header advert
The Bone & Joint Journal Logo

Receive monthly Table of Contents alerts from The Bone & Joint Journal

Comprehensive article alerts can be set up and managed through your account settings

View my account settings

Full Access


National in-hospital morbidity and mortality trends after lumbar fusion surgery between 1998 and 2008

Download PDF


Increasing numbers of posterior lumbar fusions are being performed. The purpose of this study was to identify trends in demographics, mortality and major complications in patients undergoing primary posterior lumbar fusion. We accessed data collected for the Nationwide Inpatient Sample for each year between 1998 and 2008 and analysed trends in the number of lumbar fusions, mean patient age, comorbidity burden, length of hospital stay, discharge status, major peri-operative complications and mortality. An estimated 1 288 496 primary posterior lumbar fusion operations were performed between 1998 and 2008 in the United States. The total number of procedures, mean patient age and comorbidity burden increased over time. Hospital length of stay decreased, although the in-hospital mortality (adjusted and unadjusted for changes in length of hospital stay) remained stable. However, a significant increase was observed in peri-operative septic, pulmonary and cardiac complications. Although in-hospital mortality rates did not change over time in the setting of increases in mean patient age and comorbidity burden, some major peri-operative complications increased. These trends highlight the need for appropriate peri-operative services to optimise outcomes in an increasingly morbid and older population of patients undergoing lumbar fusion.

Posterior lumbar fusion is one of the most common procedures performed by spinal surgeons.1 Over recent decades the number of these procedures has increased dramatically, especially among older patients.2

Although lumbar fusion is considered relatively safe, the shift towards elderly and medically compromised patients is likely to affect peri-operative morbidity and mortality.3,4 However, there is a paucity of literature on trends in peri-operative outcomes over time. Knowledge about such trends allows health-care providers to assess the impact of changes in medical practice on outcomes and adjust the allocation of resources to meet changing conditions.

We used the Nationwide Inpatient Sample (NIS) to analyse linear trends in demographics as well as morbidity and mortality associated with posterior lumbar fusion surgery between 1998 and 2008. The NIS has been widely used in the past to analyse peri-operative morbidity and mortality,4,5 as well as to establish trends in lumbar fusion surgery.2 Given the rapid change in the study population and an increase in the use of the procedure, more recent data are required.

We hypothesised that significant changes in patient demographics and peri-operative major complications occurred between 1998 and 2008.

Materials and Methods

Data source

Nationwide Inpatient Sample (NIS) data files were analysed for this study. The NIS represents the largest all-payer inpatient database in the United States. Information on the design of the NIS is available on the internet.6,7 This study was exempt from review by the institutional review board, as the data used are sufficiently anonymised.

Selection of study sample and statistical methods

Data from 1998 to 2008 were accessed. Admissions for primary posterior lumbar fusion were identified using the International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) codes.8 The ICD-9 code 81.08 includes ‘lumbar and lumbosacral fusion, posterior technique’, ‘arthrodesis of lumbar or lumbosacral region: posterior (interbody) technique’, ‘posterolateral technique’, ‘posterior lumbar interbody fusion (PLIF)’ and ‘transforaminal lumbar interbody fusion (TLIF)’.

Linear trends in patient demographics, including mean age, comorbidity burden as indicated by the mean Deyo index9,10 and gender were analysed. The Deyo comorbidity index10 is a modification of the Charlson index9 and was developed to predict in-hospital outcomes based on a number of comorbidities when using administrative data, such as are used in this paper. In brief, a number of comorbidities, defined by their ICD-9 codes, are assigned a weight when present.11 The sum of these weights is converted to an overall score (the higher the score, the higher the chance of adverse outcomes). The Deyo index was originally validated in spinal patients, making it especially attractive for this study. Linear trends in discharge destination (routine versus non-routine, i.e. short-and long-term facilities and home health-care) were computed.

