Population-Based Estimates of Transbronchial Lung Biopsy Utilization and Complications

Abstract

Background

Little is known about trends in the utilization or complication rates of transbronchial lung biopsy, particularly in community hospitals.

Methods

We used the Healthcare Cost and Utilization Project Florida State Inpatient and State Ambulatory Surgical Databases to assess trends in transbronchial lung biopsy utilization in adults from 2000 to 2009. We subsequently calculated population based estimates of complications associated with transbronchial lung biopsy (iatrogenic pneumothorax and procedure-related hemorrhage) and identified characteristics associated with complications.

Results

From 2000 to 2009, the age-adjusted rate of transbronchial biopsies per 100,000 adults in Florida decreased by 25% from 74 to 55 (p<0.0001), despite stability in the overall utilization of bronchoscopy. Analysis of 82,059 procedures revealed that complications associated with transbronchial biopsy were uncommon and stable over the study period, with 0.97% (95% CI 0.94–1.01%) of procedures complicated by pneumothorax, 0.55% (95% CI 0.52–0.58%) by pneumothorax requiring chest tube placement, and 0.58% (95% CI 0.55–0.61%) by procedure-related hemorrhage. Patients with COPD (OR 1.51, 95% CI 1.31–1.75) and women (OR 1.32, 95% CI 1.15–1.52) were at increased risk for pneumothorax, while renal failure (OR 2.85, 95% CI 2.10– 3.87), cirrhosis (OR 2.31, 95% CI 1.18– 4.52), older age (OR 1.17, 95% CI 1.09–1.25) and female sex (OR 1.40, 95% CI 1.17–1.68) were associated with higher risk of procedure-related hemorrhage.

Conclusions

Utilization of transbronchial lung biopsy is decreasing relative to the overall use of bronchoscopy. Nevertheless, it remains a safe procedure with low risk of complications.

INTRODUCTION

According to professional guidelines, bronchoscopy with transbronchial lung biopsy may be considered as a less invasive alternative to surgical biopsy for the evaluation of pulmonary nodules and parenchymal lung disease¹. Little is known, however, about how commonly and for which patients this procedure is currently used. In particular, it is not known how the introduction and proliferation of other less invasive methods of obtaining lung specimens have affected utilization of transbronchial lung biopsy.

In light of reported mortality rates as high as 1–2 per 1000 patients undergoing transbronchial biopsy^2,3, concerns for complications related to transbronchial biopsy, including pneumothorax and procedure-related hemorrhage, may influence the use of this procedure. There are only limited data about which characteristics place individuals at increased risk for complications following transbronchial lung biopsy.

We sought to address this gap in the literature by using the Florida State Inpatient Databases and State Ambulatory Surgery Databases from 2000 to 2009 to examine trends in the utilization of transbronchial lung biopsy and to calculate population-based estimates of complications associated with this procedure.

METHODS

Data Source

We used the Healthcare Cost and Utilization Project Agency for Healthcare Research and Quality Florida State Inpatient Databases and State Ambulatory Surgery Databases from 2000 to 2009. The State Inpatient Database includes discharge data from approximately 93% of non-federal acute care community hospitals in Florida⁴. The State Ambulatory Surgery Database includes data from all ambulatory surgical centers in Florida that experience more than 200 visits per quarter⁵. Each record in the State Inpatient Database contains multiple data elements including patient demographic characteristics, and up to 31 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses and up to 31 ICD-9-CM procedure codes. The State Ambulatory Surgical Database contains similar demographic elements and up to 15 Current Procedural Terminology (CPT) codes and 5 ICD-9-CM diagnosis codes.

Study Population

We identified adults (age ≥ 18) who had undergone bronchoscopy with transbronchial biopsy in Florida from 2000 to 2009 by the presence of the code for transbronchial lung biopsy (ICD-9-CM 33.27; CPT 31628) on the discharge record. For comparison we also identified patients who had undergone other diagnostic pulmonary procedures including any bronchoscopy, bronchoscopy with other bronchoscopic biopsy (bronchoalveolar lavage, bronchial washing, bronchial brushing or endobronchial biopsy), or other lung biopsy procedures (mediastinoscopy, open lung biopsy, transthoracic needle biopsy). In addition to basic demographic characteristics, we identified medical conditions associated with transbronchial lung biopsy, which were not considered mutually exclusive, by the presence of select codes in any of the diagnosis code fields.

