Respiratory illnesses are a leading cause of morbidity, mortality, school absenteeism, and increased health care expenditures in children worldwide.1,2 Bronchoalveolar lavage (BAL) via flexible bronchoscopy is an essential tool that can help to establish an early diagnosis and guide appropriate management in children with respiratory illnesses of different etiologies (including immunologic, inflammatory, and infectious processes).3–6 Thus, if performed correctly, it can decrease the risk of potential complications and reduce the burden associated to these conditions.4,7
The clinical value of the BAL is thought to be linked to the volume of fluid recovered, as a lower percentage of volume retrieved may represent a proximal (nonalveolar) sample and has been associated with a decreased diagnostic yield.6,8–13 One essential aspect that can affect the volume of fluid recovered is the lobe where the BAL is performed. The available pediatric and adult guidelines for this procedure suggest that the right middle lobe (RML) and lingula (LIN) are the preferred sites as they have the highest fluid return6,7,13–15; however, there is very little evidence to support this opinion.3
We hypothesized that the site of BAL would be an important determinant of the volume return from BAL in children. To test this hypothesis, we conducted a retrospective review of all flexible bronchoscopies with BAL performed at a single institution between January 2011 and December 2012.
MATERIALS AND METHODS
Study Setting and Data Sources
Children’s Hospital of Pittsburgh of UPMC is a tertiary-care pediatric hospital located in southwestern Pennsylvania, United States. At this institution, BALs via flexible bronchoscopy are performed by a member of the Division of Pediatric Pulmonology (either an attending physician or a fellow in-training supervised by an attending physician) and assisted by a respiratory therapist. Procedures are performed under sedation or anesthesia in an endoscopy suite, an operating room, or one of the intensive care units (ICUs) (pediatric ICU, neonatal ICU, or cardiac ICU). Postprocedure notes are entered into the electronic medical record by the physician doing the bronchoscopy using the Olympus EndoWorks suite version 7.3 (Olympus America Inc., Center Valley, PA) within 1 hour after completing the procedure. A separate electronic database containing information about all flexible bronchoscopies is maintained at the hospital’s pulmonary function laboratory. Information in this electronic database is entered on the same day of the procedures by the assisting respiratory therapist.
The size of the flexible bronchoscope, the type of airway used (eg, laryngeal mask airway, endotracheal tube, or tracheostomy), and the site of the BAL varies according to the procedure’s indication and/or the patient’s clinical status. No specific protocol exists for conducting BALs at Children’s Hospital of Pittsburgh of UPMC. In general, 1 to 3 aliquots of 1 mL/kg of nonbacteriostatic normal saline (up to a maximum of 50 mL per aliquot) at room temperature are instilled through the working channel of the bronchoscope. Fluid is recovered immediately after each aliquot is instilled (ie, no dwell time) using either continuous wall suction or handheld syringe suction, according to the physician’s preference. Samples are then pooled and sent to the hospital’s laboratory for cytologic and microbiological analyses.
We reviewed the 2 electronic databases described above to gather information on all flexible bronchoscopies with BAL conducted at Children’s Hospital of Pittsburgh of UPMC between January 2011 and December 2012 as described previously.10 We collected information on patient’s demographics, personnel performing the procedure, indication for and location of the procedure in the hospital, and technical aspects of the flexible bronchoscopy and BAL. We excluded from the analyses procedures in patients 21 years and above and those without information on the site of the BAL, the total volume instilled for the BAL, or the total volume recovered from the BAL. For patients who underwent >1 BAL during the study period, we only included the earliest procedure in analyses to maintain independence among observations. Likewise, in children who had >1 BAL during a single procedure, we only included data from the first lavage. This study was approved by the Institutional Review Board of the University of Pittsburgh (Pittsburgh, PA).
Our predictor of interest was the site of the BAL. Our outcome of interest was the percent of fluid return (calculated as total volume recovered×100/total volume instilled). To assess whether there was any difference in the percent of fluid return between lobes, we first used 1-way analysis of variance with subsequent pairwise comparisons for unadjusted analyses. We tested the equality of variances between groups for the analysis of variance with the Levene’s test.
Next, we used multivariable linear regression for adjusted analysis. Because of the small sample size per group and to avoid model overfitting, we selected a priori the variables to be included in our main multivariable model [child’s age, sex, and type of suction (continuous wall vs. handheld syringe)] based on published literature.10,16 Lastly, to control for multiple comparisons, we used the Tukey-Kramer test.
