Methods

Top Methods Results Discussion Author & Article Info References

Procedure and Data Collection

All inpatients 18 years of age or older having thoracentesis in the Internal Medicine Service at Walter Reed Army Medical Center were eligible to participate in this study. All patients gave consent. Thoracentesis was done using standard operating procedure with a needle or angiocath and syringe or with a prepackaged needle-catheter tray (Teflon catheter with 14 gauge x 15 cm needle, Kendall Curity, Mansfield, Massachusetts), according to operator preference. After the procedure, each operator completed a preprinted procedure note. The variables recorded are listed in the Appendix. After thoracentesis, patients had portable anteroposterior or standard posteroanterior and lateral chest roentgenography. Chart review, completed by the investigators within 72 hours after the procedure, identified any further complications that had occurred in the interim. Patients were excluded if they were younger than 18 years of age, did not have a procedure note before chest roentgenography, did not have a chest roentgenogram within 4 hours of thoracentesis, had a concurrent pleural biopsy, had thoracentesis assisted by ultrasonography, or were using mechanical ventilation. Several thoracenteses could be done on one patient if they were done on different dates. In six cases, the roentgenogram obtained immediately after the procedure was not formally interpreted by the radiology department and could not be located for our review. These six films were interpreted only by the house officer, but subsequent roentgenograms showed no evidence of pneumothorax. All other films were available for interpretation.

Data Analysis

Exploratory analysis comparing the demographic and procedural characteristics of patients having a pneumothorax with those of patients not having a pneumothorax was done using the Wilcoxon ranksum test for continuous and ordinal variables and the Fisher exact test (two-tailed) for nominal data. We did not adjust for the number of univariate statistical tests done. For demographic characteristics, the sampling unit for analysis was the patient's first tap (n = 110). We analyzed procedural characteristics using the patient's first tap (n = 110) and all taps that had similar results (n = 174). Relative risk ratios were calculated as the incidence of pneumothorax in the group at risk divided by the incidence of pneumothorax in the group without risk factors. If the incidence of pneumothorax was 0, we added 0.5 to each cell and estimated the ratios. We calculated 95% CIs for each relative risk [2] and proportion. Data were analyzed using SPS 5.0 for Windows (SPS, Chicago, Illinois).

Results

Top Methods Results Discussion Author & Article Info References

We analyzed 174 thoracenteses done on 110 patients admitted to the Walter Reed Army Medical Center Department of Medicine between March 1991 and June 1993. These were approximately 95% of all thoracenteses done during the study period. The study sample consisted of 43 women and 67 men with a mean age of 62.4 ± 13.2 years.

Patients had either 1 (n = 71), 2 (n = 24), 3 (n = 10), 4 (n = 3), 6 (n = 1), or 7 (n = 1) thoracenteses. Nine pneumothoraces occurred in this series of 174 taps, for an incidence rate of 5.2% (CI, 2.4% to 9.6%). Among the 110 patients, six pneumothoraces occurred on the first tap (incidence of 5.5%; CI, 2.0% to 11.5%). Of the 39 patients having a second tap, 2 had pneumothoraces, for an incidence rate of 5.1% (CI, 0.6% to 17.3%). On the third tap, 1 of 15 patients had a pneumothorax, for an incidence rate of 6.7% (CI, 0.1% to 31.9%). One patient had pneumothoraces on both his first and second thoracenteses, which were done 6 months apart.

The chest roentgenogram obtained immediately after the procedure identified eight of the nine pneumothoraces. Pneumothorax was suspected in five of the eight instances; tube thoracostomy was done in four of the five cases. The unsuspected pneumothoraces diagnosed in the other three cases were small (estimated by the radiologist to be less than 20%). One resolved spontaneously, the second was treated with a chest tube only to resolve a loculated effusion, and the third was untreated because the severely ill patient died approximately 6 hours after the procedure. An autopsy ruled out tension pneumothorax as the cause of death, and, because of the severity of the underlying illness, the pneumothorax was not believed to have contributed to the patient's death. In the one case in which roentgenography done immediately after the procedure did not identify pneumothorax, a moderately sized (50%) ipsilateral pneumothorax was found on a chest roentgenogram obtained 3 days later. To prevent further enlargement, tube thoracostomy was done at the request of the attending physician.

The pertinent characteristics of patients developing pneumothoraces are shown in Table 1. Correlations of patient and procedure characteristics with pneumothorax are shown in Table 2. Three procedure variables and one patient variable were strongly associated with the occurrence of pneumothorax. During the first thoracentesis (n = 110), aspiration of air occurred in 13 patients, and pneumothorax developed in 4 (relative risk ratio, 14.9; CI, 3.0 to 73.6). Patients who had more than one pass of the thoracentesis needle had an increased risk for developing pneumothorax (relative risk ratio, 5.6; CI, 1.1 to 28.9). Operator suspicion and existence of pneumothorax were also highly associated (relative risk ratio, 42.0; CI, 9.9 to 177.4). Five of the eight pneumothoraces confirmed by immediate chest roentgenography were suspected. In four of the five, the operator based suspicion on aspiration of air during the procedure. In the fifth case, air was audibly entrained through the catheter by the patient because of poor compliance with instructions. Finally, pneumothorax developed in 2 of 5 patients with a history of thoracic radiation therapy (relative risk ratio, 10.5; CI, 2.5 to 44.4).

