Improved capture of adverse events after spinal surgery procedures with a longitudinal administrative database

Full access

OBJECT

The significant medical and economic tolls of spinal disorders, increasing volume of spine surgeries, and focus on quality metrics have made it imperative to understand postoperative complications. This study demonstrates the utility of a longitudinal administrative database for capturing overall and procedure-specific complication rates after various spine surgery procedures.

METHODS

The Thomson Reuters MarketScan Commercial Claims and Encounters and the Medicare Supplemental and Coordination of Benefits database was used to conduct a retrospective analysis of longitudinal administrative data from a sample of approximately 189,000 patients. Overall and procedure-specific complication rates at 5 time points ranging from immediately postoperatively (index) to 30 days postoperatively were computed.

RESULTS

The results indicated that the frequency of individual complication types increased at different rates. The overall complication rate including all spine surgeries was 13.6% at the index time point and increased to 22.8% at 30 days postoperatively. The frequencies of wound dehiscence, infection, and other wound complications exhibited large increases between 10 and 20 days postoperatively, while complication rates for new chronic pain, delirium, and dysrhythmia increased more gradually over the 30-day period studied. When specific surgical procedures were considered, 30-day complication rates ranged from 8.6% in single-level anterior cervical fusions to 27.3% in multilevel combined anterior and posterior lumbar spine fusions.

CONCLUSIONS

This study demonstrates the usefulness of a longitudinal administrative database in assessing postoperative complication rates after spine surgery. Use of this database gave results that were comparable to those in prospective studies and superior to those obtained with nonlongitudinal administrative databases. Longitudinal administrative data may improve the understanding of overall and procedure-specific complication rates after spine surgery.

ABBREVIATIONSCPT = Current Procedural Terminology; ICD-9-CM = International Classification of Disease, Ninth Revision, Clinical Modification; NIS = National (Nationwide) Inpatient Sample.

OBJECT

The significant medical and economic tolls of spinal disorders, increasing volume of spine surgeries, and focus on quality metrics have made it imperative to understand postoperative complications. This study demonstrates the utility of a longitudinal administrative database for capturing overall and procedure-specific complication rates after various spine surgery procedures.

METHODS

The Thomson Reuters MarketScan Commercial Claims and Encounters and the Medicare Supplemental and Coordination of Benefits database was used to conduct a retrospective analysis of longitudinal administrative data from a sample of approximately 189,000 patients. Overall and procedure-specific complication rates at 5 time points ranging from immediately postoperatively (index) to 30 days postoperatively were computed.

RESULTS

The results indicated that the frequency of individual complication types increased at different rates. The overall complication rate including all spine surgeries was 13.6% at the index time point and increased to 22.8% at 30 days postoperatively. The frequencies of wound dehiscence, infection, and other wound complications exhibited large increases between 10 and 20 days postoperatively, while complication rates for new chronic pain, delirium, and dysrhythmia increased more gradually over the 30-day period studied. When specific surgical procedures were considered, 30-day complication rates ranged from 8.6% in single-level anterior cervical fusions to 27.3% in multilevel combined anterior and posterior lumbar spine fusions.

CONCLUSIONS

This study demonstrates the usefulness of a longitudinal administrative database in assessing postoperative complication rates after spine surgery. Use of this database gave results that were comparable to those in prospective studies and superior to those obtained with nonlongitudinal administrative databases. Longitudinal administrative data may improve the understanding of overall and procedure-specific complication rates after spine surgery.

The medical and economic burdens of spine disorders pose a growing challenge for patients, medical providers, and society. In 2005, 33.3 million US adults (15%) had spinal ailments whose treatment and other expenditures resulted in an estimated total cost of $85.9 billion. The most common International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses of these spinal disorders included “other and unspecified disorders of the back” (52.9%), “intervertebral disc disorders” (15.9%), and “sprains or strains of the back” (9.3%). Expenditures for managing or treating these conditions, after adjustment for age, sex, and inflation, increased by 65% from 1997 to 2005, but health outcomes did not improve commensurate with these rising costs.12 One strategy to improve quality of spine surgical care is to gain improved and generalizable knowledge of adverse events after spine surgery to aid in their prevention and in improved patient selection.

