Oriainal Research

INTERVENTIONAL PULMONOLOGY

Prospective Risk-Adjusted Morbidity

and Mortality Outcome Analysis After
Therapeutic Bronchoscopic Procedures*
Results of a Multi-institutional Outcomes Database
Armin Emst, MD, FCCP; Michael S i m MD, ~ FCCP; David Ost, MD, FCCP;
Yaron Goldman, MD; and Felix J. F. Herth, MD, FCCP

Introduction: Interest in databases is growing to allow for outcomes research, assess health-care
quality, and determine best practices and resource docation, and they are increasinglyconsidered as
a tool to potentially tie reimbursement to outcome parameters. Little is known about resource use
and risk-adjusted morbidity and mortality after therapeutic bronchoscopic interventions.
Methoda: Data were extracted and reviewed from an ongoing prospective, multi-institutional
outcomes database for therapeutic bronchoscopic interventions.AU consecutive patients are entered
into this database, and information on demographics, indications, procedures and anesthesia,
comorbidities and general health status, urgency of intervention, morbidity and mortality to 30 days,
increase in levels of care, and procedural resources is documented.
Rerru2f.s: From December 2005 to May 2007,554 therapeutic procedures were performed in four
hospitals. Most procedures were done under general anesthesia (n = 362)and rigid bronchoscopy
(n = 483), and the most common intervention was airway stent placement (n = 258). Forty-two
percent of procedures were done urgently or emergently. Complications were common (19.8%),and
3O-day mortality was 7.8%,correlating with underlying health status and urgency of intervention.
Discussion: Prospective and ongoing data analysis for bronchoscopic procedures is feasible and
valuable. Risk-adjusted and disease-specific outcomes can be documented and potentially used for
quality assessment,benchmarking, and quality improvement initiatives. Appropriate use of resources
and effect of interventionscan be documented. Extending the number of participating centers as well
as inclusion of quality of life tools and technical success are the next steps.
(CHEST2008; 134:514519)

Key words: bronchoscopy; complications; interventional pulmonology; outcomes

Abbreviations:ASA = American Society of Anesthesiologists; CI = confidence interval; OR = odds ratio

nterest in the use of medical databases or patient
Ito collect
registries is increasing rapidly. Databases are used
information in a structured fashion and can
be used in many different ways, all depending on the
intended use of the collected information. Most
frequently, data are collected on medical and surgi-
cal outcomes as well as adherence to guidelines and
resource use.14 Patient registries allow for outcomes
assessment from the natural progression of disease to
assessing success (or lack thereof) of medical or
device-based and surgical interventions. Databases
are not clinical trials with fxed inclusion and exclu-
sion criteria, and they do not have control groups.
Well-designed databases allow for population-based
research and determination of best practices (best
results with the use of the least resources necessary)?
Participating in trials and outcomes databases is cred-
ited with decreasing morbidity and mortality after
surgical procedures by identifying outliers and offering
support for improving performance.7.8
With the increase of health-care spending moving
into the public eye and the forefront of discussions,
there is also an increased interest in tying reimburse-
ment to performance or value.QJ0An often quoted
database is the outcomes database maintained by the
Society of Thoracic Surgery, which has been used to

