- Email: [email protected]
Evaluation of Patient-Related and Procedure-Related Factors Contributing to Pneumothorax Following Thoracentesis
Evaluation of Patient-Related and Procedure-Related Factors Contributing to Pneumothorax Following Thoracentesis* Henri G. Colt, MD, FCCP; Nancy Brewer, RVT; and Edward Barbur, MPH
Objective: To evaluate patient-related and procedure-related risk factors for thoracentesisrelated pneumothorax. Design: Prospective, nonrandomized cohort study. Setting: Pulmonary Special Procedures Unit of a university medical center. Methods: Thoracentesis using either a 22-gauge, a Boutin, or a Cope needle (depending on availability and operator preference) was performed by the pulmonary faculty or by pulmonary physicians-in-training under faculty supervision. In order to control for effusion size and the presence of loculations, chest radiography and pleural ultrasonography were performed prior to each thoracentesis. Potential patient-related and procedure-related risk factors for pneumothorax were analyzed at the procedure level using the presence or absence of pneumothorax on the postprocedure chest radiograph as the sole outcome variable. Results: Two hundred fifty-five thoracenteses were performed in 205 adult patients (113 men and 92 women; mean age, 58.8 6 18 years) over a 31⁄2-year period. One hundred fifty procedures were performed for diagnostic purposes, 28 procedures were performed for therapeutic purposes, and 77 procedures were performed for both diagnostic and therapeutic purposes. Based on the radiographic criteria, 152 effusions (60%) were small. Loculations were present in 76 patients (30%). Pneumothoraces occurred in 14 instances (5.4%), and chest tube drainage was required in 2 instances (0.78%). Hospitalization status, critical illness, effusion size or type, presence of loculations, operator, needle type, amount of fluid withdrawn, occurrence of dry tap, and type of thoracentesis were not associated with an increased frequency of pneumothorax. The only predictor variable demonstrating statistical significance was repeated thoracentesis. Conclusion: The results of a bivariate analysis suggest that pneumothorax following thoracentesis is a rare event that is not easily predictable when the procedure is performed by experienced operators in a controlled setting. (CHEST 1999; 116:134 –138) Key words: complications; pleural ultrasonography; pneumothorax; thoracentesis Abbreviation: PSPU 5 Pulmonary Special Procedures Unit
is among the most common iatroP neumothorax genic complications following invasive chest procedures. Thoracentesis-related pneumothorax was prospectively identified, for example, in 20% of 538 iatrogenic pneumothoraces in 13 institutions participating in a Department of Veterans Affairs cooperative study.1 Shortly thereafter, Despars et al2 reported that thoracentesis-related pneumothorax *From the Interventional Pulmonology Section, Pulmonary and Critical Care Medicine Division, University of California, San Diego Medical Center/Thornton Hospital, La Jolla, CA. Manuscript received September 30, 1999; revision accepted February 2, 1999. Correspondence to: Henri G. Colt, MD, FCCP, Associate Professor of Medicine, University of California, San Diego Medical Center/Thornton Hospital, 9310 Campus Point Dr 0976, La Jolla, CA 92037; e-mail: [email protected] 134
made up 28% of 106 iatrogenic pneumothoraces identified in a single institution over a 5-year period. Almost half of these patients required chest tubes. The authors of several reviews3–5 state that the incidence of thoracentesis-related pneumothorax varies between 3% and 19%. In fact, the frequency of thoracentesis-related pneumothorax in three large retrospective studies totaling 679 procedures was only 7.6%.6 – 8 Chest tubes were needed in 3.6% of all thoracenteses, representing almost half of all patients (48%) in whom pneumothorax occurred. Similar figures have been cited in at least seven recently published prospective studies9 –15 totaling 612 thoracenteses in 463 patients. In these studies, the overall frequency of thoracentesis-related pneumothorax was only 8.6% (53 pneumothoraces). Chest tubes Clinical Investigations
were necessary in 2.6% of all thoracenteses performed, corresponding to 16 pneumothoraces (30.1%) of all patients in whom pneumothorax occurred. Conventional wisdom would suggest, therefore, that thoracentesis is a safe procedure infrequently associated with adverse events that affect clinical management. Still, investigators strive to identify risk factors that, once controlled, could further decrease the incidence of thoracentesis-related complications. Interestingly, risk factors for pneumothorax in these studies are not consistently identified, but operator inexperience,9,16,17 the use of a large needle,7,9 therapeutic as opposed to diagnostic thoracentesis,10,12 the evacuation of a large volume of pleural fluid,7,10 the need for multiple needle passes,7,15 the presence of loculations,8 underlying neoplastic or chronic obstructive lung disease,17 and repeated thoracenteses in the same patient15 have each been described. Intrigued by the diversity of results that potentially link various patient-related and procedure-related risk factors to pneumothorax, we intentionally adopted a sophisticated approach to all patients referred to our unit for this procedure, in order to further investigate the relationship between risk factors and thoracentesis-related pneumothorax. Our experience with this approach forms the basis of this report.