Pre-operative spinal diagnoses were tabulated and each entity was analysed over time. The diagnoses included degenerative disc disease, spinal stenosis, scoliosis, spondylolisthesis, trauma, tumour, multiple spine diagnoses (two or more) and other diagnoses. Rates of major complications were tabulated by identifying entries that listed ICD-9-CM codes consistent with post-operative cerebral infarction, sepsis, shock/cardiorespiratory arrest, acute myocardial infarction, cardiac complications (other than myocardial infarction), pneumonia and pulmonary embolism.12 In-hospital mortality rates were determined for each year. Morbidity and mortality rates were adjusted for changes in length of hospital stay by computing the number of events per 1000 inpatient days.

Statistical analysis

Weighted means and percentages were calculated and plotted for continuous and categorical variables, respectively. As the NIS represents only a stratified sample of approximately 20% of all annual hospitalisations in the United States, each actual entry in the database (unweighted data) is assigned a weight that allows for the extrapolation of that particular entry to generate a national estimate (weighted number). Different regression methods for trend analysis were used to characterise the linear association between the outcomes of time.13 In particular, Poisson regression, linear regression and logistic regression were applied to person–time data (i.e. in-hospital mortality per 1000 inpatient days), continuous data (i.e. mean age, mean comorbidity burden, mean length of stay) and binary data (i.e. incidences of spinal pathologies), respectively. Pearson’s correlation coefficient (r) was assessed to evaluate the association between the total number of lumbar fusion operations and time, and the association between the gender differences in proportions and time. The regression coefficient (β) associated with time, which illustrates the direction and the magnitude of the trend over time, is reported along with its p-value. A p-value < 0.05 was considered significant.

The linear trend analysis was performed using Stata version 11 (StataCorp, College Station, Texas) and other descriptive statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina).14 In order to facilitate data analysis and to obtain consistent estimates of mean and variance parameters taking into account the complex survey data setting, SAS procedures SURVEYMEANS and SURVEYFREQ were used for descriptive analyses. SVYSET and SVY LINEARIZED from Stata procedures were used for linear trend analysis.


A total of 1 288 496 primary posterior lumbar fusions were performed between 1998 and 2008, during which time the number of procedures performed per year increased almost threefold, to 182 981 in 2008 (r = 0.987, p < 0.0001) (Fig. 1). A similar increase was seen in the population growth-adjusted use. Use in the civilian population increased from 23 per 100 000 in 1998 to 60 per 100 000 in 2008 (r = 0.987, p < 0.0001) (Fig. 2). A significant increase in the mean age of the patients from 52.47 to 56.85 years (β = 0.496, p < 0.0001) and comorbidity burden from 0.32 to 0.55 was seen over time (β = 0.027, p < 0.0001) (Figs 3 and 4).

Fig. 1 
          Graph showing the total number of lumbar
fusions undertaken between 1998 and 2008 (r = 0.987; p <

Fig. 1

Graph showing the total number of lumbar fusions undertaken between 1998 and 2008 (r = 0.987; p < 0.0001).

Fig. 2 
          Graph showing the population growth-adjusted
use of primary posterior lumbar fusions undertaken per 100 000 civilians
in the United States (r = 0.987; p <

Fig. 2

Graph showing the population growth-adjusted use of primary posterior lumbar fusions undertaken per 100 000 civilians in the United States (r = 0.987; p < 0.0001).

Fig. 3 
          Graph showing the increase in mean patient
age at operation for each year (β = 0.496; p <

Fig. 3

Graph showing the increase in mean patient age at operation for each year (β = 0.496; p < 0.0001).

Fig. 4 
          Graph showing the increase in the mean
Deyo comorbidity index between 1998 and 2008 (β = 0.027; p <

Fig. 4

Graph showing the increase in the mean Deyo comorbidity index between 1998 and 2008 (β = 0.027; p < 0.0001).

The mean length of stay decreased from 5.1 days in 1998 to 4.7 days in 2008 (β = -0.098, p < 0.0001) (Fig. 5). The percentage of routine discharges to the patients’ customary residence remained stable at around 70% during this time.