514 Original Research

clearly demonstrate resource use and risk-adjusted
outcomes,11-*4 which was in the end also used to
positively affect reimbursement negotiations.
Bronchoscopy is one of the most commonly per-
formed procedures in the United States. Interest in
therapeutic interventions is growing and could ex-
pand exponentially if, for example, endoscopic lung
volume reduction or endoscopic asthma treatment
will prove to be beneficial and clinically available.
Despite the very frequent use and the significant
risk associated with some of these procedures, few
recommendations regarding proper training exist,
little is known about risk-adjusted or disease-specific
outcomes, and outcome benchmarks have not been
established. Additionally, very little is known about
what resources are needed for best results. In this
article, we present a multi-institutional effort to
establish a comprehensive outcomes database for
therapeutic bronchoscopic procedures.
MATERIALS AND METHODS
The Interventional Pulmonology Programs at the participating
institutions (Beth Israel Deacxmess Medical Center, Boston; Henry
Ford Hospital, Detroit; New York University Hospital, New York
City: and Thoraxklinik Heidelberg, Germany) maintain an ongoing,
common prospective database on outcomes in therapeutic broncho-
scopic intewentions. The Institutional Review Boards at the partic-
ipating institutions approved the data collection. Informed consent
was waived. Data are collected prospectively and retrospectively
reviewed. The current database was preceded by an earlier version,
which after a p-trial phase was upgraded to the currently presented
database. All data-entry fields (such as complications, which in-
cluded hypoxia at < 90% saturation for > 1 min, pneumothorax,
escalation of care, bleeding, and hypotension) were predefined and
agreed on by the group. Similarly, all potential data-collection items
were reviewed for relevance and inclusion and agreed on by the
group members.
All submitted forms were controlled for completion before
entering into the database to ensure collection as complete as
possible. No effort was made to cross check the submitted data
with the actual medical records at the participating institutions.
All centers agreed to submission of all cases.
*From Interventional Pulmonolo and Thoracic Surgery (Drs.
Y
Ernst and Goldman), Beth Israe Deaconess Medical Center,
Harvard Medical School, Boston, MA; Division of Pulmonary and
Critical Care Medicine (Dr. Simoff), Henry Ford Hospital,
Detroit, MI; Division of Pulmona and Critical Medicine (Dr.
Ost), New York University Hospitx New York, NY; and Pulmo-
nary and Critical Care Medicine (Dr. Herth), Thoraxklinik
Heidelberg, Germany.
The authors have no conflicts of interest relating to the contents
of this article to declare.
Manuscript received March 3, 2008; revision accepted April 10,
2008.
Reproduction of this article is rohibited without written permission
org/misdre rints shtml).
P
from the American College o Chest Physicians (www.chestjoumal.
Correspon&nce to: A m i n Emst, MD, FCCP, Chief, Intemen-
tional P i t h n o ~ O g, Beth Israel Deaconess Medical Center, Har-
uard Medical Sc!ool, One Deaconess Ad, Deaconess 201A,
Boston, MA 0221 5; e-wmil: aernst@bidmc.hamard.edu
DOI: 10.1378/chest.08-0580
www.chestjournal.org
For this study, outcomes data from December 2005 to May
2007 were analyzed. The data collected included patient demo-
graphics (deidentified), indication for the procedure, timing, and
all details of the procedure performed. Comorbidities are docu-
mented as well as preprocedure general health status and anesthesia
risk. All intraoperative and postoperative complications or increased
levels of care are documented to 30 days. Additionally, prncedural
detail including time required, anesthesia used, and staff involved is
collected to estimate effort and resources going into specific proce-
dures. All procedures were entered into a database (Access, Mi-
crosoft Office; Microsoft; Redmond, WA) by one individual (Y.G.).
The database form is shown in the addendum to this article.
Statistical Methods
Continuous variables are expressed as means and SDs. Dichot-
omous variables are summarized as simple proportions. Baseline
characteristics were analyzed using two-sample t tests for contin-
uous variables and Fisher exact test for categorical variables.
Stratified analysis was done using the Cochrane Mantel-Haenzel