Materials and Methods All patients referred to the Pulmonary Special Procedures Unit (PSPU) of the University of California, San Diego Medical Center for thoracentesis between January 1993 and June 1996 were eligible for this study. Patients with trapped lung or patients undergoing closed-needle pleural biopsy were excluded. Thoracentesis was performed with patients in the seated position. Routine vital signs were monitored. The presence or absence of loculations was confirmed using dynamic, real-time pleural ultrasonography (Ultra Mark 4 scanner; Advanced Technologies Laboratories; Bothell, WA). Scanning was done through the intercostal spaces posteriorly and laterally using a 3.0-MHz, long-focus mechanical transducer (Advanced Technologies Laboratories). Thoracentesis was performed, for the most part, by pulmonary physiciansin-training under the direct supervision of PSPU faculty. Chest radiographs were routinely obtained within 2 h after each procedure. The following patient-related and procedure-related predictor variables were noted prospectively: the inpatient or outpatient hospitalization status; whether the patient was critically ill and hospitalized in the ICU; the size of the effusion, based on a review of preprocedure chest radiographs (large if fluid extended to or above the dome of the hemidiaphragm, and small if fluid or thickening was confined to the costophrenic angle or below the level of the dome of the diaphragm); whether thoracentesis was warranted for diagnostic or therapeutic purposes; the physician performing the procedure; the thoracentesis needle that was used, based on random physician preferences and needle availability (a simple 22-gauge needle attached to a 60-mL syringe, a reusable 3-mm Boutin pleural puncture needle equipped with a
stopcock and blunt inner cannula, or a Cope inner trocar and hollow outer cannula); the indication for the procedure; whether thoracentesis was an initial or repeated procedure; whether loculations were present on sonography; the amount of fluid removed; whether the procedure was a dry tap; whether fluid was exudative or transudative; and the etiology of the pleural effusion. The major outcome variable, thoracentesis-related pneumothorax, was defined as new evidence of air in the pleural space on postprocedure chest radiographs in patients with or without clinical symptoms, but in the absence of known trapped lung. Other complications were noted as part of our ongoing quality assurance program, but the results were not used for statistical analysis. Dry taps, defined as an attempt at thoracentesis without the recovery of enough pleural fluid for laboratory analysis, were not considered complications. The data were analyzed at the procedure level (each patient could contribute more than one procedure) using the presence or absence of thoracentesis-related pneumothorax as the dichotomous outcome variable. A bivariate analysis using contingency tables was performed to examine associations between predictor variables and pneumothorax. x2 and associated p values were examined using p # 0.05 as the threshold for statistical significance.
Results Two hundred fifty-five consecutive procedures were performed on 205 patients referred to our PSPU for thoracentesis during this 42-month period. There were 113 men and 92 women (mean age, 58.8 6 18 years). Indications for thoracentesis were parapneumonic effusion (n 5 54), suspected empyema (n 5 8), effusion of other known benign cause (n 5 4), suspected malignant effusion (n 5 89), known malignant effusion (n 5 4), and effusion of unclear etiology (n 5 83). None of the patients underwent failed attempts at thoracentesis prior to referral. Thirty-two patients underwent more than one thoracentesis, having either two (n 5 25), three (n 5 6), or four (n 5 1) procedures. The etiologies for all 255 pleural effusions are presented in Table 1. Thoracentesis-related pneumothoraces were
Table 1—Etiologies of 255 Pleural Effusions in 205 Patients Undergoing Thoracentesis Etiology of Pleural Effusion
No. of Cases
Malignancy Uncomplicated parapneumonic Paramalignant Other benign etiologies* Congestive heart failure Dry tap/no etiology determined Complicated parapneumonic Empyema Unknown etiology
70 54 39 38 22 15 9 6 2
*Includes postsurgical sympathetic (n 5 9), postpulmonary thromboembolism surgery (n 5 7), tuberculosis (n 5 7), renal disease (n 5 7), collagen vascular or rheumatoid disease (n 5 6), pancreatitis (n 5 1), and hemothorax (n 5 1). CHEST / 116 / 1 / JULY, 1999
135
noted in 14 instances (5.4%) but required chest tube drainage in only 2 instances (0.78%). In both cases, the patients had chest pain and dyspnea shortly after the procedure. The outcome was satisfactory in each case after a short period of chest tube drainage. Other complications included vasovagal reactions (n 5 10) and subcutaneous hematoma (n 5 1). There were no deaths, hemothoraces, inadvertent liver or splenic lacerations, or cases of reexpansion pulmonary edema. No statistically significant associations were found between patient-related risk factors and the occurrence of thoracentesis (Table 2). One hundred fiftytwo procedures (60%) were performed primarily for diagnostic purposes. Although pneumothorax occurred more frequently in patients undergoing more than one thoracentesis (p , 0.05), none of the other procedure-related factors were significantly associated with the occurrence of pneumothorax (Table 3). In addition, after coding pneumothorax as a binary variable, with 1 equaling the presence of pneumothorax and 0 equaling no pneumothorax, a correlation was run between pneumothorax and the amount of fluid withdrawn. No statistical significance was found (r 5 0.3). Dry taps occurred on 15 occasions (overall frequency, 6.7%), but they were not associated with a greater incidence of pneumothorax. Discussion This prospective study provides additional evidence that thoracentesis, when performed in a controlled setting by experienced operators, is a very
Table 2—Patient-Related Factors Potentially Contributing to Pneumothorax Following Thoracentesis
Predictor Variables Hospitalization status Outpatient Inpatient Critical illness status Non-ICU ICU Effusion size Small Large Fluid type Transudate Exudate Not done Loculations present Loculations No loculations
136
No. of Procedures n 5 255
No. of Pneumothoraces (%) n 5 14
99 156
6 (6.0) 8 (5.1)
211 44
12 (5.7) 2 (4.5)
152 103
7 (4.6) 7 (6.8)
58 161 36
3 (5.1) 9 (5.6) 2 (5.6)
76 179
3 (3.9) 11 (6.1)
Table 3—Procedure-Related Factors Potentially Contributing to Pneumothorax Following Thoracentesis
Predictor Variables Physician Fellow Attending Needle type Needle/angiocatheter Boutin Cope Fluid removed No fluid (dry tap) Fluid removed Amount of fluid removed, mL ,60 60 to 350 350 to 1,000 .1,000 First or subsequent thoracentesis Initial Subsequent ($ 2) Type of procedure Diagnostic Therapeutic Both diagnostic and therapeutic
No. of No. of Procedures Pneumothoraces (%) n 5 255 n 5 14 194 61
10 (5.1) 4 (6.6)
100 84 71
5 (5.0) 4 (4.8) 5 (7.0)
15 240
1 (6.7) 13 (5.4)
59 60 60 61
3 (5.0) 2 (3.3) 5 (8.3) 3 (4.9)
215 40
9 (4.2) 5 (12.5)*
150 28 77
7 (4.7) 2 (7.1) 5 (6.5)
*Statistically significant difference at p , 0.05.