As the number of surgical treatments for spinal disorders increase and new health care regulations emphasize specific postoperative metrics, it is critical to understand the array of complications and expected incidences that may occur after surgical intervention. Although the incidences of complications and adverse events after spine surgery are becoming more defined, a large amount of disparate findings remains that varies by surgical level and study design.14 Variation in reporting and analysis both lead to an unclear understanding of the complications after spine surgery.

Administrative databases have been used extensively to assess complications associated with spine surgery. The National (Nationwide) Inpatient Sample (NIS) is a widely used database compiled from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality (see http://www.hcup-us.ahrq.gov). The NIS offers patient-level data, but only for individual hospital admissions. Because of a lack of individual patient identifiers that may be tracked over time after discharge or transfer, follow-up in the NIS database is lost. Cases of readmissions cannot be captured, and cases of wound infections and other complications that may be present after discharge cannot be assessed.

Longitudinal databases offer patient-specific identifiers that may be tracked over time.19 Hence, health care claims after discharge, hospital readmissions, and adverse events developing after the time of index procedure are captured. This capture may provide a more accurate assessment of any postoperative adverse events.

Improving the understanding of procedure-specific complication rates is useful to surgeons, patients, ancillary providers, and payers. Here, we show that longitudinal databases enable capture of overall and procedure-specific complications at rates approaching those observed in prospective studies. This study also places our longitudinal data within the context of other retrospective studies of adverse events after spine surgery. Longitudinal databases provide a robust collection of patient data and offer a valuable retrospective assessment of complications occurring in the perioperative and postoperative periods of spine surgery.

Methods

Literature Review

We searched the National Center for Biotechnology Information PubMed database for studies analyzing complications, adverse events, and outcomes after spine surgery. We initially queried the PubMed database by using the Boolean operator “AND” with the terms “spine surgery,” “adverse event,” and “complication.” We expanded our survey to include searches for specific events that we analyzed with longitudinal data and queried the database for procedure and complication combinations such as “lumbar fusion” AND “hematoma.” We then selected studies that considered complications investigated in our present analysis. We included both prospective and retrospective studies that drew data from several different databases and patient samples. Articles in languages other than English and studies conducted to record events not analyzed in our present study were excluded. This selection resulted in 20 studies whose scope and data represented a number of complications after spine surgery, diverse databases, and different study designs.

Study Design and Database

This study has been approved by the Stanford Institutional Review Board. All patient information and followup data were obtained and analyzed in accordance with the Helsinki Declaration, patient privacy was protected according to the US Health Insurance Portability and Accountability Act (HIPAA), and all procedures were in compliance with an institutional review board-approved protocol.

We performed a retrospective observational administrative database study of records from patients in the United States who underwent spine surgical procedures. Inpatient and outpatient longitudinal data for this study were obtained from the Thomson Reuters MarketScan Commercial Claims and Encounters and the Medicare Supplemental and Coordination of Benefits database from 2006 through 2010. These annual data sets include data from 100 payers and comprised inpatient, outpatient, and pharmacy services from large employers, health plans, governmental, and public organizations.

Cohort Selection

We queried the MarketScan database using Current Procedural Terminology (CPT) and ICD-9-CM codes (Supplemental Table 1). In the studied period between 2006 and 2010, there were 188,965 patients who met our primary inclusion criteria.

Complication Assessment

To define the time point of complication onset, we chose the date of surgery according to the presence of a CPT code indicating an operative procedure. Patients were excluded if they were in an associated diagnosis-related group or if their condition had an ICD-9-CM code indicating trauma or neoplastic involvement. The aggregation format in MarketScan allowed for identification of the initial, or index, hospital stay for a given procedure. To recreate the limitations of incident-only, nonlongitudinal data, in our “index” group we truncated follow-up to the initial hospitalization where the procedure in question was performed. This eliminated the potential benefits of the longitudinal data capturing complication rates associated with the procedure at follow-ups. The rates of complications at the index hospitalization were then compared with the rates at various postoperative time points. These postoperative complications were identified through readmissions, clinic visits, and other claims not registered during the patient’s initial inpatient stay.

