SPINE Volume 39, Number 5, pp 417-423 ©2014, Lippincott Williams Sc Wilkins
HEALTH SERVICES RESEARCH
Complications After Lumbar Spine Surgery Between Teaching and Nonteaching Hospitals Sreeharsha V. Nandyala, BA, Alejandro Marquez-Lara, MD, Steven J. Fineberg, MD, Hamid Hassanzadeh, MD, and Kern Singh, MD
Study Design. Retrospective national database analysis. Objective. A national population-based database was analyzed to characterize the perioperative complications of lumbar spine procedures performed at teaching and nonteaching hospitals. Summary of Background Data. Perception biases exist regarding the complications of lumbar spine surgery based upon the hospital teaching environment. M e t h o d s . Data from the Nationwide Inpatient Sample was queried from 2 0 0 2 - 2 0 1 1 . Patients undergoing an anterior lumbar interbody fusion, posterior lumbar interbody fusion, anterior/posterior lumbar fusion, or lumbar decompression to treat lumbar degenerative pathology were identified and separated into cohorts based upon the teaching status of the hospital. Patient demographics, Charlson Comorbidity Index, length of stay, complications, mortality, and costs were assessed. Results. A total of 658,616 lumbar procedures were identified from 2 0 0 2 - 2 0 1 1 , of which 367,875 (55.9%) were performed at teaching hospitals. An older patient population comprised the teaching hospital cohort and demonstrated a greater comorbidity burden than the nonteaching group (Charlson Comorbidity Index 2.90 vs. 2.55; P < 0.001). In addition, the teaching hospital cohort was associated with a significantly greater number of multilevel fusion cases ( P < 0.001) and incurred a greater mean length of stay (3.7 vs. 3.0 d; P < 0.001). Patients treated atteaching hospitals demonstrated a significantly greater incidence of postoperative pulmonary embolism, deep vein thrombosis, infection, and neurological complications than the nonteaching cohort (P < 0.001). Overall, there were no significant differences in the mean total hospital costs or mortality between the hospital cohorts. Regression analysis
From the Department of Orthopaedic Surgery, Rush University Medical Center, Chicago, IL. Acknowledgment date: September 4, 2013. First revision date: October 21, 2013. Second revision date: November 18, 2013. Acceptance date: November 22, 2013. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: board membership, consultancy, royalties. Address correspondence and reprint requests to Kern Singh, MD, Department of Orthopaedic Surgery, Rush University Medical Center, 1611 W. Harrison St, Suite 300, Chicago, IL 60612; F-mail:
[email protected] DOI : 10.1097/BRS.OOOOOOOOOOOOOl 49 Spine
demonstrated that teaching status was not a significant predictor of mortality (OR, 1.02; confidence interval 0.8-1.2; P = 0.8). Conclusion. Patients treated in teaching hospitals for lumbar spine surgery incurred a longer hospitalization and a greater incidence of postoperative complications including pulmonary embolism, deep vein thrombosis, infection, and neurological events. These findings may be explained by an increased complexity of procedures performed at teaching hospitals along with an older and a more comorbid patient population. Despite these differences, the teaching status was not a significant predictor of in-hospital mortality after a lumbar spine surgery. K e y w o r d s : teaching hospital, lumbar spine surgery, complications, mortality, anterior lumbar fusion, posterior lumbar fusion, lumbar decompression, risk factors.
Level of Evidence: 3 Spine 2014;39:417-423
L
umbar decompression and lumbar fusion procedures are frequently performed in both academic and private institutions. A considerable amount of speculation is focused upon the differences in outcomes and the quality of health care provided between teaching and nonteaching hospitals. Perception biases against teaching institutions have traditionally centered upon the potential for errors by trainees. ^'The published literature provides conflicting reports regarding the surgical outcomes based upon hospital teaching status. Some authors have suggested that teaching hospitals have superior surgical outcomes with regard to complex procedures than nonacademic centers.'"' In contrast, others have demonstrated a greater incidence of postoperative mortality, complications, and costs associated with academic institutions.'"^'- The majority of these reports involve emergent or oncological procedures, which may have confounded the medical outcomes. In summary, there is a lack of evidence comparing the perioperative surgical outcomes of elective lumbar spine surgery in teaching versus nonteaching institutions. In this national study, a population-based database was analyzed to characterize the differences based upon the hospital teaching status about patient demographics, clinical outcomes, mortality, and costs of lumbar spine surgery. www.spinejournal.com
417
Spine
HEALTH SERVICES RESEARCH
MATERIALS AND METHODS The Nationwide Inpatient Sample (NIS) database is part of the Healthcare Cost and Utilization Project (HCUP) and is governed by the Agency for Healthcare Research and Quality.'^ The NIS represents the largest all-payer health care database in the United States, approximating a 20% stratified sample of all hospital discharges. Fach entry contains data from a single hospital admission. The NIS issues admission diagnosis, in-hospital complications, and procedural data using the International Classification of Disease, Ninth Edition, Clinical Modification (ICD-9-CM), codes.