safe procedure associated with few complications. In fact, our pneumothorax rate (5.4%) confirms that this complication is infrequent and is seldom of clinical significance; chest tubes were required in only 0.78% of all procedures, a lower frequency than is reported by other investigators, particularly when considering that 60% of our effusions were small by radiographic criteria and 30% showed loculations on pleural ultrasonography. One limitation of this study, however, was our liberal use of pleural ultrasonography to control for the presence of loculations. It is unclear whether our use of pleural ultrasonography alone could explain the low incidence of pneumothorax or the infrequent need for tube thoracostomy. It is noteworthy that thoracentesis was always performed regardless of the results of the ultrasound examination, and that dry taps (n 5 15) were not associated with an increased incidence of pneumothorax. Although dynamic signs such as anechoic appearance, flapping movements of atelectatic lung, and the visualization of internal septa or intrapleural debris suggest a pleural effusion,18 –20 many pleural or chest wall abnormalities are hypoechoic, and some fluid collections appear as uniform echogenic structures.21–23 At the same time, pleural ultrasonography is not 100% specific, so dry taps are not always avoidable. Another potential limitation of our study was Clinical Investigations
(functionally speaking) that the procedures were performed at the attending level, because trainees performed the thoracentesis under direct faculty guidance, making the addressed issue of operator experience less clear. In fact, it is likely that operator skill was a major factor contributing to the low incidence of pneumothorax in our study. Because our complication rate was less than the pneumothorax rate (11 to 19%) described in studies10,12,14 in which house officers performed thoracentesis without faculty supervision, our results could be used to validate the findings of Bartter et al9 that thoracentesis is a low-risk procedure when performed by trained operators. The demonstrated safety of thoracentesis in a well-controlled environment also provides indirect evidence to support the argument of Doyle et al15 that routine chest radiography after thoracentesis is probably unwarranted. Our study is larger than the retrospective studies of Jenkins et al6 or Moore et al.8 It is also larger than each of the seven prospective studies9 –15 that addressed the risk for thoracentesis-related pneumothorax (Table 4). In contrast to the results presented by these and other investigators, we found no statistically significant association between the occurrence
of pneumothorax and the type of needle used, the size of the effusion, the amount of fluid drained, the presence of loculations, the type of thoracentesis, or experience of the operator. Our complication rate is insufficient, however, to provide the statistical power necessary to refute these previously demonstrated relationships between risk factors and pneumothorax. We cannot explain why pneumothorax occurred most frequently in patients undergoing multiple thoracenteses. The majority of these repeated thoracenteses were performed in order to evacuate recurrent or large pleural effusions. This group of patients did not differ in any other way (including etiology for pleural effusion) from the group undergoing a single procedure. This potential association has not been previously described. A few proposed causes for pneumothorax could include the use of a more aggressive approach to thoracentesis during repeated attempts, the patient predisposition due to chest wall thickness and configuration (cachexia vs hyperinflation or obesity), and the duration of the pleural effusion. Although a procedural technique-related explanation is possible, studies of pathophysiologic processes, including pleural pressure differentials
Table 4 —Summary of Prospective Studies Reporting Thoracentesis-Related Pneumothorax Study
No. of No. of No. of No. Needing No. of Dry Patients Procedures Pneumothoraces (%) Chest Tubes Taps (%)
Current study
205
255
14 (5.1)
2
Bartter et al9
33
50
2 (4)
0
Collins and Sahn10
86
129
15 (12)
5
Grodzin and Balk11
23
57
2 (3.5)
1
Grogan et al12
52
52
10 (19)
2
Yu et al13
25
25
1 (4)
Seneff et al14
91
125
14 (11)
3
Doyle et al15
110
174
9 (5.2)
4
1(4%)
Comments
15 (5.8)
3 different needles. All with ultrasound. Performed mostly by supervised pulmonary trainees. No correlation between potential risk factors and complications. 1 (2) Needle not described. No ultrasound. Performed by pulmonary trainees and faculty. Thoracentesis by experienced operators is low risk. 9 (7) Needle not described. No ultrasound. Performed mostly by medical house staff. Therapeutic taps, cough, and malignancy were principal risk factors for pneumothorax. 1 (1.75) Turkel needle-catheter. No ultrasound. Large freeflowing fluid only. Performed by single faculty operator. Safety of Turkel needle demonstrated. 3 (5.7) Comparative study of 2 needles with or without ultrasound. Large, free-flowing effusions only. Performed by medical house staff. Procedures under ultrasound guidance safest. 1 (4) Needle not specified. All with ultrasound. Risk factors not addressed. 16 (13) Needle-catheter with or without ultrasound. Performed mostly by medical house staff. Some major complications with dry taps. Risk factors included faulty technique and inadequate identification of landmarks. NA* Needle-catheter. No ultrasound. Risk factors included number of passes, aspiration of air, operator suspicion, history of radiation.