The complications were chosen according to clinical relevance and because they were most commonly assessed in the literature. To avoid attributing preexisting diagnoses as a new onset in the analysis, a postoperative ICD-9-CM code in the patient’s record was only considered if it was not present within a year before the operation. Complications for which this restriction applied included chronic pain, deep vein thrombosis, pulmonary embolism, dysphagia, dysrhythmia, neurological complications, and pulmonary complications.

To measure the increase in captured complication rates, time points were chosen at which to calculate the cumulative frequency of each complication within 30 days postoperatively. The index complication rate represented the incidence of a complication during the operative hospital stay. The rate at each time point (5, 10, 20, or 30 days postoperatively) represented the percentage of the patients who had a given complication diagnosed on the basis of administrative claims data within that postoperative window. If a complication occurred before discharge, and if the length of stay was longer than a given time point, the complication was counted as occurring within the initial stay. From these rates, we calculated the percent increase relative to the index rate for each complication.

Statistical Analysis

Data were prepared and analyzed with SAS software (version 9.3; SAS Institute, Inc.).

Results

Literature Review of Complication Rates in Spine Surgery

We identified 19 studies that reported spine surgery procedures and complications comparable to those considered in the current analysis (Table 1). Studies varied by procedure performed, adverse events recorded, sample size, time point of outcomes, and type of study. Of these 20 studies, 13 had been performed retrospectively, and 13 drew on large national databases for patient data. The studies’ investigators determined and presented the rates of both overall and specific complications for a range of spine surgery procedures and follow-up periods. Pulmonary (13.0%) and hematological (10.8%) complications were the more common adverse events in spine surgery,10 while pulmonary embolisms (1.0%) and wound-related complications (1.2%–3.2%) were the least common.6,7

TABLE 1.

Summary results of literature review of postoperative spine surgery complications