If we combined these criteria in our sample, 110 attempts would not have required roentgenograms, and 109 (99.1%; CI, 95.0% to 99.9%) would have been negative for pneumothorax. Among the 64 attempts requiring roentgenograms, pneumothorax would have been diagnosed in eight cases (12.5%; CI, 5.5% to 23.2%). Vital signs and physical examination findings were infrequently documented by the operators and therefore could not be interpreted.

Two hemothoraces (1.2%) and two subcutaneous hematomas occurred in the study sample. Neither hemothorax was identified by roentgenography done immediately after the procedure; these diagnoses were made by evaluating pleural fluid cell counts. Both patients required chest tube drainage.

Discussion

Top Methods Results Discussion Author & Article Info References

Many articles have documented the various complications that occur after thoracentesis, but few have specifically evaluated or commented on the need for routine chest roentgenography after thoracentesis. In 1983, Collins and Sahn [3] reported their experience with complications after thoracentesis. They prospectively evaluated 74 hospitalized patients having thoracentesis and found a 12% incidence of pneumothorax. Routine chest roentgenograms obtained after thoracentesis showed four small "clinically insignificant" pneumothoraces not expected by the operators. None of these patients required further intervention. Collins and Sahn concluded their abstract by stating that "routine post thoracentesis chest x-rays are unnecessary unless complications are suspected clinically" [3]. In 1989, the American Thoracic Society [1] published guidelines for thoracentesis and pleural biopsy and recommended that "a chest film should be performed after therapeutic thoracentesis in most instances."

The necessity of chest roentgenography after thoracentesis was not specifically addressed in the medical literature again until Gerardi and colleagues [4] reported their retrospective chart review of all thoracenteses done during a 1-year period at a large community hospital. Among 134 procedures, they found a 7.5% incidence of pneumothorax. Three patients were asymptomatic and required no further intervention. Four of the seven patients with pneumothorax had new or increased dyspnea and required tube thoracostomy. In their conclusion, Gerardi and colleagues questioned the necessity of obtaining routine chest roentgenograms after thoracentesis in hospitalized patients who remained free of symptoms.

The incidences of pneumothorax (5.2%; CI, 2.4% to 9.6%) and hemothorax (1.2%; CI, 0.1% to 4.1%) in our study are similar to those in other published studies [4-8]. However, the presence of pneumothorax was not predicted by the development of new symptoms. Only two of the nine patients developed new symptoms during or after the procedure. Therefore, symptoms alone could not predict the presence of pneumothorax or the need for intervention, although the development of new symptoms should never be ignored.

Although other investigators have reported associations between pneumothorax and cancer and cough [4], needle-catheter technique [9], volume of aspirated fluid and needle size [10], clinical indication (diagnostic compared with therapeutic) [4, 9-11], and operator experience [4, 7, 11, 12], we found four factors associated with pneumothorax. However, because of the small number of events in our sample, we cannot determine whether each factor independently predicts the incidence of pneumothorax. Further research involving more patients is needed.

Because the incidences of both pneumothorax and associated morbidity after thoracentesis in our study were low and because a chest roentgenogram obtained immediately after the procedure may not identify all serious complications associated with thoracentesis, indiscriminate use of chest roentgenography after thoracentesis does not appear warranted. As a conservative approach, we recommend ordering a chest roentgenogram if the patient entrains air during the procedure; if the physician aspirates air back; if the patient has more than one pass, has new symptoms indicative of pneumothorax, or has had previous radiation therapy to the chest; or if pneumothorax is suspected by the operator for any other reason. If these criteria had been used in our sample, 110 attempts would not have required roentgenograms; 109 of those would have been negative for pneumothorax.

We must point out that for the 109 attempts that would not have required roentgenograms and were negative for pneumothorax, the lower limit of the CI is 95.0%. This suggests that the frequency of pneumothorax in this group could be as high as 5%; but even this number may be acceptably low given the minimal consequences of unsuspected pneumothorax. When the data from our study are combined with data from Collins and colleagues [3] and Gerardi and colleagues [4], no patient with unsuspected pneumothorax had serious clinical consequences as a result of their pneumothoraces. Ultimately, the individual clinician must decide what frequency of unsuspected pneumothorax is acceptable. A larger study that more precisely defines the frequency of unsuspected pneumothorax and confirms that the clinical consequences are minimal is needed.

Our study of hospitalized patients shows that the overall risk for significant morbidity after thoracentesis is low, that the chest roentgenogram obtained immediately after the procedure does not identify all serious complications associated with thoracentesis, and that the indiscriminate use of chest roentgenography after thoracentesis does not appreciably alter patient management. The absence of pneumothorax after thoracentesis can be predicted in hospitalized patients who are clinically stable, have not previously received chest irradiation, have had only one pass at thoracentesis attempted without the aspiration of any air, and have no other indications of pneumothorax.

Appendix

Variables recorded during thoracentesis were the postgraduate year of the operator, operator experience, decubitus film results, results from the left side compared with those from the right side, clinical indication (diagnostic compared with therapeutic), needle gauge, needle-catheter use, number of needle passes required, volume of aspirated fluid, time required, air aspiration, cough during procedure, signs and symptoms, vital signs (both before and after procedure), complication suspected, age, sex, smoking history (pack-years), cancer history, thoracic cancer history, previous thoracic surgery, previous thoracic radiation, and existence of active pneumonia.

Presented in part in November 1994 at the annual meeting of the American College of Physicians, U.S. Army Region, Orlando, Florida. The opinions contained herein solely represent the views of the authors and are not to be construed as representing the views of the Department of Defense or the Department of the Army.