*NA 5 not available. CHEST / 116 / 1 / JULY, 1999
137
within the pleural space after repeated thoracenteses, are probably warranted. In summary, we cannot unquestionably refute the relationships between potential risk factors and pneumothorax that have been described by previous investigators. Nevertheless, the results from this prospective bivariate analysis suggest that thoracentesis-related pneumothorax is rare and not easily predictable when the procedures are performed under faculty supervision in a controlled environment. References 1 Light RW, Hara VS, Moritz TE. Iatrogenic pneumothorax: etiology and morbidity; results of a Department of Veterans Affairs cooperative study. Respiration 1992; 59:215–220 2 Despars JA, Sassoon CS, Light RW. Significance of iatrogenic pneumothoraces. Chest 1994; 105:1147–1150 3 Bartter T, Santarelli R, Akers SM, et al. The evaluation of pleural effusion. Chest 1994; 106:1209 –1214 4 American College of Physicians. Diagnostic thoracentesis and pleural biopsy in pleural effusions. Ann Intern Med 1985; 103:799 – 802 5 Sokolowski JW, Burgher LW, Jones FL, et al. Guidelines for thoracentesis and needle biopsy of the pleura: ATS position paper. Am Rev Respir Dis 1989; 140:257–258 6 Jenkins DW Jr, McKinney MK, Szpak MW, et al. Veres needle in the pleural space. South Med J 1983; 76:1383–1385 7 Raptopoulos V, Davis LM, Lee G, et al. Factors affecting the development of pneumothorax associated with thoracentesis. AJR Am J Roentgenol 1991; 156:917–920 8 Moore PV, Mueller PR, Simeone JF, et al. Sonographic guidance in diagnostic and therapeutic interventions in the pleural space. AJR Am J Roentgenol 1987; 149:1–5 9 Bartter T, Mayo PD, Pratter MR, et al. Lower risk and higher yield for thoracentesis when performed by experienced operators. Chest 1993; 103:1873–1876
138
10 Collins TR, Sahn SA. Thoracentesis: clinical value, complications, technical problems, and patient experience. Chest 1987; 91:817– 822 11 Grodzin CJ, Balk RA. Indwelling small pleural catheter needle thoracentesis in the management of large pleural effusions. Chest 1997; 111:981–988 12 Grogan DR, Irwin RS, Channick R, et al. Complications associated with thoracentesis: a prospective randomized study comparing three different methods. Arch Intern Med 1990; 150:873– 877 13 Yu CJ, Yang, PC, Chang DB, et al. Diagnostic and therapeutic use of chest sonography: value in critically ill patients. AJR Am J Roentgenol 1992; 159:695–701 14 Seneff MG, Corwin RW, Gold LH, et al. Complications associated with thoracentesis. Chest 1986; 90:97–100 15 Doyle JJ, Hnatiuk OW, Torrington KG, et al. Necessity of routine chest roentgenography after thoracentesis. Ann Intern Med 1996; 124:816 – 820 16 Swinburne AJ, Bixby AJ, Lee D, et al. Pneumothorax after thoracentesis. Arch Intern Med 1991; 151:2095–2096 17 Brandstetter RD, Karetzky M, Rastogi R, et al. Pneumothorax after thoracentesis in chronic obstructive pulmonary disease. Heart Lung 1994; 23:67–70 18 Eibenberger KL, Dock WI, Ammann ME, et al. Quantification of pleural effusions: sonography versus radiography. Radiology 1994; 191:681– 684 19 Kohan JM, Poe RH, Israel RH, et al. Value of chest ultrasonography versus decubitus roentgenography for thoracentesis. Am Rev Respir Dis 1986; 133:1124 –1126 20 Lomas DJ, Padley SG, Flower CDR. The sonographic appearances of pleural fluid. Br J Radiol 1993; 66:619 – 624 21 McLoud TC, Flower CDR. Imaging the pleura: sonography, CT, and MR imaging. AJR Am J Roentgenol 1991; 156:1145– 1153 22 Weingardt JP, Guico RR, Nemcek AA, et al. Ultrasound findings following failed, clinically directed thoracenteses. Clin Ultrasound 1994; 22:419 – 426 23 Wu RG, Yuan A, Liaw YS, et al. Image comparison of real-time gray-scale ultrasound and color Doppler ultrasound for use in diagnosis of minimal pleural effusion. Am J Respir Crit Care Med 1994; 150:510 –514
Clinical Investigations