Authors & YearPMIDProcedureComplicationNo. of PtsTime PointNo. of Pts w/ Complication (%)Study TypeDatabase
Yadlaetal., 20112221445020Thoracolumbar opAny adverse event that produced transient or permanent sequelae12830 days59.4ProspectiveNA
Campbell et al., 201021492571Ant & pst cervical spine opAny adverse event that produced transient or permanent sequelae11930 days44.5ProspectiveNA
Yadla et al., 201020809731Spine opAny adverse event that produced transient or permanent sequelae24830 days53.2ProspectiveNA
Yadlaetal., 20112320124909Cervical spine opAny adverse event that produced transient or permanent sequelae12130 days47.1ProspectiveNA
Bazaz et al., 200212435974Ant cervical spine opDysphagia22430 days50.2ProspectiveNA
2 mos32.2
6 mos17.8
12 mos12.5
Lee et al., 201222245448Spine opCardiac15912yrs8.4ProspectiveSpine End Results Registry (UW local registry)
Pulmonary13.0
Gastrointestinal3.9
Neurological7.4
Hematological10.8
Urological9.2
Chen et al., 201323848353Same-day combined anteroposterior cervical spine opDysphagia13Index 12 mos43.4RetrospectiveLocal patient series
110.0
Nandyala et al., 201424216394Spinal fusion opPeriop vision loss541,4850.02RetrospectiveNIS
Veeravagu et al., 201424126076Single-level ACDF Multilevel ACDFPostop infection, wound dehiscence, chronic pain, DVT, PE, or dysphagia12,74430 days2.2RetrospectiveMarketScan (2006–2010)
16,0333.0
Lovecchio et al., 201424108298ACDFWound, UTI, blood transfusion, sepsis/ shock, or any other systemic complication232030 days2.9RetrospectiveACS NSQIP
Deyo et al., 201323778366Op for lumbar stenosisWound12,15430 days3.2RetrospectiveACS NSQIP
Medical30 days2.1
Death90 days0.6
Fineberg et al., 201323462577ACF w/ BMPDysphagia, postop shock, neurological, CSF leak, hemorrhage/hematoma, cardiac, infection, or DVT/PE213,4216.0RetrospectiveNIS
ACF w/o BMP34.0
PCF w/ BMP20,3348.4
PCF w/o BMP7.3
Chitale et al., 201222960417Lumbar spine opInfection, pneumonia, renal failure, myo-cardial infarction, pulmonary complications, neurological complications, CHF, DVT/PE, hematoma, wound complications, UTI, cardiac dysrhythmia, seroma, or delirium185,32713.6RetrospectiveNIS
Lumbar spine op185,32713.6
Pugely et al., 201322814304Inpatient single-level lumbar discectomyUTI, bleeding requiring transfusion, graft failure, sepsis, septic shock, wound, pulmonary, neurological, hematologi-cal, cardiac, or renal complication265830 days5.4RetrospectiveACS NSQIP
Outpatient single-level lumbar discectomy16523.5
Patii et al., 201424291409Interspinous deviceDVT/PE, pulmonary, systemic, wound infections, or renal, cardiac, pulmonary, or neurological etiologies174Index 90 days3.5RetrospectiveMarketScan (2007–2009)
3.5
Laminectomy174Index 90 days7.5
9.2
Smith et al., 201121192289Correction of thoracolumbar fixed sagittal plane deformityOverall578NA29.4RetrospectiveSRS M&M database
Wound infection3.8
Neurological deficit3.8
PE1.0
Epidural hematoma1.4
Wound hematoma1.6
Fu et al., 201120739916Op Tx of pediatrie isthmic & dysplastic spondylolisthesisWound infection, neurological injury, implant related, durai tear, DVT/PE605NA10.4RetrospectiveSRS M&M database
Smith et al., 201020581760Lumbar microdiscectomyWound, hematoma, DVT/PE, acute neurological, cardiorespiratory96924yrs3.6RetrospectiveSRS M&M database
ACDF67352.4
Lumbar stenosis decompression10,3297.0
Deyo et al., 201020371784Lumbar decompression, simple fusion, & complex fusionMajor medical complications Wound complication32,15230 days3.1RetrospectiveMedicare claims data (2002–2007)
1.2

ACDF = anterior cervical discectomy and fusion; ACF = anterior cervical fusion; ACS NSQIP = American College of Surgeons National Surgical Quality Improvement Program; ant = anterior; BMP = bone morphogenetic protein; CHF = congestive heart failure; DVT = deep venous thrombosis; NA = not applicable or not available; PCF = posterior cervical fusion; PE = pulmonary embolism; PMID = PubMed identifier; pst = posterior; Pts = patients; SRS M&M = Scoliosis Research Society Morbidity and Mortality; Tx = treatment; UTI = urinary tract infection; UW = University of Washington.

In a retrospective analysis, Chitale et al. observed that lumbar spine surgery resulted in greater complication rates (13.6%) than cervical spine surgery (4.5%).4 In a prospective assessment, Yadla et al. reported a similar finding, noting that complication rates were higher (59.4%) in operations in the thoracolumbar spine than in the cervical spine (47.1%).22,23 Several of the prospective studies identified reported overall complication rates ranging from 44.5% to 59.4% at 30 days postoperatively. These rates were uniformly greater than overall complication rates determined by any retrospective study at 30 days postoperatively, including our data (27%). In a prospective study by Bazaz et al., the rate of dysphagia after anterior cervical spine surgery fell greatly from 50.2% to 12.5% over the 1-year study period.1 A retrospective study by Patil et al. reported complication rates to be relatively stable or slightly increased between immediately after the operation and 90 days postoperatively.15

Total Complication Rates After Spine Surgery

We queried the MarketScan database for the 12 most common complications after spine surgery (Table 2). For a range of spinal procedures (Table 3), we determined rates of adverse events at the index time point. We also estimated cumulative complication rates at 5, 10, 20, and 30 days postoperatively to measure the increased capture of complications otherwise undetected by databases limited to the immediate inpatient setting.