To further investigate the possibility of confounding, we then built exploratory models by adding one of the following covariates to our main model: total volume instilled for the BAL (mL), location of the procedure in the hospital (any ICU vs. other), size of the flexible bronchoscope’s suction channel (1.2 vs. 2.0 mm), physician’s level of training (fellow vs. attending), and indication for the procedure (cough or wheezing vs. other). All statistical analyses were carried out using SAS 9.4 (SAS Institute, Cary, NC).
A total of 647 BALs via flexible bronchoscopy were performed at our institution during the study period. After excluding those performed in adults (n=5), those without information on the site of BAL (n=10) or percentage of volume recovered from BAL (n=10), and those conducted as follow-up procedures (n=93), 529 (∼81.8%) of the 647 observations remained for analyses.
The baseline characteristics of children included in this study can be found in Table 1. The majority of BALs were performed in males, in a non-ICU setting, by fellows, using a flexible bronchoscope with a 1.2-mm suction channel, and through a laryngeal mask airway. A total of 20 physicians (10 fellows and 10 attending physicians) participated in the BALs. The RML was the most common site for the BAL (n=424, 80.2%). This was followed by the LIN (n=46, 8.7%). All the other lobes accounted for <12% of the procedures.
To facilitate statistical analyses, we collapsed the different lobes into 4 groups on the basis of their anatomic location as follows: (1) right or left upper lobe other than LIN (R/L-UL, n=21), (2) RML (n=424), (3) LIN (n=46), and (4) right or left lower lobe (R/L-LL, n=38). The variances of these 4 groups seemed to be equal despite the differences in sample sizes (P=0.08 for the Levene’s test).
The mean (SD) percent of fluid return was 52.1 (14.4) for the RML, 50.7 (16.0) for the LIN, 50.5 (18.6) for the R/L-UL, and 42.2 (18.7) for the R/L-LL (Fig. 1). The difference in these means was statistically significant (P=0.002). The R/L-LL had a significantly lower fluid return when compared with each of the other lobes (P<0.05 for all pairwise comparisons involving the R/L-LL); in contrast, there was no significant difference in fluid return between the other lobes (P>0.6 for all other pairwise comparisons).
In view of the above findings, we then proceeded to perform multivariable linear regression models using the R/L-LL as the reference category. The significant lower fluid return of the R/L-LL persisted even after adjustment for potential confounders. For instance, there was an estimated 11.1% (95% confidence interval, 6.2-16.1; P<0.001) increase in the fluid return in the RML and an estimated 9.5% (95% confidence interval, 3.2-15.8; P=0.003) in the LIN when compared with the R/L-LL in our main multivariable analysis adjusting for the child’s age, sex, and the type of suction (Table 2). We obtained similar results in our exploratory analyses adjusting for other potential confounders (data not shown).
The differences in fluid return of the R/L-LL with the RML and LIN remained significant after adjustment for multiple comparisons (P<0.001 and 0.02 in our main model, respectively), although the difference with the R/L-UL did not (P=0.1), likely because of the small sample size.
Despite being one of the most common invasive procedures performed by pediatric pulmonologists worldwide, the technique for BAL via flexible bronchoscopy is not standardized in children.7,17–19 Although the decision of where to perform a BAL must be on the basis of the child’s clinical and/or radiologic findings, the site of the BAL can substantially affect the efficacy of this procedure. Unfortunately, only scant evidence on the optimal site of the BAL exists.6 In our study, we found that the RML and LIN seem to have a similar BAL fluid return. In contrast, the lower lobes have a lower fluid return when compared with the other ones.