Fig. 5 
          Graph showing the mean length of hospital
stay after lumbar fusion (β = -0.098; p <

Fig. 5

Graph showing the mean length of hospital stay after lumbar fusion (β = -0.098; p < 0.0001).

More women than men underwent lumbar fusion at any given time point. The gender gap increased over the period from 5% in 1998 to 13% in 2008 (Pearson’s correlation coefficient (between the gender differences in proportions and time) r = 0.91, p < 0.0001) (Fig. 6). The in-hospital mortality rates after primary posterior lumbar fusion per 1000 inpatient days decreased from 0.34 in 1998 to 0.30 in 2008, but this was not significant (β = -0.009, p = 0.584) (Fig. 7).

Fig. 6 
          Graph showing the proportions of the
genders over the study period 
(r = 0.91; p <

Fig. 6

Graph showing the proportions of the genders over the study period (r = 0.91; p < 0.0001).

Fig. 7 
          Graph showing the in-hospital mortality
rate per 1000 in-patient days (β = -0.009; p = 0.584).

Fig. 7

Graph showing the in-hospital mortality rate per 1000 in-patient days (β = -0.009; p = 0.584).

Table I gives a detailed analysis of linear trends in spinal pathology. Although the proportions of procedures addressing spinal stenosis and multiple spine diagnoses (≥ 2) each significantly increased over time (p < 0.0001), degenerative disc disease, scoliosis, spondylolisthesis, trauma, tumour and other diagnosis each decreased significantly (p < 0.0001).

Table I

Linear trend analysis showing trends in percentages of spinal pathologies among patients undergoing posterior lumbar spine fusion

Pathology 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Regression coefficient p-value
Degenerative disc disease 43.7 41.9 44.8 47.8 48.4 47.6 45.1 44.6 45.3 44.1 43.5 -0.006 < 0.0001
Spinal stenosis 10.4   9.7 10.7 10.0   9.4 9.8   10.0   9.8 9.9 10.7 11.3  0.011 < 0.0001
Scoliosis   1.7   2.1   1.9   2.0   1.5   1.2   1.4   1.4 1.2   1.2   1.1 -0.064 < 0.0001
Spondylolisthesis   6.8   7.9   7.1   6.0   6.0   6.0   5.5   5.5 5.1   4.7   4.2 -0.056 < 0.0001
Trauma   2.5   2.4   1.9   1.4 1  .4   1.7   1.9   1.3 1.6   1.6   1.2 -0.049 < 0.0001
Oncologic   0.4   0.3   0.3   0.3   0.4   0.3   0.2   0.3 0.2   0.3   0.2 -0.045 < 0.0001
Other   5.9   6.3   6.2   5.0   5.6   4.7   4.1   4.3 4.6   3.7   4.5 -0.045 < 0.0001
Multiple spinal diagnoses (≥ 2) 28.5 29.4 27.1 27.4 27.3 28.7 31.8 32.8 32.1 33.8 33.8  0.035 < 0.0001

The in-hospital complication rates after primary posterior lumbar fusion per 1000 inpatient days were also calculated. No significant changes in rates of peri-operative complications were observed for cerebral (β = 0.035, p = 0.051) and myocardial infarction (β = -0.007, p = 0.527); however, cardiac complications (other than myocardial infarction) (β = 0.059, p < 0.0001), pneumonia (β = 0.063, p < 0.0001), pulmonary embolism (β = 0.060, p < 0.0001), sepsis (β = 0.064, p < 0.0001) and shock/cardiorespiratory arrest each increased significantly. Detailed analysis of the rates of peri-operative major complications linear trends adjusted per 1000 in-hospital days is given in Table II.