Objective: To evaluate patient-related and procedure-related risk factors for thoracentesisrelated pneumothorax. Design: Prospective, nonrandomized cohort study. Setting: Pulmonary Special Procedures Unit of a university medical center. Methods: Thoracentesis using either a 22-gauge, a Boutin, or a Cope needle (depending on availability and operator preference) was performed by the pulmonary faculty or by pulmonary physicians-in-training under faculty supervision. In order to control for effusion size and the presence of loculations, chest radiography and pleural ultrasonography were performed prior to each thoracentesis. Potential patient-related and procedure-related risk factors for pneumothorax were analyzed at the procedure level using the presence or absence of pneumothorax on the postprocedure chest radiograph as the sole outcome variable. Results: Two hundred fifty-five thoracenteses were performed in 205 adult patients (113 men and 92 women; mean age, 58.8 6 18 years) over a 31⁄2-year period. One hundred fifty procedures were performed for diagnostic purposes, 28 procedures were performed for therapeutic purposes, and 77 procedures were performed for both diagnostic and therapeutic purposes. Based on the radiographic criteria, 152 effusions (60%) were small. Loculations were present in 76 patients (30%). Pneumothoraces occurred in 14 instances (5.4%), and chest tube drainage was required in 2 instances (0.78%). Hospitalization status, critical illness, effusion size or type, presence of loculations, operator, needle type, amount of fluid withdrawn, occurrence of dry tap, and type of thoracentesis were not associated with an increased frequency of pneumothorax. The only predictor variable demonstrating statistical significance was repeated thoracentesis. Conclusion: The results of a bivariate analysis suggest that pneumothorax following thoracentesis is a rare event that is not easily predictable when the procedure is performed by experienced operators in a controlled setting. (CHEST 1999; 116:134 –138) Key words: complications; pleural ultrasonography; pneumothorax; thoracentesis Abbreviation: PSPU 5 Pulmonary Special Procedures Unit
is among the most common iatroP neumothorax genic complications following invasive chest procedures. Thoracentesis-related pneumothorax was prospectively identified, for example, in 20% of 538 iatrogenic pneumothoraces in 13 institutions participating in a Department of Veterans Affairs cooperative study.1 Shortly thereafter, Despars et al2 reported that thoracentesis-related pneumothorax *From the Interventional Pulmonology Section, Pulmonary and Critical Care Medicine Division, University of California, San Diego Medical Center/Thornton Hospital, La Jolla, CA. Manuscript received September 30, 1999; revision accepted February 2, 1999. Correspondence to: Henri G. Colt, MD, FCCP, Associate Professor of Medicine, University of California, San Diego Medical Center/Thornton Hospital, 9310 Campus Point Dr 0976, La Jolla, CA 92037; e-mail: [email protected] 134
made up 28% of 106 iatrogenic pneumothoraces identified in a single institution over a 5-year period. Almost half of these patients required chest tubes. The authors of several reviews3–5 state that the incidence of thoracentesis-related pneumothorax varies between 3% and 19%. In fact, the frequency of thoracentesis-related pneumothorax in three large retrospective studies totaling 679 procedures was only 7.6%.6 – 8 Chest tubes were needed in 3.6% of all thoracenteses, representing almost half of all patients (48%) in whom pneumothorax occurred. Similar figures have been cited in at least seven recently published prospective studies9 –15 totaling 612 thoracenteses in 463 patients. In these studies, the overall frequency of thoracentesis-related pneumothorax was only 8.6% (53 pneumothoraces). Chest tubes Clinical Investigations
were necessary in 2.6% of all thoracenteses performed, corresponding to 16 pneumothoraces (30.1%) of all patients in whom pneumothorax occurred. Conventional wisdom would suggest, therefore, that thoracentesis is a safe procedure infrequently associated with adverse events that affect clinical management. Still, investigators strive to identify risk factors that, once controlled, could further decrease the incidence of thoracentesis-related complications. Interestingly, risk factors for pneumothorax in these studies are not consistently identified, but operator inexperience,9,16,17 the use of a large needle,7,9 therapeutic as opposed to diagnostic thoracentesis,10,12 the evacuation of a large volume of pleural fluid,7,10 the need for multiple needle passes,7,15 the presence of loculations,8 underlying neoplastic or chronic obstructive lung disease,17 and repeated thoracenteses in the same patient15 have each been described. Intrigued by the diversity of results that potentially link various patient-related and procedure-related risk factors to pneumothorax, we intentionally adopted a sophisticated approach to all patients referred to our unit for this procedure, in order to further investigate the relationship between risk factors and thoracentesis-related pneumothorax. Our experience with this approach forms the basis of this report.