TABLE 2.

Rates of postoperative spinal surgery complications immediately after the operation (index) and in the 30-day period after discharge

ComplicationComplication Rate (%)
Index5 Days10 Days20 Days30 Days
Wound dehiscence0.210.250.370.660.86
Wound infection1.21.41.93.03.6
Wound hematoma1.01.11.31.61.8
Other wound complication0.410.500.731.21.5
New chronic pain1.11.21.21.41.5
Delirium0.340.380.420.490.56
DVT1.51.62.02.42.8
PE0.830.880.971.11.3
Dysphagia1.21.41.51.82.1
Dysrhythmia7.47.78.18.69.1
Neurological complication3.43.53.84.14.4
Pulmonary complication5.86.16.47.07.5
All13.614.017.220.022.8
Increase Over Index Rate (%)
Wound dehiscenceNA16.376.4213.5308.2
Wound infectionNA13.156.3139.9194.4
Wound hematomaNA10.531.161.780.5
Other wound complicationNA22.477.7190.3272.4
New chronic painNA5.913.227.439.2
DeliriumNA10.923.046.064.2
DVTNA10.332.165.089.0
PENA5.717.035.450.9
DysphagiaNA12.427.050.772.5
DysrhythmiaNA4.28.516.423.0
Neurological complicationNA5.111.622.030.2
Pulmonary complicationNA5.010.520.729.8
AllNA5.413.626.735.8
TABLE 3.

Postoperative complication rates by procedure at the index time point and in the 30-day period after discharge

ProcedureNo. of PtsComplication Rate (%)
Index5 Days10 Days20 Days30 Days
Ant thoracolumbar817814.414.916.218.419.7
Lumbar decompression
 Pst61,5918.08.59.511.112.1
 Pst & ant66812.713.214.215.716.9
Lumbar fusion
 Single level, pst77739.19.610.812.814.0
 Single level, pst & ant147211.812.413.615.917.3
 Single level, pst, w/ decompression26,4169.29.810.912.914.1
 Single level, pst & ant, w/ decompression165011.011.612.815.416.6
 Multiple levels744716.717.318.921.523.3
 Multiple levels, pst & ant190220.521.522.925.727.3
 Multiple levels, pst, w/ decompression21,99113.614.516.118.620.2
 Multiple levels, pst & ant, w/ decompression141215.616.618.622.223.8
ACDF
 Single level, w/ decompression21,6725.66.47.07.88.6
 Multiple levels, w/ decompression25,5087.28.39.110.010.8
Cervical corpectomy
 Single level23738.99.510.511.112.1
 Multiple levels451411.612.914.115.516.5
 Multiple levels, w/ decompression116812.213.714.616.017.0

Complication rates for each observed adverse event increased from the index to the 30-day time point (Table 2). However, the increases of the individual complications varied over time. Wound dehiscence, infection, and other wound complications showed large relative increases in frequency, with the steepest increases occurring between 10 and 20 days postoperatively, while complication rates for new chronic pain, delirium, and dysrhythmia increased more gradually over the 30-day postoperative period. Nonwound-related complications occurred earlier after the index surgery, whereas wound-related events were both present and developed over the course of the 30-day postoperative period. The overall complication rate for the all spine interventions analyzed here was 13.6% at the index time point and increased to 22.8% at 30 days postoperatively.

Differential Increases in Complication Rate

We identified 16 different spine procedures in the MarketScan database and calculated their complication rates at discrete time points ranging from the index time point to 30 days postoperatively (Table 3). This analysis indicated variation in complication rates according to surgical approach, number of levels operated on, and use of decompression. For instance, the complication rates at 30 days postoperatively ranged from 8.6% in single-level anterior cervical fusions to 27.3% in multilevel combined anterior and posterior lumbar spine fusions. Furthermore, the complications varied in the level of increase relative to the index rate, with some adverse events showing higher rates 30 days postoperatively relative to the index rate than others. Overall complication rates of the surgical procedures were also calculated to assess variability in complication rates based on procedure. The overall procedure-specific complications also varied in frequency over time.