Table II

Linear trend analysis showing the rate per 1000 in-hospital days of peri-operative complications after lumbar spinal fusion surgery

Complication 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Regression coefficient p-value
Cerebral infarction 0.34 0.28 0.26 0.28 0.23 0.27 0.28 0.41 0.33 0.30 0.42  0.035 0.051
Sepsis 0.58 0.62 0.54 0.58 0.55 0.75 0.89 0.70 0.93 0.84 1.03  0.064 < 0.0001
Shock/cardiac arrest 0.16 0.13 0.21 0.21 0.30 0.28 0.15 0.28 0.26 0.21 0.29  0.046 0.02
Acute myocardial infarction (AMI) 0.65 0.75 0.88 0.73 0.51 0.77 0.80 0.63 0.62 0.69 0.73 -0.007 0.527
Cardiac complications (non-AMI) 5.13 5.16 4.99 5.37 5.42 5.80 6.83 6.99 6.88 8.16 8.44  0.059 < 0.0001
Pneumonia 1.88 2.01 1.99 2.24 2.40 2.55 3.05 2.83 3.30 3.00 3.57  0.063 < 0.0001
Pulmonary embolism 0.31 0.47 0.47 0.41 0.48 0.38 0.61 0.63 0.53 0.65 0.69  0.059 < 0.0001


We analysed nationally representative data on linear trends of major morbidity and mortality in primary posterior lumbar fusion surgery between 1998 and 2008. Over this period the total number of operations increased significantly. Patient age and comorbidity burden also increased over time. During the ten-year study period the length of hospital stay decreased, whereas routine discharges to patients’ customary residence remained unchanged. Adjusted in-hospital mortality also remained unchanged, whereas the incidence of most peri-operative major complications increased.

Our data demonstrate a mean increase in the number of primary posterior lumbar fusions of 11.6% per year. Although regional differences in the rate of lumbar fusion in the United States have been described previously,15 our data confirm an absolute, as well as a population growth-adjusted, increase in these procedures. This increase has been reported previously.2,16,17 Further analysis of types of fusion demonstrated a 15-fold increase in complex spinal fusions between 2002 and 2007.18 This constant increase in the total number and complexity of lumbar spinal fusions might be attributed to advances in implant design and manufacture, surgical techniques and management strategies. Although the surgical indications for fusion may remain similar, new technologies and improvements in peri-operative management allow surgeons to perform complex operations on patients with higher comorbidity and advanced age.

We found an increase in mean patient age of over four years within one decade. This may be attributable to the ageing population in general as well as to a steeper increase in lumbar fusions in elderly people.2,18 The comorbidity burden also increased between 1998 and 2008. Age and the presence of comorbidities have previously been linked to increased rates of complications and mortality after surgical procedures.3,4,19 In this context, our finding of increasing mean patient age over time must be viewed with concern.

A reduction in length of hospital stay, as shown in our study, has been described in other fields of spinal surgery,20 possibly due to improved peri-operative management strategies, pain control and improved rehabilitation regimes. In addition, increasing financial pressure by insurance carriers is likely to be a significant factor. However, the rates of patient discharge to their customary residence remained stable over the ten-year study period, suggesting that advances in post-operative care may be partially responsible for this trend.

Significant gender differences were observed, with more women undergoing lumbar fusion at any given time. This gender gap increased over time to reach a level of 12.6% weighted towards women in 2008. Such increased rates in women undergoing primary and revision lumbar fusion have previously been shown.4,18,21

Unadjusted in-hospital mortality rates decreased from 0.17% in 1998 to 0.12% in 2008, and similar mortality rates were reported by Shen, Silverstein and Roth.3 These researchers also concluded that medically complicated and elderly patients are at an increased risk for in-hospital mortality and early post-operative complications. Fu et al reviewed over 10 000 lumbar operations and reported that all patients who died following these operations were over 60 years old.22 However, after adjusting for length of stay, the decrease in in-hospital mortality was not found to be significant. Our data represent unchanged adjusted in-hospital mortality, even though the mean age and comorbidity burden increased from 1998 to 2008. Shifts in the patient population and improvements in management might explain this stability of mortality rates. Although mortality risk might be increased in the elderly, this is counterbalanced by a significantly decreased mortality rate among younger patients.3 Furthermore, advances in peri-operative care might counterbalance the increased risk of peri-operative mortality posed by an older and medically unfit patient population. Our findings clearly underline the importance of ongoing research in optimising peri-operative management strategies. As there is a trend towards increasing comorbidity and age among patients undergoing lumbar fusion, future efforts to improve patient care, targeting multiple peri-operative factors, will have to be pursued in order to ensure satisfactory outcomes.