Materials and Methods All patients referred to the Pulmonary Special Procedures Unit (PSPU) of the University of California, San Diego Medical Center for thoracentesis between January 1993 and June 1996 were eligible for this study. Patients with trapped lung or patients undergoing closed-needle pleural biopsy were excluded. Thoracentesis was performed with patients in the seated position. Routine vital signs were monitored. The presence or absence of loculations was confirmed using dynamic, real-time pleural ultrasonography (Ultra Mark 4 scanner; Advanced Technologies Laboratories; Bothell, WA). Scanning was done through the intercostal spaces posteriorly and laterally using a 3.0-MHz, long-focus mechanical transducer (Advanced Technologies Laboratories). Thoracentesis was performed, for the most part, by pulmonary physiciansin-training under the direct supervision of PSPU faculty. Chest radiographs were routinely obtained within 2 h after each procedure. The following patient-related and procedure-related predictor variables were noted prospectively: the inpatient or outpatient hospitalization status; whether the patient was critically ill and hospitalized in the ICU; the size of the effusion, based on a review of preprocedure chest radiographs (large if fluid extended to or above the dome of the hemidiaphragm, and small if fluid or thickening was confined to the costophrenic angle or below the level of the dome of the diaphragm); whether thoracentesis was warranted for diagnostic or therapeutic purposes; the physician performing the procedure; the thoracentesis needle that was used, based on random physician preferences and needle availability (a simple 22-gauge needle attached to a 60-mL syringe, a reusable 3-mm Boutin pleural puncture needle equipped with a
stopcock and blunt inner cannula, or a Cope inner trocar and hollow outer cannula); the indication for the procedure; whether thoracentesis was an initial or repeated procedure; whether loculations were present on sonography; the amount of fluid removed; whether the procedure was a dry tap; whether fluid was exudative or transudative; and the etiology of the pleural effusion. The major outcome variable, thoracentesis-related pneumothorax, was defined as new evidence of air in the pleural space on postprocedure chest radiographs in patients with or without clinical symptoms, but in the absence of known trapped lung. Other complications were noted as part of our ongoing quality assurance program, but the results were not used for statistical analysis. Dry taps, defined as an attempt at thoracentesis without the recovery of enough pleural fluid for laboratory analysis, were not considered complications. The data were analyzed at the procedure level (each patient could contribute more than one procedure) using the presence or absence of thoracentesis-related pneumothorax as the dichotomous outcome variable. A bivariate analysis using contingency tables was performed to examine associations between predictor variables and pneumothorax. x2 and associated p values were examined using p # 0.05 as the threshold for statistical significance.
Results Two hundred fifty-five consecutive procedures were performed on 205 patients referred to our PSPU for thoracentesis during this 42-month period. There were 113 men and 92 women (mean age, 58.8 6 18 years). Indications for thoracentesis were parapneumonic effusion (n 5 54), suspected empyema (n 5 8), effusion of other known benign cause (n 5 4), suspected malignant effusion (n 5 89), known malignant effusion (n 5 4), and effusion of unclear etiology (n 5 83). None of the patients underwent failed attempts at thoracentesis prior to referral. Thirty-two patients underwent more than one thoracentesis, having either two (n 5 25), three (n 5 6), or four (n 5 1) procedures. The etiologies for all 255 pleural effusions are presented in Table 1. Thoracentesis-related pneumothoraces were
Table 1—Etiologies of 255 Pleural Effusions in 205 Patients Undergoing Thoracentesis Etiology of Pleural Effusion
No. of Cases
Malignancy Uncomplicated parapneumonic Paramalignant Other benign etiologies* Congestive heart failure Dry tap/no etiology determined Complicated parapneumonic Empyema Unknown etiology
70 54 39 38 22 15 9 6 2
*Includes postsurgical sympathetic (n 5 9), postpulmonary thromboembolism surgery (n 5 7), tuberculosis (n 5 7), renal disease (n 5 7), collagen vascular or rheumatoid disease (n 5 6), pancreatitis (n 5 1), and hemothorax (n 5 1). CHEST / 116 / 1 / JULY, 1999
135
noted in 14 instances (5.4%) but required chest tube drainage in only 2 instances (0.78%). In both cases, the patients had chest pain and dyspnea shortly after the procedure. The outcome was satisfactory in each case after a short period of chest tube drainage. Other complications included vasovagal reactions (n 5 10) and subcutaneous hematoma (n 5 1). There were no deaths, hemothoraces, inadvertent liver or splenic lacerations, or cases of reexpansion pulmonary edema. No statistically significant associations were found between patient-related risk factors and the occurrence of thoracentesis (Table 2). One hundred fiftytwo procedures (60%) were performed primarily for diagnostic purposes. Although pneumothorax occurred more frequently in patients undergoing more than one thoracentesis (p , 0.05), none of the other procedure-related factors were significantly associated with the occurrence of pneumothorax (Table 3). In addition, after coding pneumothorax as a binary variable, with 1 equaling the presence of pneumothorax and 0 equaling no pneumothorax, a correlation was run between pneumothorax and the amount of fluid withdrawn. No statistical significance was found (r 5 0.3). Dry taps occurred on 15 occasions (overall frequency, 6.7%), but they were not associated with a greater incidence of pneumothorax. Discussion This prospective study provides additional evidence that thoracentesis, when performed in a controlled setting by experienced operators, is a very
Table 2—Patient-Related Factors Potentially Contributing to Pneumothorax Following Thoracentesis
Predictor Variables Hospitalization status Outpatient Inpatient Critical illness status Non-ICU ICU Effusion size Small Large Fluid type Transudate Exudate Not done Loculations present Loculations No loculations
136
No. of Procedures n 5 255
No. of Pneumothoraces (%) n 5 14
99 156
6 (6.0) 8 (5.1)
211 44
12 (5.7) 2 (4.5)
152 103
7 (4.6) 7 (6.8)
58 161 36
3 (5.1) 9 (5.6) 2 (5.6)
76 179
3 (3.9) 11 (6.1)
Table 3—Procedure-Related Factors Potentially Contributing to Pneumothorax Following Thoracentesis
Predictor Variables Physician Fellow Attending Needle type Needle/angiocatheter Boutin Cope Fluid removed No fluid (dry tap) Fluid removed Amount of fluid removed, mL ,60 60 to 350 350 to 1,000 .1,000 First or subsequent thoracentesis Initial Subsequent ($ 2) Type of procedure Diagnostic Therapeutic Both diagnostic and therapeutic
No. of No. of Procedures Pneumothoraces (%) n 5 255 n 5 14 194 61
10 (5.1) 4 (6.6)
100 84 71
5 (5.0) 4 (4.8) 5 (7.0)
15 240
1 (6.7) 13 (5.4)
59 60 60 61
3 (5.0) 2 (3.3) 5 (8.3) 3 (4.9)
215 40
9 (4.2) 5 (12.5)*
150 28 77
7 (4.7) 2 (7.1) 5 (6.5)
*Statistically significant difference at p , 0.05.