test. Univariate logistic regression was used to identify risk factors
for different types of complication and their respective odds
ratios (ORs). For each type of complication, a separate univariate
analysis was performed. Multivariable analysis was then used to
determine the relationship between risk factors and complica-
tions. In multivariable analysis, for each type of complication,
only variables with a p value < 0.2 in univariate analysis were
assessed. We used backward selection with an a of 0.05 for
inclusion in the fmal model. For all statistical tests, we used a
two-tailed p value < 0.05 to define statistical significance.
RESULTS
From December 2005 to May 2007,554 therapeu-
tic bronchoscopic procedures were performed and
analyzed for this report (Table 1). All procedures
were performed in the four participating institutions.
Most data fields were filled out with very little data
loss.
Most procedures were performed under general
anesthesia (n = 362) or total IV anesthesia (n = 140)
rather than moderate sedation (n = 33) or deep seda-
tion (n = 3). Rigid bronchoscopy was most commonly
employed (n = 483) and was performed in the operat-
ing room. The most commonly performed therapeutic
intervention was placement of airway stents (n = 258).
Ablative techniques included mechanical core-out with
the rigid scope (n = la),argon plasma coagdation
(n = 124), electrocautery (n = 34), and Nd-YAG laser
interventions (n = 29).
Only a fraction of therapeutic procedures were per-
formed under moderate sedation. Of the 33 proce-
dures done under moderate sedation, 14 were stent
placements. More procedures were performed for
nonmalignant airway obstruction than for malignant
airway obstruction (n = 301 vs n = 253), and patients
generally were quite sick as reflected in their high
American Society of Anesthesiologists (ASA) and Zubrod
scores and sigdicant comorbidities. Two hundred
CHEST I 134 1 3 I SEPTEMBER, 2008 515
 Table 1-Procedures and Complications There were significant differences between patients
Total Patients With
with malignant airway disease and those with benign
Procedures, Any Complication, airways disease in terms of complications. For the
Indications No. No. (%) outcome of any complication, those with an indication
Nonmalignant airway 30 1 46 (15) related to malignant disease were more likely to have
obstruction complications as compared to those with benign in&-
Postintubation tracheal stenosis 67 15 (22) cations (OR, 1.81;95% confidence interval [CI], 1.18to
Simple (web-like) 7 3 (43) 2.76; p = 0.007). Bleeding complications were also
Complex 56 12 (21)
Not recorded 4 0 (0)
more likely in those with malignant disease (OR, 4.96;
Idiopathic subglottic stenosis 41 3 (7) 95% CI, 1.49 to 16.55; p = 0.008). Hypoxia intraoper-
Involves cricoid 10 1 (10) atively was more common in those with malignant
lloes not iiivolve cricoid 31 2 (6) disease, but this did not reach statistical si@icance
Tracheoesoplvageal fistula 15 2 (13)
Malignant 11 2 (18) (OR, 1.48; 95% CI, 0.91 to 2.39; p = 0.14).
Benign 4 0 (0) Stratified analysis of other risk factors, such as tra-
Tracheomalacia 64 7 (11) cheal location (p = 0.003), diabetes (p = 0.004), and
Malignant airway obstruction 253 62 (25) COPD (p = 0.02), indicated that there was significant
By anatomic location
Tracheal involvement 93
heterogeneity of effect depending on whether or not
31 (33)
[,eft mainstem 89 25 (28) the patient had malignant or benign disease (Table 2).
Right iliainstein 124 24 (19) Similarly, for the outcome of intraoperative hypoxia
By type of airway compromise (Table 3), the effect of different risk factors varied
Endobronchial (intrinsic) 27 10 (37) based on whether or not the patient had a malignant
Extnnsic 30 8 (27)
Mixed 102
airway indication.
22 (22)
Other 94 22 (23) We therefore performed separate uriivariate (Ta-
ble 2) and multivariable logistic regression for those
with malignant airways disease and those with be-
nign airway disease to identify risk factors for com-
thirty of all procedures (n = 554; 42%) were per- plications. In multivariable analysis, among patients
formed on an urgent or emergent basis. Complications with malignant disease the variables associated with
were not infrequent. A complication was noted in 108 risk of complications were current tobacco use (OR,
of 554 of all cases (19.5%). 2.14; 95% CI, 1.28 to 5.26; p = 0.04), any diabetes