The analysis of major peri-operative complications is shown in Table II. The most significant increase was seen in the incidence of sepsis, cardiac complications other than acute myocardial infarction, pneumonia and pulmonary embolism. This increase in pulmonary and cardiac complications has already been reported in patients undergoing cervical fusion.20 Although pulmonary embolism is relatively rare following lumbar surgery and routine screening for detection of asymptomatic thrombosis might not be indicated,23 our data show a significant increase in pulmonary embolism over the past decade, and these findings are comparable with previously published data.24 Owing to the shift towards elderly patients, prospective studies should re-evaluate the risk of pulmonary embolism in patients undergoing lumbar surgery. It is possible that the increase in pneumonia can be attributed to the increase in the mean age and comorbidity burden. Interestingly, a retrospective analysis showed that > 90% of the post-operative deaths occurring after lumbar surgery were due to respiratory complications, myocardial infarction or sepsis.21

Our findings must be interpreted according to the limitations associated with the analysis of the NIS database. The clinical information gained from the NIS is limited. For example, the exact procedural details are unknown. This is important, as it is known that posterolateral, PLIF and TLIF procedures differ significantly in their risk–benefit profile.25 Furthermore, the database only allows interpretation of inpatient complications and mortality rates. The rates of peri-operative complications are likely to be higher than shown in this study because post-operative complications and deaths outside the hospital are not captured. Additionally, possible coding bias has to be considered. Using the ICD-9-CM coding system leaves a large number of alternative code options for each entity.

In general, there is a lack of detailed information relating to spinal surgery in the available databases, and thus it seems prudent that a specialty-specific registry be pursued. With such an approach more detailed and important information could be captured that might lead to opportunities to study outcomes and linear trends in a more comprehensive manner.

In conclusion, this study confirms a nationwide increase in the number of primary posterior lumbar fusions. Generally, patients presented at an older age and had an increased comorbidity burden over time. However, adjusted mortality rates remained stable between1998 and 2008, whereas non-myocardial infarction cardiac, pulmonary and septic complications increased. These findings suggest that although advances in peri-operative medicine may have a positive impact on the extreme outcome of mortality, the increases in many major complications reflect the increasing demands on the peri-operative health-care team brought about by increases in comorbidity burden and advanced age. Future endeavours in the peri-operative arena will need to focus on interventions targeted to maintain adequate outcomes in a patient population that is increasingly medically unfit and elderly.

Correspondence should be sent to Dr S. G. Memtsoudis; e-mail:

1 Garrett WE Jr , SwiontkowskiMF, WeinsteinJN, et al.American Board of Orthopaedic Surgery Practice of the Orthopaedic Surgeon: part-II, certification examination case mix. J Bone Joint Surg [Am]2006;88-A:660667. Google Scholar

2 Deyo RA , GrayDT, KreuterW, MirzaS, MartinBI. United States trends in lumbar fusion surgery for degenerative conditions. Spine2005;30:14411447. Google Scholar

3 Shen Y , SilversteinJC, RothS. In-hospital complications and mortality after elective spinal fusion surgery in the United States: a study of the nationwide inpatient sample from 2001 to 2005. J Neurosurg Anesthesiol2009;21:2130. Google Scholar

4 Ma Y , PassiasP, Gaber-BaylisLK, GirardiFP, MemtsoudisSG. Comparative in-hospital morbidity and mortality after revision versus primary thoracic and lumbar spine fusion. Spine J2010;10:881889. Google Scholar

5 Memtsoudis SG , VougioukasVI, MaY, Gaber-BaylisLK, GirardiFP. Perioperative morbidity and mortality after anterior, posterior, and anterior/posterior spine fusion surgery. Spine2011;36:18671877. Google Scholar

