Back to 2016 Annual Meeting Posters

Objective: Prolonged Air Leaks (>5 days) are a commonly studied complication with significant patient morbidity, but the impact of any air leak is less well understood. This study assesses the incidence, risk factors, and clinical impact of air leaks during the hospital episode of care.
Methods: The Premier Perspective® Database contains billing data from over 600 hospitals in the U.S. All elective primary lobectomy, segmentectomy, and wedge resections from 2012-2014 were identified. During the index hospitalization, Air Leak Complications (ALC) were identified as a composite of air leak and pneumothorax ICD-9 diagnosis codes. Patient, procedure, and hospital factors were identified (Table 1) and included in multivariate models evaluating ALC risk factors and the impact of ALC on index hospitalization mortality and length of stay (LOS). The multivariate models accounted for the clustering of patients within hospitals; p values of <0.05 were considered to be statistically significant.
Results: A total of 21,150 patients undergoing lung surgery were included in the analysis: lobectomy (n=10,946), segmentectomy (n=1,788), and wedge (n=8,416). The overall incidence of ALC was 24.26% (95%CI[23.68,24.83]) and varied with resection type: lobectomy 29.20% (95%CI[28.35,30.05]), segmentectomy 22.04% (95%CI[20.11,23.96]) , and wedge 18.30% (95%CI[17.47,19.12]). Relevant risk factors for ALC included (Table 1): resection type, thoracotomy surgical approach, male gender, indication for surgery, and presence of chronic obstructive pulmonary disease (COPD). The overall mortality in the study sample was 1.06% (95%CI[0.92,1.20]), and the mean LOS was 5.7 (SE=0.04) days. After controlling for patient, procedure and hospital factors, ALC was associated with increased mortality OR 1.90 (95%CI[1.42,2.55]) and increased mean LOS of 2.5 days (p<0.01).
Conclusion: This analysis shows that multiple patient, procedural, and provider characteristics increase the risk for ALC. Air Leak Complications after lung resections are not only a frequent complication but are associated with increased mortality and hospital LOS.

Table 1: Logistic Regression Model for occurrence of ALC (N=21,150)
Variable	Effect	OR	95% CI
Resection Type (REF: Lobectomy)	Segmentectomy	0.769	0.668	0.885
	Wedge	0.726	0.661	0.798
Approach (REF: VATS*)	Thoracotomy	1.132	1.008	1.271
Race (REF: Caucasian)	African American	0.958	0.835	1.100
	Other	1.055	0.902	1.234
Gender (REF: Female)	Male	1.108	1.037	1.184
Age (REF: >74 years)	18-44	1.369	1.098	1.705
	45-54	0.928	0.786	1.096
	55-64	0.899	0.789	1.026
	65-74	1.012	0.926	1.105
Marital Status (REF: Single)	Married	0.928	0.858	1.004
	Other	1.035	0.805	1.331
Payer (REF: Other)	Commercial	0.915	0.708	1.182
	Medicaid	0.923	0.699	1.219
	Medicare	0.948	0.734	1.225
Year of Surgery (REF:2014)	2012	0.854	0.751	0.971
	2013	0.960	0.866	1.064
Primary Indication (REF: Cancer)	Pulmonary Fibrosis	0.803	0.687	0.938
	Other	1.177	1.026	1.351

Table 1 cont: Logistic Regression Model for occurrence of ALC (N=21,150)
Teaching Status (REF: Non Teaching)	Teaching	0.996	0.829	1.197
Hospital Bed Size (REF: >500 beds)	1-300	1.078	0.832	1.397
	301 - 500	1.084	0.842	1.395
Provider Region (REF: West)	Midwest	1.065	0.792	1.430
	South	0.905	0.707	1.158
	Northeast	0.691	0.497	0.963
Provider Urbanicity (REF: Urban)	Rural	1.088	0.854	1.385
Hospital Volume of Lung Resections	1 to 50	0.927	0.601	1.429
(REF: >300 surgeries)	51 to 150	0.927	0.589	1.460
	151 to 300	1.203	0.740	1.955
Provider Costing Type (REF: RCC**)	Procedural	1.067	0.894	1.274
Physician Specialty	Cardiac Surgeon	1.046	0.895	1.222
(REF: Thoracic Surgeon)	General Surgeon	0.939	0.767	1.148
	Other	0.996	0.812	1.221
Charlson Comorbidity Index	Score of 0	0.648	0.555	0.758
(REF: >4)	Score of 1-2	1.024	0.923	1.136
	Score of 3-4	1.260	1.156	1.373
Specific Comorbid Conditions	COPD	1.803	1.630	1.994
	Depression	0.963	0.865	1.073
	Hypertension	0.911	0.852	0.975
	Diabetes	0.821	0.746	0.903
	Obesity	0.631	0.554	0.719
* Video Assisted Thoracic Surgery (VATS)
**Ratio of cost to charge (RCC) method