safe procedure associated with few complications. In fact, our pneumothorax rate (5.4%) confirms that this complication is infrequent and is seldom of clinical significance; chest tubes were required in only 0.78% of all procedures, a lower frequency than is reported by other investigators, particularly when considering that 60% of our effusions were small by radiographic criteria and 30% showed loculations on pleural ultrasonography. One limitation of this study, however, was our liberal use of pleural ultrasonography to control for the presence of loculations. It is unclear whether our use of pleural ultrasonography alone could explain the low incidence of pneumothorax or the infrequent need for tube thoracostomy. It is noteworthy that thoracentesis was always performed regardless of the results of the ultrasound examination, and that dry taps (n 5 15) were not associated with an increased incidence of pneumothorax. Although dynamic signs such as anechoic appearance, flapping movements of atelectatic lung, and the visualization of internal septa or intrapleural debris suggest a pleural effusion,18 –20 many pleural or chest wall abnormalities are hypoechoic, and some fluid collections appear as uniform echogenic structures.21–23 At the same time, pleural ultrasonography is not 100% specific, so dry taps are not always avoidable. Another potential limitation of our study was Clinical Investigations
(functionally speaking) that the procedures were performed at the attending level, because trainees performed the thoracentesis under direct faculty guidance, making the addressed issue of operator experience less clear. In fact, it is likely that operator skill was a major factor contributing to the low incidence of pneumothorax in our study. Because our complication rate was less than the pneumothorax rate (11 to 19%) described in studies10,12,14 in which house officers performed thoracentesis without faculty supervision, our results could be used to validate the findings of Bartter et al9 that thoracentesis is a low-risk procedure when performed by trained operators. The demonstrated safety of thoracentesis in a well-controlled environment also provides indirect evidence to support the argument of Doyle et al15 that routine chest radiography after thoracentesis is probably unwarranted. Our study is larger than the retrospective studies of Jenkins et al6 or Moore et al.8 It is also larger than each of the seven prospective studies9 –15 that addressed the risk for thoracentesis-related pneumothorax (Table 4). In contrast to the results presented by these and other investigators, we found no statistically significant association between the occurrence
of pneumothorax and the type of needle used, the size of the effusion, the amount of fluid drained, the presence of loculations, the type of thoracentesis, or experience of the operator. Our complication rate is insufficient, however, to provide the statistical power necessary to refute these previously demonstrated relationships between risk factors and pneumothorax. We cannot explain why pneumothorax occurred most frequently in patients undergoing multiple thoracenteses. The majority of these repeated thoracenteses were performed in order to evacuate recurrent or large pleural effusions. This group of patients did not differ in any other way (including etiology for pleural effusion) from the group undergoing a single procedure. This potential association has not been previously described. A few proposed causes for pneumothorax could include the use of a more aggressive approach to thoracentesis during repeated attempts, the patient predisposition due to chest wall thickness and configuration (cachexia vs hyperinflation or obesity), and the duration of the pleural effusion. Although a procedural technique-related explanation is possible, studies of pathophysiologic processes, including pleural pressure differentials
Table 4 —Summary of Prospective Studies Reporting Thoracentesis-Related Pneumothorax Study
No. of No. of No. of No. Needing No. of Dry Patients Procedures Pneumothoraces (%) Chest Tubes Taps (%)
Current study
205
255
14 (5.1)
2
Bartter et al9
33
50
2 (4)
0
Collins and Sahn10
86
129
15 (12)
5
Grodzin and Balk11
23
57
2 (3.5)
1
Grogan et al12
52
52
10 (19)
2
Yu et al13
25
25
1 (4)
Seneff et al14
91
125
14 (11)
3
Doyle et al15
110
174
9 (5.2)
4
1(4%)
Comments
15 (5.8)
3 different needles. All with ultrasound. Performed mostly by supervised pulmonary trainees. No correlation between potential risk factors and complications. 1 (2) Needle not described. No ultrasound. Performed by pulmonary trainees and faculty. Thoracentesis by experienced operators is low risk. 9 (7) Needle not described. No ultrasound. Performed mostly by medical house staff. Therapeutic taps, cough, and malignancy were principal risk factors for pneumothorax. 1 (1.75) Turkel needle-catheter. No ultrasound. Large freeflowing fluid only. Performed by single faculty operator. Safety of Turkel needle demonstrated. 3 (5.7) Comparative study of 2 needles with or without ultrasound. Large, free-flowing effusions only. Performed by medical house staff. Procedures under ultrasound guidance safest. 1 (4) Needle not specified. All with ultrasound. Risk factors not addressed. 16 (13) Needle-catheter with or without ultrasound. Performed mostly by medical house staff. Some major complications with dry taps. Risk factors included faulty technique and inadequate identification of landmarks. NA* Needle-catheter. No ultrasound. Risk factors included number of passes, aspiration of air, operator suspicion, history of radiation.