Outcomes

Our main outcome was the risk of complications associated with transbronchial biopsy. For our analysis of risk of complications associated with transbronchial biopsy, we excluded patients who had undergone other common procedures within the chest cavity during the same visit (e.g., central line placement, thoracentesis). We identified pneumothorax by the presence of the ICD-9-CM diagnosis code for “iatrogenic pneumothorax” (ICD-9-CM 512.1). The outcome of pneumothorax requiring chest tube placement was defined as the presence of codes for both iatrogenic pneumothorax and chest tube placement during a single visit in patients without co-existing pleural effusion. We identified procedure related hemorrhage by the presence of the ICD-9-CM diagnosis code for hemorrhage complicating a procedure (ICD-9-CM 998.11). Additional details on the coding strategy can be found in Appendix 1.

Statistical Analysis

The annual rate of transbronchial lung biopsy per 100,000 adults in Florida for each year from 2000 to 2009 was calculated and age standardized to the US census distribution from 2000^6,7. Age-adjusted rates of competing procedures were calculated in the same manner. Annual percent change (APC) in procedure utilization and procedure-related complications were calculated using the JoinPoint Regression Program, version 3.5.2 (Statistical Research and Applications Branch, National Cancer Institute).

We used logistic regression analysis to examine characteristics associated with iatrogenic pneumothorax and procedure-related hemorrhage. We used single-predictor logistic regression to identify significant associations between biopsy characteristics and complications. For the multivariable models, covariates with p<0.2 were entered into the model and remained via backward selection if p<0.05, the predetermined level of significance. To account for interactions among variables we examined interaction terms. None of the interactions were significant in the multivariable models.

We performed a series of sensitivity analyses to address potential limitations of using administrative databases. Across the study period, the maximum number of ICD-9-CM codes and CPT codes recorded in the Florida databases was not consistent, ranging from 10 to 31. The primary analysis allows the maximum number of procedure and diagnosis codes available for each year. To address the possibility that changes in the number of codes falsely affected our results over time, the analysis was repeated limiting the number of ICD-9-CM and CPT codes to 10 per record across all years. To address the possibility that a change in the rate of transbronchial biopsy over time represents a change in coding practice instead of a change in procedure utilization (i.e., a possible “crowding out” of minor procedures from the discharge record if the number of procedures per visit increased over time), we also analyzed the combined frequency with which other bronchoscopic biopsy procedures were coded during the study period. To address the possibility that the code for iatrogenic pneumothorax was too specific, we also analyzed risk of pneumothorax using the non-specific “any pneumothorax” code (ICD-9-CM 512). Unless specified, all analyses were performed using the SAS version 9.1.3 statistical software (Cary, NC). This study was reviewed and determined to be exempt by the Boston University Medical Campus Institutional Review Board.

RESULTS

Utilization of transbronchial biopsy, bronchoscopy and other lung biopsy procedures

From 2000 to 2009, the Florida State Inpatient Database and State Ambulatory Surgery Databases contained data from 48,553,802 adult patient visits, 99,533 of which involved transbronchial lung biopsy (Figure 1). Table 1 shows characteristics of the study population. The age adjusted annual rate of transbronchial biopsy decreased by 25% over the 10-year study period (74 versus 55 procedures per 100,000 adults, APC −3.2%, p<0.0001). Demographic characteristics such as age, race and gender were similar during the study period. The most common diagnosis codes associated with transbronchial biopsy were related to pulmonary nodules or lung cancer. While the frequency of most diagnosis codes associated with transbronchial biopsy was consistent during the study period, the proportion of transbronchial biopsies performed in the setting of lung transplantation increased from 2.0% to 11.1%. The number of lung transplant surgeries performed in Florida similarly increased from 23 in 2000 to 125 in 2009 (0.19 versus 0.85 procedures per 100,000 adults).