Table 2-ORs for Any Complication*

Test of Homogeneity
Variables Malignant Disease Benign Disease (Mantel-Haenzel),p Valne
Type of obstruction
Extrinsic 1.14 (0.49-2.66) 2.82 (0-22.1) 0.49
Intrinsic 1.97 (0.86-4.49) 1.17 (0.47-2.94) 0.42
Mixed 0.76 (0.42-1.38) 0.85 (0-3.51) 0.90
Location of disease
Trachea 2.08 (1.17-3.71) 0.55 (0.29-1.05) 0.003
Left mainstem 1.34 (0.75-2.41) 0.59 (0.25-1.35) 0.12
Right mainstein 0.<57(0.32-1.03) 0.50 (0.21-121) 0.80
Tobacco use
Any tobacco we 1.44 (0.71-2.89) 0.99 (0.53-1 .85) 0.44
Current tobacco use 2.38 (1.22-4.67) 0.92 (0-3.84) 0.27
Zubrod score > 2 1.46 (0.82-2.59) 0.91 (0.47-1.76) 0.30
ASA score > 2 0.78 (0.42-1.46) 0.41 (0.22-0.77) 0.16
Urgent or emergent case 1.73 (0.97-3.08) 1.39 (0.74-2.63) 0.62
Any coroiiary artery disease 0.66 (0.28-1.55) 0.79 (0.37-1.72) 0.29
Any history of congestive heart failure 0.98 (0.41-2.36) 1.25 (0.55-2.84) 0.70
Any diabetes 3.34 (1.60-7.00) 0.67 (0,31-1.45) 0.004
Any hypertension 0.31 (0.16-0.62) 0.38 (0.18-0.81) 0.71
COPD
Any COPD 1.48 (0.82-2.69) 0.56 (0.26-1.19) 0.05
Moderatelsevere COPD 1.98 (1.05-3.75) 0.56 (0.25-1.24) 0.02
Any renal failure 1.39 (0.44-4.45) 0 (0-1.47) 0.43
*Data are presented as OR (95% CI) unless otherwise indicated.