In a study of 5 children aged 2 years and below, Midulla et al20 also found that the fluid return from the RML is similar to that of the LIN. However, this study was limited by the very small sample size, the lack of comparison with other lobes, and the use of unadjusted analysis. Furthermore, 4 of these 5 children had no evidence of lower airway or parenchymal lung disease, and it is believed that the fluid recovery may differ between healthy children and those with acute or chronic respiratory illnesses.18 To the best of our knowledge, no other studies in this field have been published in children. There is also very limited data in adults, with most16,21,22 but not all23 studies reporting a higher fluid return in the RML or LIN when compared with other lobes. Nonetheless, the RML and LIN are frequently cited as the lobes with the best fluid return in both pediatric and adult guidelines.6,7,24
We can only speculate on the reasons for our findings as the dynamics of fluid return are complex.25 It is possible that the low recovery in the lower lobes is related to their larger surface area, the difficulty in wedging with the flexible bronchoscope, or their dependent situation in the supine position.15,22
Our study has considerable strengths, such as the large number of procedures reviewed and the statistical methods accounting for potential confounders. Most importantly, our study adds to the small but necessary literature comparing the different techniques of BAL in children. Indeed, as noted in the recently published American Thoracic Society guidelines for flexible bronchoscopy in children,6 most current recommendations for this procedure are based on clinical experience, as only very few pediatric hypothesis-driven studies have been conducted.
We also acknowledge several limitations to our findings. First, we were unable to determine whether the difference in percent of fluid return between the lower lobes and the other ones (which ranged from 8.7% to 11.1% in our study) was associated with a higher diagnostic yield, as we lacked patient-related outcomes. It has been previously suggested that a higher fluid return is an adequate marker of optimal alveolar sampling and procedure adequacy,13,15,26 although this has not been sufficiently studied. In an adult study comparing 2 different BAL suction techniques, a higher fluid return of only 8% was associated with 17% higher number of final diagnoses.8 In all groups studied, a higher percentage of BAL fluid return was associated with an increased likelihood of establishing a diagnosis based on the BAL, which suggests that our results may be clinically significant.8 In addition, larger BAL samples could potentially provide investigators with more cells for culture and analyses and, thus, be important to advance high-quality pediatric pulmonary research.9,26 Second, we were unable to analyze the right and left upper or lower lobes separately because of the small sample sizes in each group. Third, as it is the case with any other observational study, selection bias could have affected our results. The majority of the procedures in our sample (80.2%) were performed in the RML. This is likely due to the fact that most pulmonologists are taught that this is the best lobe for BAL (despite the little evidence to support this belief, as noted above).7 However, it is also possible (albeit unlikely) that these BALs were performed in the RML because this was the site most severely affected or the lung disease was only localized to this lobe (which could have biased our results). Fourth, there could be residual confounding by measured or unmeasured variables. Because of the small sample sizes in each group, we were not able to include more than a few variables per multivariable model. However, we obtained almost identical results to our main model (that included a priori selected covariates) in our exploratory models that adjusted for a variety of potential confounders, which further supports our conclusions. Therefore, it is unlikely that any of the variables included in our exploratory models (such as child’s age, sex, type of suction, total volume instilled for the BAL, ICU status, size of the flexible bronchoscope’s suction channel, or physician’s level of training) could have biased our results. Unfortunately, we were unable to fully adjust for the diagnostic indication for the procedure, as the databases we used did not accurately record the patient’s specific lung disease. For instance, nonspecific symptoms (such as wheezing and cough) accounted for ∼55% of the indications, thus residual confounding by the specific lung disease is a possibility. Likewise, the interprovider and intraprovider variability, the dwell time to aspirate the fluid, the generation number where the tip of the flexible bronchoscope was wedged, or the presence of positive-pressure ventilation are all factors that were not included in our analyses and may have affected our results.
In summary, our data support the concept that BAL of the lower lobes will have a decreased percent fluid return than a lavage performed elsewhere. Although widely stated in both pediatric and adult literatures, there have been little data to date, to back up the belief that the RML or LIN are the sites with the highest fluid return. The decision of where to perform a BAL should be on the basis of the patient’s clinical and/or radiologic findings. However, when there is no clear reason for an alternate choice, our objective data suggest that the BAL should be performed in the RML or LIN. More studies comparing the different techniques for flexible bronchoscopy with BAL in children are needed to standardize this common procedure, enhance its diagnostic yield, and improve the care of children with respiratory diseases.
1. Anderson HR. Respiratory disease in childhood. Br Med Bull. 1986;42:167–171.
3. Connett GJ. Bronchoalveolar lavage
. Paediatr Respir Rev. 2000;1:52–56.
4. Wood RE. The diagnostic effectiveness of the flexible bronchoscope in children
. Pediatr Pulmonol. 1985;1:188–192.