Figure 1.

Study Flow Diagram

Table 1.

Demographic Characteristics of Patients Undergoing Transbronchial Biopsy in Florida in 2000 compare to 2009

	2000	2009
Age (mean, SD)	64.9 (14.7)	63.6 (14.3)
Female (%)	45.5	47.0
Race (%)
White	82.3	78.7
Black	10.4	11.2
Hispanic	5.8	6.6
Asian or Pacific Islander	0.3	0.8
Native American	0.1	0.1
Other	1.0	2.6
Associated Conditions (%)
Pulmonary Nodule or Malignancy	43.4	39.5
Unspecified Pneumonia	16.0	19.3
Hemoptysis	9.3	8.0
Pulmonary Eosinophilia	5.5	5.3
Pulmonary Fibrosis	8.0	9.5
Lung Transplant	2.0	11.1
HIV	5.0	2.5
Sarcoidosis	1.3	1.7

In contrast to the significant decrease seen in the utilization of transbronchial lung biopsy, the age adjusted overall utilization of bronchoscopy was unchanged (268 versus 275 procedures per 100,000 adults, APC 0.19%, p=0.39). The combined age adjusted utilization of other bronchoscopic biopsy techniques, however, increased during this period (153 versus 195 procedures per 100,000 adults, APC 2.7%, p<0.0001), see Figure 2. The utilization of mediastinoscopy (12 versus 12 procedures per 100,000 adults, APC 0.0%, p=0.99) and transthoracic needle lung biopsy (54 versus 52 procedures per 100,000 adults, APC 0.0%, p=0.99) were unchanged. The rate of surgical lung biopsy (combined total of open and thoracoscopic procedures) decreased during the study period (12 versus 9.1 procedures per 100,000 adults, APC −3.6%, p=0.0003), see Figure 3.

Figure 2.

Age-Adjusted Rates of Utilization of Transbronchial Biopsy and Other Bronchoscopic Procedures in Florida from 2000 to 2009

Figure 3.

Age-Adjusted Rates of Utilization of Lung Biopsy Procedures in Florida from 2000 to 2009

Risk of Complications

After excluding patients who had undergone additional procedures during the encounter associated with risk for the complications of interest, we calculated that 0.97% (95% CI 0.94% to 1.01%) of transbronchial lung biopsies were complicated by iatrogenic pneumothorax, 0.55% (95% CI 0.52% to 0.58%) by pneumothorax requiring chest tube placement and 0.58% (95% CI 0.55% to 0.61%) by procedure related hemorrhage. Using a less specific definition of any pneumothorax, the risks calculated were slightly higher at 1.45% (95% CI 1.41% to 1.49%). Overall complication rates per 1000 procedures were not significantly changed throughout the study period: iatrogenic pneumothorax (APC 2.04%, p=0.31), pneumothorax requiring chest tube (APC 3.19%, p=0.31), procedure related hemorrhage (APC 2.33%, p=0.23).

Factors Associated with Complications

Table 2 shows characteristics associated with iatrogenic pneumothorax and procedure-related hemorrhage in univariate and multivariate logistic regression analysis. Chronic obstructive pulmonary disease (COPD) and female gender were significantly associated with higher risk of iatrogenic pneumothorax in multivariate analysis. Kidney failure, cirrhosis, age and female gender were significantly associated with increased risk of procedure-related hemorrhage. Lung transplantation was significantly associated with lower risk for both iatrogenic pneumothorax and procedure-related hemorrhage.

Table 2.