6 No authors listed. Healthcare Cost and Utilization Project (HCUP): overview of the Nationwide Inpatient Sample (NIS), 2008. http://www.hcup-us.ahrq.gov/nisoverview.jsp (date last accessed 1 November 2010). Google Scholar

7 No authors listed. Healthcare Cost and Utilization Project (HCUP): introduction to the HCUP National Inpatient Sample (NIS) 2008. http://www.hcup-us.ahrq.gov/db/nation/nis/2006NIS_INTRODUCTION.pdf (date last accessed 1 November 2010). Google Scholar

8 No authors listed. Healthcare Cost and Utilization Project (HCUP): HCUP Comorbidity Software. http://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp (date last accessed 1 November 2011). Google Scholar

9 Charlson ME , PompeiP, AlesKL, MacKenzieCR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis1987;40:373383. Google Scholar

10 Deyo RA , CherkinDC, CiolMA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol1992;45:613619. Google Scholar

11 No authors listed. Healthcare Cost and Utilization Project (HCUP): changes in the NIS sampling and weighting information. http://www.hcup-us.ahrq.gov/db/nation/nis/reports/Changes_in_NIS_Design_1998.pdf (date last accessed 1 November 2011). Google Scholar

12 Iezzoni LI , DaleyJ, HeerenT, et al.Identifying complications of care using administrative data. Med Care1994;32:700715. Google Scholar

13 Ely JW , DawsonJD, LemkeJH, RosenbergJ. An introduction to time-trend analysis. Infect Control Hosp Epidemiol1997;18:267274. Google Scholar

14 No authors listed. SAS institute: sample survey design and analysis. http://support.sas. com/rnd/app/da/new/dasurvey.html (date last accessed 1 November 2011). Google Scholar

15 Cherkin DC , DeyoRA, LoeserJD, BushT, WaddellG. An international comparison of back surgery rates. Spine1994;19:12011206. Google Scholar

16 Weinstein JN , LurieJD, OlsonPR, BronnerKK, FisherES. United States’ trends and regional variations in lumbar spine surgery: 1992-2003. Spine 2006;312707. Google Scholar

17 Cowan JA Jr , DimickJB, WainessR, et al.Changes in the utilization of spinal fusion in the United States. Neurosurgery2006;59:1520. Google Scholar

18 Deyo RA , MirzaSK, MartinBI, et al.Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in older adults. JAMA2010;303:12591265. Google Scholar

19 Coe JD , SmithJS, BervenS, et al.Complications of spinal fusion for scheuermann kyphosis: a report of the scoliosis research society morbidity and mortality committee. Spine2010;35:99103. Google Scholar

20 Marawar S , GirardiFP, SamaAA, et al.National trends in anterior cervical fusion procedures. Spine2010;35:14541459. Google Scholar

21 Cassinelli EH , EubanksJ, VogtM, et al.Risk factors for the development of perioperative complications in elderly patients undergoing lumbar decompression and arthrodesis for spinal stenosis: an analysis of 166 patients. Spine2007;32:230235. Google Scholar

22 Fu KM , SmithJS, Polly DW Jr, et al.Morbidity and mortality in the surgical treatment of 10,329 adults with degenerative lumbar stenosis. J Neurosurg Spine2010;12:443446. Google Scholar

23 Smith MD , BresslerEL, LonsteinJE, et al.Deep venous thrombosis and pulmonary embolism after major reconstructive operations on the spine: a prospective analysis of three hundred and seventeen patients. J Bone Joint Surg [Am]1994;76-A:980985. Google Scholar

24 Smith JS , FuKM, PollyDW Jr, et al.Complication rates of three common spine procedures and rates of thromboembolism following spine surgery based on 108,419 procedures: a report from the Scoliosis Research Society Morbidity and Mortality Committee. Spine2010;35:21402149. Google Scholar

25 Mehta VA , McGirtMJ, Garcés AmbrossiGL, et al.Trans-foraminal versus posterior lumbar interbody fusion: comparison of surgical morbidity. Neurol Res2011;33:3842. Google Scholar

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.