Discussion

In this study, we demonstrate the value of a longitudinal database for capturing rates of complications arising from spine surgery. We also present these results in the context of both prospective and retrospective studies, which have investigated complications of a broad range of spinal procedures.

Database Comparison

Effective use of longitudinal databases stands to improve the quality of retrospective studies on spine surgery outcomes. The advantages of observational data over randomized controlled trial data include being able to study a more diverse patient population and observing the effects of a treatment under real-world, rather than idealized, conditions. Although randomized trials are the gold standard for medical evidence, the surgical community is increasingly recognizing that in many circumstances this “Level A” evidence is costly to generate and may not be timely. Often, study participants are narrowly selected in these trials, and the results of such studies are therefore not generalizable.

For specific demographic groups that might be under-represented in trials, large anonymized data sets gleaned from existing electronic medical records or claims data may enable a clinician to address patient-centered questions in a practical manner and to study the effectiveness and risk factors of a treatment. Accordingly, our results do not represent only a single university’s or a few institutions’ experience; rather, our use of a longitudinal database provided a wide-angle lens of a patient population independent of institution-specific constraints.

Yet despite the utility of retrospective and longitudinal databases for studying outcomes in large patient samples, we note that clinical data are aggregated and structured in ways specific to the organization or company administrating the database. Thus, retrospective databases are not monolithic, with higher subscription fees generally relating to higher-quality clinical or patient-level data. For example, the NIS draws from a broad group of hospitals (that is, from approximately 20% of hospitals nationally), but it does not offer longitudinal patient tracking after discharge. In contrast, the MarketScan database has patient-level, longitudinal tracking as well as detailed information on insurance status, while still drawing on a large nationwide hospital sample. The high level of detail at the patient level translates to a higher cost for use of the MarketScan database than that of NIS. The MarketScan database is composed primarily of private payer records; however, the Medicare Provider Utilization and Payment Data has similar patient-level, longitudinal data with similarly high user fees. In addition, the database providers may differ in how procedures are tracked, such as with ICD-9-CM codes in addition to diagnosis-related groups versus with CPT codes directly.

Other databases used retrospectively are being adapted from prospectively collected data, such as the Spine End Results Registry and the American College of Surgeons National Surgical Quality Improvement Program. The advantage of these databases is the granularity of the assessed variables, which often may be more clinically relevant and do not rely on algorithms to extract patient comorbidities from administrative data. However, these types of databases lack insurance and payer information, making it more difficult to conduct cost comparisons and address other socioeconomic questions. The Scoliosis Research Society has yet another paradigm in retrospective database construction: surgeon-reported outcomes. This specific scope has the advantage of focused clinical outcomes most relevant to surgical practitioners, yet also suffers from a lack of socioeconomic data in addition to poor delineations of patient comorbidity measures.

With focused and statistically sound leveraging of database research, randomized controlled trials may be better planned to match surgeons’ most pressing questions and to clarify conflicting results identified in retrospective data assessments. If a given question is adequately addressed with retrospective data, then resources need not necessarily be dedicated to answering it with resource-intensive prospective trials. Furthermore, retrospective data may be used to determine whether the results from trials hold in practice, or to assess complications that were not initially measured and that original studies were not powered to detect.5,20

Results of This Study

Using an administrative, claims-based, longitudinal database, we found an overall complication rate of 13.6% at index hospitalization (that is, complications identified during a patient’s initial admission with a billing claim for surgery) that increased to a cumulative 22.8% at the 30-day follow-up. These rates were comparable to what has been encountered in practice and closely reflected rates observed in prospective studies on spinal surgery complications. Adverse events such as wound dehiscence, infection, and other wound complications showed large relative increases during discrete postoperative time periods, while rates for complications such as new chronic pain, delirium, and dysrhythmia displayed more even and moderate increases over the observed postoperative period.