Risk Factors for Complications Associated with Transbronchial Biopsy

	Unadjusted* Odds Ratio (95% CI), p	Adjusted¶ Odds Ratio (95% CI),p
Iatrogenic Pneumothorax
COPD	1.54 (1.34 to 1.78), < 0.0001	1.51 (1.31 to 1.75), <0.0001
Age (per 10 years)	1.05 (1.01 to 1.11), 0.03	NS
Hospital Yearly Procedure Volume (per 100 TBBX)	0.93 (0.87 to 1.00), 0.04	NS
Lung Transplant	0.37 (0.24 to 0.56), <0.0001	0.40 (0.26 to 0.62), <0.0001
Female Gender	1.30 (1.13 to 1.50), 0.0002	1.32 (1.15 to 1.52), 0.0001
Procedure Related Hemorrhage
COPD	1.10 (0.90 to 1.33), 0.35	NS
Age (per 10 years)	1.20 (1.12 to 1.29), <0.0001	1.17 (1.09 to 1.25), <0.0001
Hospital Yearly Procedure Volume (per 100 TBBX)	0.92 (0.84 to 1.00), 0.06	NS
Lung Transplant	0.31 (0.17 to 0.56), 0.0001	0.39 (0.21 to 0.72), 0.002
Female Gender	1.36 (1.13 to 1.62), 0.001	1.40 (1.17 to 1.68), 0.0003
Kidney failure	2.94 (2.17 to 3.99), <0.0001	2.85 (2.10 to 3.87), <0.0001
Cirrhosis	2.40 (1.23 to 4.66), 0.01	2.31 (1.18 to 4.52), 0.01
Thrombocytopenia	1.94 (1.06 to 3.53), 0.03	NS

^*Derived from single-predictor logistic regression analyses

^¶Derived from multivariate logistic regression models

NS indicates not significant.

Variables with adjusted p-values < 0.05 are included in the final model.

DISCUSSION

In this study we demonstrate that utilization of transbronchial lung biopsy in Florida decreased by 25% from 2000 to 2009 (p<0.0001). This decrease occurred despite stability in the annual rate of bronchoscopy and cannot be explained by a concomitant increase in the rate of competing minimally invasive lung biopsy procedures such as video-assisted thoracoscopic surgery. We did identify an increase in the annual rate of other bronchoscopic biopsy techniques, which raises the possibility that bronchoscopists are increasingly favoring these procedures over transbronchial biopsy. Along these lines, another possibility for decreasing utilization of transbronchial lung biopsy is supplantation by newer bronchoscopic techniques such as endobronchial ultrasound and electromagnetic navigational bronchoscopy. However, because these techniques do not have dedicated ICD-9-CM procedure codes, we were unable to assess their use. Although no studies have previously addressed the utilization of transbronchial biopsy, the stable rate of fiberoptic bronchoscopy observed contrasts with the only previous study assessing utilization of bronchoscopy, a single center study that showed decreasing use of bronchoscopy in the late 1990s⁸.

The major advantage to using comprehensive state administrative databases for analysis of trends of procedure utilization is the ability to capture nearly all procedures performed in a population-based sample. A number of disadvantages, however, exist. It is likely that our estimate of the annual rate of transbronchial biopsy underestimates the actual rate. Using ICD-9-CM codes, transbronchial lung biopsy must be coded in addition to flexible or rigid bronchoscopy. Particularly in inpatients, it is possible that coding for a minor procedure such as transbronchial lung biopsy might be omitted. We attempted to account for this possibility by also analyzing the combined rate of other bronchoscopic biopsies, which would be subject to similar coding omission. The fact that the rate of these other bronchoscopic techniques is increasing suggests that the decrease in the annual rate of transbronchial lung biopsy is not an artifact of coding omission. We were also concerned that variation in the maximal number of procedure codes available across the study period might result in spurious estimates of procedure utilization over time. The similar trends in the rates of procedure utilization seen when the analysis was repeated using a constant number of procedure codes available across all study years allows us to be confident that the substantial decrease in transbronchial biopsy utilization we observed is real.

The majority of transbronchial biopsies were performed in patients with a diagnosis of a pulmonary nodule, lung cancer or parenchymal lung disease. Of note, these ICD-9-CM codes are non-specific, as would be expected in association with a diagnostic procedure performed to establish a diagnosis. Because the discharge records had no information on pathologic results and could not be linked to follow-up visits, we were unable to ascertain, for example, how often transbronchial biopsy performed for “pulmonary nodules” or “parenchymal lung disease” yielded a specific diagnosis (e.g., sarcoidosis). While the frequency of most associated conditions evaluated were stable throughout the study period, the proportion of biopsies performed in patients who had undergone lung transplantation, increased 5-fold from 2000 to 2009. This increase corresponded with a significant rise in the number of lung transplant surgeries being performed in the state of Florida during this time period.