*NA 5 not available. CHEST / 116 / 1 / JULY, 1999
137
within the pleural space after repeated thoracenteses, are probably warranted. In summary, we cannot unquestionably refute the relationships between potential risk factors and pneumothorax that have been described by previous investigators. Nevertheless, the results from this prospective bivariate analysis suggest that thoracentesis-related pneumothorax is rare and not easily predictable when the procedures are performed under faculty supervision in a controlled environment. References 1 Light RW, Hara VS, Moritz TE. Iatrogenic pneumothorax: etiology and morbidity; results of a Department of Veterans Affairs cooperative study. Respiration 1992; 59:215–220 2 Despars JA, Sassoon CS, Light RW. Significance of iatrogenic pneumothoraces. Chest 1994; 105:1147–1150 3 Bartter T, Santarelli R, Akers SM, et al. The evaluation of pleural effusion. Chest 1994; 106:1209 –1214 4 American College of Physicians. Diagnostic thoracentesis and pleural biopsy in pleural effusions. Ann Intern Med 1985; 103:799 – 802 5 Sokolowski JW, Burgher LW, Jones FL, et al. Guidelines for thoracentesis and needle biopsy of the pleura: ATS position paper. Am Rev Respir Dis 1989; 140:257–258 6 Jenkins DW Jr, McKinney MK, Szpak MW, et al. Veres needle in the pleural space. South Med J 1983; 76:1383–1385 7 Raptopoulos V, Davis LM, Lee G, et al. Factors affecting the development of pneumothorax associated with thoracentesis. AJR Am J Roentgenol 1991; 156:917–920 8 Moore PV, Mueller PR, Simeone JF, et al. Sonographic guidance in diagnostic and therapeutic interventions in the pleural space. AJR Am J Roentgenol 1987; 149:1–5 9 Bartter T, Mayo PD, Pratter MR, et al. Lower risk and higher yield for thoracentesis when performed by experienced operators. Chest 1993; 103:1873–1876
138
10 Collins TR, Sahn SA. Thoracentesis: clinical value, complications, technical problems, and patient experience. Chest 1987; 91:817– 822 11 Grodzin CJ, Balk RA. Indwelling small pleural catheter needle thoracentesis in the management of large pleural effusions. Chest 1997; 111:981–988 12 Grogan DR, Irwin RS, Channick R, et al. Complications associated with thoracentesis: a prospective randomized study comparing three different methods. Arch Intern Med 1990; 150:873– 877 13 Yu CJ, Yang, PC, Chang DB, et al. Diagnostic and therapeutic use of chest sonography: value in critically ill patients. AJR Am J Roentgenol 1992; 159:695–701 14 Seneff MG, Corwin RW, Gold LH, et al. Complications associated with thoracentesis. Chest 1986; 90:97–100 15 Doyle JJ, Hnatiuk OW, Torrington KG, et al. Necessity of routine chest roentgenography after thoracentesis. Ann Intern Med 1996; 124:816 – 820 16 Swinburne AJ, Bixby AJ, Lee D, et al. Pneumothorax after thoracentesis. Arch Intern Med 1991; 151:2095–2096 17 Brandstetter RD, Karetzky M, Rastogi R, et al. Pneumothorax after thoracentesis in chronic obstructive pulmonary disease. Heart Lung 1994; 23:67–70 18 Eibenberger KL, Dock WI, Ammann ME, et al. Quantification of pleural effusions: sonography versus radiography. Radiology 1994; 191:681– 684 19 Kohan JM, Poe RH, Israel RH, et al. Value of chest ultrasonography versus decubitus roentgenography for thoracentesis. Am Rev Respir Dis 1986; 133:1124 –1126 20 Lomas DJ, Padley SG, Flower CDR. The sonographic appearances of pleural fluid. Br J Radiol 1993; 66:619 – 624 21 McLoud TC, Flower CDR. Imaging the pleura: sonography, CT, and MR imaging. AJR Am J Roentgenol 1991; 156:1145– 1153 22 Weingardt JP, Guico RR, Nemcek AA, et al. Ultrasound findings following failed, clinically directed thoracenteses. Clin Ultrasound 1994; 22:419 – 426 23 Wu RG, Yuan A, Liaw YS, et al. Image comparison of real-time gray-scale ultrasound and color Doppler ultrasound for use in diagnosis of minimal pleural effusion. Am J Respir Crit Care Med 1994; 150:510 –514
Clinical Investigations