For the index time point of our study, we truncated our data analysis to the initial hospital stay where a procedure of interest was performed. By limiting our analysis to the index hospitalization only, we eliminated the potential benefits of longitudinal patient-level data and modeled the output of the MarketScan database to parallel assessments with databases such as the NIS. As shown in Table 2, this produced a much more limited assessment of postoperative complications, with many later adverse events being missed because of this focus on only the index time point. This finding indicates that the use of longitudinal data results in potentially superior accuracy in identifying postoperative complications.

In comparing our results to those in studies investigating similar spine surgery outcomes in Table 1, we call attention to the various study types selected and data sources leveraged. Our selection included both prospective and retrospective studies, with the latter drawing on a range of database types including surgeon reported (Scoliosis Research Society Morbidity and Mortality Database), retrospective nonlongitudinal (NIS), retrospective longitudinal (National Surgical Quality Improvement Program), and longitudinal administrative (MarketScan). Overall, the prospective studies indicated higher complication rates than the retrospective studies. This was expected, as investigators actively collect patient information at regular intervals, reducing recall error.

However, the degree of symptom transiency or which recorded complications specifically correlate with patient quality of life is unclear. Despite the lower recall of complications associated with retrospective data, these data generally represent complications that are more representative of what care providers must address clinically. The MarketScan database used in the current study has the advantage of containing longitudinal administrative data across different facilities and payers; thus, any changes in location of service or insurance status do not affect the data collection. With this method of data curation, we find that administrative data can assess complications more reflective of those assessed with prospective data (such as less recall error) yet still represent data that are encountered in clinical practice and likely to represent significant patient events.

Limitations

Despite the aforementioned advantages of using longitudinal administrative data, it is important to consider the limitations of this study. First, we note that the direct comparison of complication rates from our data to the rates obtained from the literature should be viewed with caution given the different methods of data selection and analysis of the various series included. Comparisons of complication rates among surgeries were also limited because indications for the procedures and more detailed contextual information were not included in this analysis. For example, MarketScan does not provide information about which level of the spine was surgically treated or which specific levels have an increased incidence of complications.

In addition, administrative database studies rely on accurate coding of both the index procedures and the complications; coding errors may lead to inappropriate over-or underreporting of events and may vary depending on treatment outcomes. Moreover, the MarketScan database used in this study primarily contained records from individuals with private insurance, which may limit the generalizability of our results. Although we evaluated some of the most common postoperative complications, others are not well assessed with administrative data. Thus, our analysis did not include all potential complications nor did it achieve granularity beyond the ICD-9-CM definitions. As database design and content quality continue to improve, these limitations will become less significant when retrospective data are applied to answering clinical questions. Taken together, longitudinal databases stand to markedly improve our understanding of the postoperative course following spine surgery.

Conclusions

This study demonstrates that postoperative complication rates of spine surgery obtained with data from a large longitudinal administrative database are comparable to rates observed in prospective studies. Longitudinal administrative data are superior to nonlongitudinal administrative data and may contribute to an improved understanding of overall and procedure-specific complication rates after spine surgery.

Author Contributions

Conception and design: all authors. Acquisition of data: Veeravagu, Cole, Azad. Analysis and interpretation of data: Veeravagu, Cole, Azad. Drafting the article: Ratliff, Cole, Azad. Critically revising the article: Ratliff, Veeravagu. Reviewed submitted version of manuscript: Ratliff, Veeravagu. Statistical analysis: Cole. Administrative/technical/material support: Ratliff. Study supervision: Ratliff.

Appendix
SUPPLEMENTAL TABLE 1.

ICD-9 and CPT codes used to define the procedures and complications in the study cohort