One explanation for the decrease in utilization of transbronchial biopsy may be concern for complications related to this procedure, specifically hemorrhage and/or pneumothorax. Estimates of complications of transbronchial biopsy have been highly variable. Estimates of the risk of pneumothorax associated with transbronchial biopsy typically range from 0 to 5%, with higher rates reported in mechanically ventilated patients^2,3,9–18. The definition of procedure-related hemorrhage is less standardized, but clinically significant hemorrhage has been reported in 0.6 to 1.3% in large case series^2,3,16,17. In this study we analyzed 82,059 transbronchial biopsies and calculated overall rates of iatrogenic pneumothorax of 0.97% (0.55% requiring chest tube placement) and procedure-related hemorrhage rate of 0.58%. These complication rates were stable throughout the study period. The complication rates calculated in this study are lower than some previously reported estimates. In contrast to prior studies, which were conducted primarily in single academic centers, our study includes patients from both large academic hospitals and community hospitals and is more representative of the general population. It is therefore likely that our estimates are closer to the true complication rates within the community, where procedures are typically performed in healthier individuals. It is also likely, however, that our results underestimate the risk of complications associated with transbronchial lung biopsy. While in general, estimates of complications such as iatrogenic pneumothorax and procedure-related hemorrhage from administrative data are considered reliable and precise, both the procedure and the complication must be accurately coded which may lead to events not being captured^19–21. Reassuringly, our findings are consistent with a recent survey of complications of bronchoscopy procedures performed throughout Japan, which found a 0.6% risk of pneumothorax requiring chest tube after transbronchial biopsy, as well as a large prospective cohort study addressing bleeding complications of transbronchial biopsy in patients on aspirin which reported serious bleeding complications in 0.7% of cases^17,18.

This study is one of the first to identify specific factors associated with complications from transbronchial biopsy. The presence of COPD has consistently been identified as a risk factor for pneumothorax associated with transthoracic needle biopsy^22–28. It is therefore not surprising that COPD was also significantly associated with risk for pneumothorax associated with transbronchial biopsy. Similarly, cirrhosis and uremia are well-described risks for procedure related hemorrhage associated with bronchoscopy^1,29,30. Thrombocytopenia was identified as a risk factor for procedure related hemorrhage in univariate analysis but did not reach statistical significance in multivariate analysis. In our series only 1.2% of all transbronchial biopsies were performed in patients with thrombocytopenia which may reflect both a reluctance to perform the procedure in patients with low platelet counts as well as well as failure to capture secondary diagnoses such as thrombocytopenia in the discharge record. In this study, female gender was also associated with increased risk for both pneumothorax and procedure-related hemorrhage. While studies of transthoracic needle biopsy have not identified gender as a risk factor for pneumothorax, one study identified female gender as a risk factor for procedure-related hemorrhage²⁷. The reason for the association of gender with complications from transbronchial biopsy is unclear. In contrast, lung transplantation was associated with a decreased risk for complications associated with transbronchial biopsy. Multivariate analysis including interaction variables suggests that this is independent of hospital annual procedure volume. One study of mechanically ventilated patients undergoing transbronchial biopsy, which included 34 lung transplant patients, also identified a decreased risk of pneumothorax in the lung transplant population although this association did not reach statistical significance¹⁴. It is possible that this decreased risk is related to individual physician experience independent of annual hospital volume or related to the quality of the transplanted organs (for example lack of obstructive lung disease)³¹.

This research, which analyzes all transbronchial lung biopsies conducted in Florida over a decade, represents the single largest study of transbronchial lung biopsy. Although utilization of transbronchial lung biopsy is decreasing relative to the use of fiberoptic bronchoscopy and other minimally invasive lung biopsy procedures, our data suggest that transbronchial lung biopsy is a safe procedure with a low risk of complications for most individuals. Our findings on characteristics associated with higher risk of complications can help physicians and patients make more informed decisions about whether to pursue this diagnostic procedure.