 (61)0.22 DLCO% (<80%)131 (63)17 (94)0.01 Anatomic lung resection144 (70)15 (83)0.29 Continuous variables are expressed as median with interquartile range

9 Prolonged Air Leak Scoring System QuestionsIf answer = Yes Male gender?1 point History of smoking?2 points Dyspnea when hurrying or walking up a hill? (MRC > 1) 1 point BMI ≤ 25?½ point DLCO% < 80%?2 points Total score > 4? At risk for prolonged air leak

10 ROC Curve (n=225) Score > 4

11 Performance: Air Leak Score > 4 Model Derivation (n=225) Model Validation (n=100) p Prolonged Air Leak18 (8)7 (7)0.83 Sensitivity83% (58-96)86% (42-99)1.0 Specificity65% (58-71)46% (36-57)0.07 PPV18% (11-28)11% (4-22)0.34 NPV98% (93-99)98% (86-99)1.0

12 Study Limitations  Single-institution  Relatively small number of patients  Retrospective validation cohort  Needs external validation  ↓ PPV and ↑ false-positive rate

13 Conclusions  Model reliably classifies patients into high- risk versus low-risk groups  Model uses simple, widely available preoperative parameters  Clinical applications:  Preoperative counselling  Streamline trials of preventative interventions:  ↓ number of patients to adequately power  ↓ cost  ↓ time to completion

14 Sonam Maghera 3 BSc, Sudhir Sundaresan 1,2 MD, P James Villeneuve 1,2 MD PhD, Andrew J Seely 1,2 MD PhD, Donna E Maziak 1 MD MSc,, Farid M Shamji 1 MD, and Sebastien Gilbert 1,2 MD, 1 Division of Thoracic Surgery, The Ottawa Hospital 2 The Ottawa Hospital Research Institute 3 Faculty of Medicine, University of Ottawa Ottawa, Canada Predictors of Prolonged Air Leak After Lung Resection Thank You


Download ppt "Sonam Maghera 3 BSc, Sudhir Sundaresan 1,2 MD, P James Villeneuve 1,2 MD PhD, Andrew J Seely 1,2 MD PhD, Donna E Maziak 1 MD MSc,, Farid M Shamji 1 MD,"

Similar presentations


Ads by Google