Data Collection
Lumbar Surgery in Teaching Hospitals • Nandyala et al
fusion of 3 to 7 levels (81.63), and fusion of 8 or greater levels (81.64). For this analysis, we used the cost-to-charge conversion files as published by the HCUP to convert all "charges" to "costs." All dollar amounts reported in this manuscript are reflective of the hospital costs only. At the time of this study, the hospital cost to charge ratios were not available for 2011. Therefore, the costs that are reported in this study reflect the 2002-2010 NIS data.
Data Analysis Statistical analysis was performed with SPSS version 20 (Chicago, IL). Student t test was used to compare discrete variables and X" test for categorical data. An alpha level of 0.001 or less denoted statistical significance, given the large sample size. Multivariate logistic regression with a 95% confidence interval was performed to identify the independent risk factors for mortality.
Data from the NIS was obtained from 2002 to 2011. Patients undergoing an anterior lumbar interbody fusion (81.06), posterior lumbar interbody fusion (81.07-81.08), APLF (81.06 + 81.07-81.08 and/or 81.36-81.38), or lumbar decompression without fusion (80.51, 03.09) were identified. The procedural codes 03.09 and 80.51 are not specific to the lumbar spine. To include only lumbar procedures, entries RESULTS were further stratified with concurrent ICD-9-CM diagnosis A total of 658,616 lumbar procedures were identified from codes for degenerative etiologies of the lumbar spine includ2002-2011, of which 367,875 (55.9%) procedures were ing the following: lumbar radiculopathy, herniated nucleus performed at teaching hospitals (Table 1). The majority of pulposus, degenerative disc disease, and lumbar spinal stenocases were lumbar decompressions accounting for 354,226 sis. Patients younger than 18 years were excluded from the (53.8%) procedures, followed by 222,949 (33.9%) posterior study. Only raw numbers are reported in this analysis rather lumbar interbody fusions, 48,891 (7.4%) anterior lumbar than weighted data. interbody fusions, and 23,250 (3.5%) APLFs (Table 2). Patients were then separated into 2 cohorts based on the Several demographic differences were characterized teaching status of the hospital. The NIS designates a hospital between the hospital cohorts (Table 1). The teaching hosto have teaching status if the following criteria are met: (1) pital cohort was significantly older (56.7 vs. 55.6 yr; P < approval for residency training by the Accreditation Council 0.001) and demonstrated a predilection toward more males for Graduate Medical Fducation, (2) member in the Council than those treated in nonteaching institutions. In addition, of Teaching Hospitals, or (3) has a resident to beds ratio of the patients treated in academic institutions demonstrated 0.25 or greater.'^ a significantly greater comorbidity burden (mean CCI, 2.90 Age, sex, race, comorbid risk factors, primary payer, vs. 2.55; P < 0.001) and comprised a greater proportion of and disposition were assessed between the hospital cohorts. Medicaid beneficiaries (P < 0.001). Comorbidities were quantified using a modified Charlson Overall, the teaching hospitals performed a greater perComorbidity Index (CCI) with data from the NIS Disease centage of 3 to 7 levels (12.5 vs. 9.9%) and 8-h level (3.3 vs. Severity Measure Files.'"^ The CCI predicts the 10-year mor0.7%) lumbar fusions than the nonacademic institutions (P < tality of patients based upon 22 comorbidities and age.'