516 Original Research

 Table 3-ORa for Hypoxia *
Test of Homogeneity
Variables Malignant Disease Benign Disease Coinbined (Mantel-Haenzel), p Value

Type of obstruction
Extrinsic 2.16 (1.05-4.45) 3.98 (0-19.49) 2.33 (1.18-4.58) 0.52
Intrinsic 2.16 (1.054.45) 1.98 (0.81-4.82) 2.09 (1.18-3.69) 0.88
Mixed 0.51 (0.24-1.12) 0.63 (03.90) 0.53 (0%-1.11) 0.86
Location of disease
Trachea 3.31 (1.67-6.55) 0.63 (0.31-1.29) 0.001
Left inainstem 0.95 (0.47-1.90) 0.55 (0.21-1.44) 0.77 (0.44-1.37) 0.38
Right mainstem 0.55 (0.28-1.08) 0.44 (0.15-1.23) 0.51 (0.28-0.90) 0.73
TObdccO use
Any tobacco use 1.38(0.613.10) 0.79 (0.39-1.59) 1.01 (0.59-1.71) 0.32
Current tobacco use 2.92 (1.4M.11) 0.57 (0-3.57) 2.21 (1.1M.33) 0.15
Zubrod score > 2 1.32 (0.67-2.57) 0.74 (0.341.58) 1.02(0.61-1.68) 0.27
ASA score > 2 1.17 (0.542.50) 0.39 (0.19-0.78) 0.04
Urgent or emergent case 3.36 (1.62-6.97) 1.68(0.8X3.40) 2.40 (1.45-3.97) 0.19
Any coroiiary artery disease 1.20 (0.50-2.90) 1.19 (0.54-2.65) 1.20 (0.66-2.19) 0.34
Any history of congestive 1.41 (0.55-3.62) 1.51 (0.633.62) 1.46 (0.76-2.82) 0.92
heart failure
Any diabetes 4.22 (1.93-9.29) 1.01 (0.46-2.24) 0.01
Any hypertension 0.14 (0.050.40) 0.26 (0.10-0.67) 0.20 (0.10-0.40) 0.42
COPD
Any history of COPD 2.36 (1.20-4.63) 0.71 (0.31-1.59) 0.03
Moderate-to-severe COPD 3.05 (1.51-6.16) 0.69 (0.30-1.62) 0.009
Any rend failure 2.44 (0.76-7.91) 0 (0-2.02) 1.09 (0.363.29) 0.20
*Data are presented as OR (95% CI) unless otherwise indicated.

(OR, 3.78; 95% CI, 1.65 to 8.63; p = 0.002), any
hypertension (OR, 0.28; 95% CI, 0.13 to 0.59;
p = 0.001), and endobronchial disease (OR, 2.59;
95% CI, 1.28 to 5.26; p = 0.008). Among patients
with benign disease, multivariable analysis indicated
only postintubation trachea1 stenosis (OR, 2.01; 95%
CI, 0.99 to 4.08; p = 0.05), an ASA score > 2 (OR,
0.49; 95% CI, 0.25 to 0.95; p = 0.04), and any
hypertension (OR, 0.46; 95% CI, 0.21 to 1.03;
p = 0.06) were associated with risk of complications.
In multivariable analysis, among those with malig-
nant disease, the variables associated with risk of
hypoxia intrinsic airway obstruction (OR, 3.86; 95%
CI, 1.19 to 12.53; p = 0.03), tracheal involvement
(OR, 3.31; 95% CI, 1.45 to 7.57; p = 0.005), urgent
or emergent procedure (OR, 2.79; 95% CI, 1.21 to
6.42; p = 0.02), moderate-to-severe COPD (OR,
4.09; 95% CI, 1.68 to 10.0; p = 0.002), diabetes (OR,
3.40; 95% CI, 1.24 to 9.33; p = 0.018), and hyper-
tension (OR, 0.13; 95% CI, 0.04 to 0.39; p < 0.001)
were associated with hypoxia. Among patients with
benign disease, inultivariable analysis indicated only
hypertension (OR, 0.32; 95% C1, 0.12 to 0.87;
p = 0.03) was associated with hypoxia.
Thirty-day mortality for all patients was 43 of
554 patients (7.8%). Of these deaths, only one was
procedure related and occurred perioperatively.
The remainder were related to progression of
underlying disease.
Among patients with malignant disease, 30 of 185
patients (16%) with an ASA score > 2 died within 30
days, as compared to 0 of 68 patients with an ASA
score 5 2 (risk difference, 0.16; 95% CI, 0.11 to
0.22; p = 0.0004). Zubrod score also predicted 30-
day mortality (p = 0.0001). No other factors pre-
dicted 30-day mortality.
Among patients with benign disease, only Zubrod
score > 2 predicted 30-day mortality: 3 of 198 patients
(1.5%)with a score < 2 died, as compared to 10 of 104
patients (9.6%) with a score > 2. The ASA score did
not predict 30-day mortality among those with benign
disease (p = 0.39).
DISCUSSION
In this report, we present for the first time prospec-
tive, risk-adjusted, disease-specific outcome data for
therapeutic bronchoscopic procedures across several
institutions and in different countries. Our results show
that even in experienced centers, morbidity and mor-
tality for therapeutic bronchoscopic interventions
are not trivial. The patient population we encoun-
tered is frequently acutely ill with numerous co-
morbidities and significant predicted anesthetic
risk as evidenced by the high ASA scores. Not
surprisingly, the sicker the patient, the higher the
likelihood for an adverse outcome.