5. Balfour-Lynn IM, Spencer H. Bronchoscopy—how and when? Paediatr Respir Rev. 2002;3:255–264.
6. Faro A, Wood RE, Schechter MS, et al.. Official American Thoracic Society technical standards: flexible airway endoscopy in children
. Am J Respir Crit Care Med. 2015;191:1066–1080.
7. de Blic J, Midulla F, Barbato A, et al.. Bronchoalveolar lavage
. ERS Task Force on bronchoalveolar lavage
. European Respiratory Society. Eur Respir J. 2000;15:217–231.
8. Rosell A, Xaubet A, Agusti C, et al.. A new BAL fluid instillation and aspiration technique: a multicenter randomized study. Respir Med. 2006;100:529–535.
9. Singletary ML, Phillippi-Falkenstein KM, Scanlon E, et al.. Modification of a common BAL technique to enhance sample diagnostic value. J Am Assoc Lab Anim Sci. 2008;47:47–51.
10. Rosas-Salazar C, Walczak SA, Winger DG, et al.. Comparison of two aspiration techniques of bronchoalveolar lavage
. Pediatr Pulmonol. 2014;49:978–984.
11. Klech H, Haslam P, Turner-Warwick M. Worldwide clinical survey on bronchoalveolar lavage
(BAL) in sarcoidosis: experience in 62 centers in 19 countries. Sarcoidosis. 1986;3:113–117.
12. Meyer KC, Raghu G, Baughman RP, et al.. An official American Thoracic Society clinical practice guideline: the clinical utility of bronchoalveolar lavage
cellular analysis in interstitial lung disease. Am J Respir Crit Care Med. 2012;185:1004–1014.
13. Meyer KC. Bronchoalveolar lavage
as a diagnostic tool. Semin Respir Crit Care Med. 2007;28:546–560.
14. Baughman RP. Technical aspects of bronchoalveolar lavage
: recommendations for a standard procedure. Semin Respir Crit Care Med. 2007;28:475–485.
15. Technical recommendations and guidelines for bronchoalveolar lavage
(BAL). Report of the European Society of Pneumology Task Group. Eur Respir J. 1989;2:561–585.
16. Olsen HH, Grunewald J, Tornling G, et al.. Bronchoalveolar lavage
results are independent of season, age, gender and collection site. PLoS One. 2012;7:e43644.
17. Riedler J, Grigg J, Robertson CF. Role of bronchoalveolar lavage
with lung disease. Eur Respir J. 1995;8:1725–1730.
18. Shields MD, Riedler J. Bronchoalveolar lavage
and tracheal aspirate for assessing airway inflammation in children
. Am J Respir Crit Care Med. 2000;162:S15–S17.
19. Midulla F, Nenna RPriftis KN, Anthracopoulos MB, Eber E, Koumbourlis AC, Wood RE. Bronchoalveolar lavage
: indications and applications. Paediatric Bronchoscopy Progress in Respiratory Research. Basel, Switzerland: Karger; 2010:30–41.
20. Midulla F, Villani A, Merolla R, et al.. Bronchoalveolar lavage
studies in children
without parenchymal lung disease: cellular constituents and protein levels. Pediatr Pulmonol. 1995;20:112–118.
21. O’Riordan TG, Baughman RP, Dohn MN, et al.. Lobar pentamidine levels and Pneumocystis carinii
pneumonia following aerosolized pentamidine. Chest. 1994;105:53–56.
22. Limper AH, Specks U, Brutinel WM, et al.. Interlobar variation in the recovery of bronchoalveolar lavage
fluid, cell populations, and angiotensin-converting enzyme in normal volunteers. J Lab Clin Med. 1993;121:785–791.
23. Baughman RP, Dohn MN, Shipley R, et al.. Increased Pneumocystis carinii recovery from the upper lobes in pneumocystis pneumonia. The effect of aerosol pentamidine prophylaxis. Chest. 1993;103:426–432.
24. Henderson AJ. Bronchoalveolar lavage
. Arch Dis Child. 1994;70:167–169.
25. Kelly CA, Fenwick JD, Corris PA, et al.. Fluid dynamics during bronchoalveolar lavage
. Am Rev Respir Dis. 1988;138:81–84.
26. Pingleton SK, Harrison GF, Stechschulte DJ, et al.. Effect of location, pH, and temperature of instillate in bronchoalveolar lavage
in normal volunteers. Am Rev Respir Dis. 1983;128:1035–1037.