^ 0.001) (Table 1). Similarly, in the posterior lumbar interbody Length of hospitalization, hospital costs, and the incidence fusion (11.9 vs. 10.6%) and APLF (14.2 vs. 12.9%) groups, of in-hospital complications and mortality were compared teaching hospitals performed a greater percentage of 3 -I- level between the cohorts. fusions ( P < 0.001) (Table 2). We used the "comorbidity variables" as provided by the There were notable differences in the early postoperaNIS Disease Severity Measures Files. This ensured that we distive complications between the hospital cohorts (Table 1). tinguished preoperative comorbidities {e.g.., CM_BLDLOSS, The teaching hospital cohort demonstrated a significantly CM_COAG, CM_RFNLFAIL, etc.). These comorbidity varigreater incidence of postoperative pulmonary embolism (1.81 ables are available for 2002 to 2011. By using these variables, vs. 1.26 per 1000), deep vein thrombosis (2.27 vs. 1.34 per we were able to distinguish between preoperative comorbidi1000), wound infections (4.02 vs. 3.41 per 1000), and neuroties and in-hospital complications (identified through ICD-9 logical events (7.99 vs. 6.70 per 1000) (Table 1). In teaching coding). To identify the primary admission diagnosis as degenhospitals, the APLF technique was associated with the greaterative lumbar spine disease, we only included patients with est incidence of postoperative complications (90.5 per 1000) codes for degenerative pathology if they were listed under the (Table 2). Although the teaching hospital cohort demonvariable "DXl." This variable is designated as the primary, or strated a greater incidence of postoperative mortality, this did admission, diagnosis variable by the NIS. not reach statistical significance in any of the surgical cohorts. The number of levels fused was categorized using ICDThe patient discharge disposition did not significantly vary 9-CM codes for fusion of 1 to 2 intervertebral levels (81.62), between the hospital cohorts. 418
www.spinejournal.com
March 2014
Lumbar Surgery in Teaching Hospitals • Nandyala et al
HEAETH SERVICES RESEARCH
TAE^ ^ ^ H J e n t Characteristics, Complicatid
TABLE Msmírih
^ ^ ^ n d Mortality of Lumbar Spine | ^ ^ ^ Surgery at Teaching and Nontea^W m ,
Ho«jnitak-Fm|f||ï|îf]||f.^^2|)t1
•
..• -.•Í
NonTeaching teaching
P
Count (n)
367,875
290,741
Male (%)
51.1
50.7
1
.55
o.
5
.001
L E
LO
CO
CM
54.04
Tea chii
PI
ro Lri
LO LO
b
q
.70
ce
LO
b
0
u
K
Lfl
0
.89
DC
1
d V
C
.86
LO
q d V
.08
LO
^^ vO
o V
0
.47
f\
O O
.00
C •?
•5 "5
d V
o o
.00
CO
O O
.94
DC C
•
û
o V
.00
o
.00
O Q,
Lumbar Surgery in Teaching Hospitals • Nandyala et al
54.5
Spine
Kj
'?
LO
u
Q^
CO .^ '^
(3
rM uo LO
0^ LO LO
rM
rM
LO
oj
CN
u-j rM
'~; "^
rM
0
rM
b
b
•
0
1
0
d
^
CN
b
LO [\
V
rM
l\
LO
CO
.
CO '
ro "^
r\
rM
co" rM
rsj
LO
cq
rN
rsi
0
420
www.spinejournal.com
S
CN
CM
-Si
cd
ro
ro
ao
ro
• s
01
1 ro
Lu D.
u ro
Cardi
ro
tn 0 U c ro
-S
E
0)
0 ro
Infection
•JD C OJ
c
ro
DVT
-C
ompli cations (per
01
"o
S-
, ^
0 0 0
— ro
-1- lev(2I fusior
) juno U
îmale
es
tean c omorbi dity s
y
o^ aj ro
b
r-i
b
u
ea
rM
CM
(\
0) tyi ro O
LU
b
a; I
c
u
1
ty (per 1000
ro
ro 0
S
ro
Neur ologica
ro
t
b
—J
LO
T
cr. ro ro ro""
en
b
in thiom
0
10.
ro
b
DVT,
,—
ro
.00
.00
rM
V
28,9
ro
o O V b
'^'^•
ro
LO
o V
13.
Tea
u
LO
o b
orrhage
oc
rM
1 '*' •
c
c •=
No
ALI
oc
1 teacP 1 25,5
1 LL
O
1
b
1 52.
E
iE
o o
a.
.00
^ ^
nter.ior
1
0
ro
onary
oc
.00
m
CL
ro •t 0
U March 2014
Spine
HEALTH SERVICES RESEARCH
Lumbar Surgery in Teaching Hospitals • Nandyala et al
OR
95% CI
P
Teaching hospital
1.02
0.8-1.2
0.8
3-1- levels
1.2
0.95-1.5
0.13
CHE
2.2
1.5-3.3