www.chestjournal.org CHEST/134/3/SEPTEMBER, 2008 517

 A surprising finding in our study was that patients
with malignant and nonmalignant airway compro-
mise undergoing interventional bronchoscopic pro-
cedures should not be considered as a single group,
but should rather be considered distinctly different
populations and analyzed and reported separately.
Interestingly, we saw a higher proportion of nonma-
lignant indications for airway interventions in our
cohort. This may be a reflection of the participating
centers’ dedicated interventional airway services. It
stands to argue that the number of interventions for
nonmalignant indications may rise even further, if
other procedures such as endoscopic lung volume
reduction enter clinical practice.
Data available to date have either been retrospec-
tively collected, addressed only the safety of single-
type interventions, were a single-institution experi-
ence, only included few patients, or did not go out to
30 days, and most importantly did not report risk-
adjusted or disease-specific outcome^.^"-'^ In our
model, multiple different interventions are collected
prospectively, and a 30-day window was chosen to
reflect current beliefs that procedure outcomes data
cannot just be limited to a few days perioperatively.18
A longer timeline is also reflective of institutional
systems of care and would more reliably detect
problems with patient selection for the intervention
studied.
We found the time required for data collection to
be acceptable, and the data submission was near
totally complete. This represents a significant differ-
ence to many other reports on outcome that often
suffer from significant incomplete data submission,
resulting in less robust models. A solution for this
problem would be Web-based data submission struc-
tured to require complete input.
Overall, we find that this database has the potential
for multiple uses. It can be used as a quality control
instrument allowing institutions to assess their longitu-
dinal changes in perfonnance or how they compare to
other institutions. A database that collects information
such as this one is uniquely positioned for it because it
provides risk-adjusted data as well as information re-
garding resource use. If a bronchoscopist finds that
he/she varies sigdcantly from peer institutions, this
would be a signal that can be used to assess the reasons
and improve the quality of care.
Quality in medical care and how to improve is not
the only potential use for databases such as this one.
Careful data collection on patient acuity, outcomes,
and expended resources to achieve good-quality out-
comes (benchmarks)are hard data that can and need to
be used for rational discussion about appropriate reim-
bursements for institutions and physicians.
A dreaded and frequently discussed subject is the
effort to tie the level of payment to the quality of
518
care. It has been shown that, for example, plain
procedural volume is predictive of good outcomes;
but this is a model that would punish smaller institu-
tions performing smaller volumes, even if the outcomes
were comparable. Participation in outcome databases
and efforts to improve quality may present a more
attractive and equitable opportunity for most institu-
tions and physicians interested in perfonning particular
procedures.
This database in its current form has shortcom-
ings. For one, only four institutions participated and
all four have high-volume dedicated interventional
pulmonary programs. As a reflection of this fact, a
majority of patients were treated under general
anesthesia and with the use of rigid endoscopy. It is
also conceivable that the patient population was
sicker than is the case in many smaller programs,
therefore limiting general interpretation of the re-
sults. Additionally, longer-term outcomes, quality of
life, functional performance, and technical success
have not been documented, limiting our ability to
address the actual value or impact of intervention for
the patient.
We believe that for a pilot project such as this one,
centers with a high volume and dedicated operators
who are motivated to participate in this effort would
best guarantee data submission and set a standard. Our
next steps are to add additional centers representing
different practice patterns to address the outlined
concerns. Additionally, functional and quality of life
parameters will be studied, as well as technical success
for different interventions. This will derive the addi-
tional benefit to conduct studies about the comparative
value of interventions for similar indications and estab-
lish best practices and value for any given indication.
In conclusion, we present for the first time risk-
adjusted and disease-specific outcome data for thera-
peutic bronchoscopy derived from a multi-institutional
database. Interventional bronchoscopy is associated
with significant risk and as a specialty probably more
complex than frequently assumed. Risk-adjusted data
such as presented here can be used for quality assess-
ment and improvement, benchmarking, and to assess
resource use and allocation. A Web-based model in-
cluding additional sites will be necessary to manage the
data most efficiently and to provide even more robust
modeling.
REFERENCES
1 Ellis SG, Omogui N, Bittle JA, et al. Analysis and comparison
of operator specific outcomes in interventional cardiology:
from a multicenter database of 4860 quality controlled pro-
cedures. Circulation 1996; 93:431-439
2 McGreevey C, Nadzam D, Corbin L. The Joint Commission
on Accreditation of Healthcare Organizations’ indicator mea-
surement system. Cornp Nursing 1997; 15:S87-S95
Original Research
 3 Little AG, Rusch VW, Bonner JA, et d.Pattarns of surgical
care of lung cancer patients. Ann Thorac Surg 2005; 80:2051-
2056
4 Reilly JJ, Chin B, Berkowitz J, et al. Use of a state-wide
administrative database in assessing a regional trauma system:
the New York City experience. J Am Coll Surg 2004; 198:
509518
5 Khuri SF, Daley J. Henderson W, et al. The Department of
Veteran’s Affairs’ NSQIP: the first national, validated, out-
come based, risk-adjusted and peer-controlled program for
the measurement and enhancement of the quality of surgical
care. Ann Surg 1998; 228:491507
6 Agency for Healthcare Research and Quality. Registries for
evaluating patient outcomes: a user’s guide; publication No.
07-EHC001-1. Available a t http:/h.effectivehealthcare.
ahrq.gov. Amessed July 8, 2008
7 You NY, Jacobs L, Martinez ED, et al. Improved surgeon
performance in clinical trials: an analysis of quality assurance
audits from the American College of Surgeons Oncology
Group. J Am Coll Surg 2006; 203:269-276
8 Shukri FK. The NSQIP: a new frontier in surgery. Surgery
2005; 138337-843
9 Smoldt RK, Cortese DA. Pay-for-performance or pay-for-
value? Mayo Clin Proc 2007; 82:210-213
10 Rosenthal MB. Nonpayment for performance? Medicare’s
new reimbursement rule. N Engl J Med 2007; 357:1573-1575
11 Ferguson TB. The Society of Thoracic Surgeons’ National
www.chestjournal.org
Cardiac Database: provider engagement in continuous quality
improvement. Am Heart Hosp J 2003; 1:180-182
12 Clark RE. The STS Cardiac Surgery National Database: an
update. Ann Thorac Surg 1995; 59:1376-1381
13 Jacobs JP, Lacour-gayet FG, Jacobs ML, et al. Initial appli-
cation in the STS congenital database to evaluate surgical case
mix and results. Ann Thorac Surg 2005; 79:163!5-1649
14 Ferguson TB, Dziubian SW, Edwards FH, et al. The STS
National Database: current changes and challenges for the
new millennium. Ann Thorac Surg 2000; 69:680-691
15 Dunagan DP, Burke HL, Aquino SL, et al. Fiberoptic
bronchoscopy in coronary care unit patients: indications,
safety and clinical implications. Chest 1998; 114:1660-1667
16 Pemn G, Colt HG, Mak MA, et al. Safety of interventional
rigid bronchoscopy using intravenous anesthesia and sponta-
neous assisted ventilation: a prospective study. Chest 1992;
102:1526-1530
17 Jimenez-Diaz JP, Martinez-Ballarin JE, Llunell A, et al.
Efficacy and safety of photodynamic therapy versus Nd-YAG
laser resection in NSCLC with ainvay obstruction. Eur Respir
J 1999; 14:800-805
18 Jacobs JP, Mavroudis C, Jacobs ML, et al. What is operative
mortality? Defining death in a surgical registry database: a
report of the STS congenital database taskforce and the joint
EACTS-STS congenital database committee. Ann Thorac
Surg 2006; 81:1937-1941
CHEST I 134 I 3 I SEPTEMBER, 2008 519