Procedure/ComplicationCode
Single-level ACDF (CPT)22554
Multilevel ACDF (CPT)22585, 63076
Ant instrumentation (CPT)22845, 22846, 22847, 22851
Ant decompression (CPT)63075
Ant instrumentation (CPT)22845, 22846, 22847, 22851
Pst cervical procedures (CPT)63001, 63015, 63050, 63051, 63020, 63045, 63040, 63035, 63048, 63043, 22210, 22220, 22216, 22226, 22326, 22590, 22595, 22600, 22614, 63250, 63265, 63270, 63275, 63280, 63285
Cervical corpectomy (CPT)63081, 63082
Ant thoracolumbar decompression (CPT)63077, 63085, 63087, 63090, 63101, 63102, 63301, 63302, 63303, 63305, 63306, 63307, 22857, 22862, 22556, 22558
Pst thoracolumbar decompression (CPT)63030, 63012, 63046, 63047, 63042, 63055, 63056, 63064
Pst thoracolumbar fusion (CPT)22612, 22630
Pst instrumentation (CPT)63290, 22840, 22848, 22849, 22841, 22842, 22843, 22844
Pst lumbar discectomy (CPT)63030, 63042
Nervous system complication, unspecified (ICD-9)997.00
CNS complication (ICD-9)997.01
Nervous system complication, stroke (ICD-9)997.02
Other nervous system complication (ICD-9)997.09
Postop infection (ICD-9)998.5
Wound dehiscence (ICD-9)998.3
Wound hematoma (ICD-9)998.1
Wound complication (ICD-9)998.3, 998.81, 998.83, 998.4
Wound complication (CPT)10140, 10160, 10180, 12020, 12021, 20005, 21501, 22010, 22015
Delirium (ICD-9)293
Chronic pain (ICD-9)338.2, 338.4
Pulmonary complication (ICD-9)997.3, 518.4, 518.5, 518.7, 518.81, 518.82, 518.83, 518.84, 518.89, 519.1
Neurological complication (ICD-9)997.0, 430, 431, 432, 433, 434, 435, 436, 438.2, 438.3, 438.4, 438.5
PE (ICD-9)415.1
DVT (ICD-9)453.4, 451, 453.8, 453.2, 453.1, 453.9
Dysrhythmia (ICD-9)427, 426.1, 426, 426.3, 426.4, 426.5, 426.6, 426.7, 426.8
Dysphagia (ICD-9)787.2

References

  • 1

    Bazaz RLee MJYoo JU: Incidence of dysphagia after anterior cervical spine surgery: a prospective study. Spine (Phila Pa 1976) 27:245324582002

    • Search Google Scholar
    • Export Citation
  • 2

    Campbell PGYadla SMalone JZussman BMaltenfort MGSharan AD: Early complications related to approach in cervical spine surgery: single-center prospective study. World Neurosurg 74:3633682010

    • Search Google Scholar
    • Export Citation
  • 3

    Chen CJSaulle DFu KMSmith JSShaffrey CI: Dysphagia following combined anterior-posterior cervical spine surgeries. J Neurosurg Spine 19:2792872013

    • Search Google Scholar
    • Export Citation
  • 4

    Chitale RCampbell PGYadla SWhitmore RGMaltenfort MGRatliff JK: International Classification of Disease Clinical Modification 9 modeling of a patient comorbidity score predicts incidence of perioperative complications in a nationwide inpatient sample assessment of complications in spine surgery. J Spinal Disord Tech 2012. [epub ahead of print]

    • Search Google Scholar
    • Export Citation
  • 5

    Cole TVeeravagu AJiang BRatliff JK: Usage of recombinant human bone morphogenetic protein in cervical spine procedures: analysis of the MarketScan longitudinal database. J Bone Joint Surg Am 96:140914162014

    • Search Google Scholar
    • Export Citation
  • 6

    Deyo RAHickam DDuckart JPPiedra M: Complications following surgery for lumbar stenosis in a veteran population. Spine (Phila Pa 1976) 38:169517022013

    • Search Google Scholar
    • Export Citation
  • 7

    Deyo RAMirza SKMartin BIKreuter WGoodman DCJarvik JG: Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in older adults. JAMA 303:125912652010

    • Search Google Scholar
    • Export Citation
  • 8

    Fineberg SJAhmadinia KOglesby MPatel AASingh K: Hospital outcomes and complications of anterior and posterior cervical fusion with bone morphogenetic protein. Spine (Phila Pa 1976) 38:130413092013

    • Search Google Scholar
    • Export Citation
  • 9

    Fu KMGSmith JSPolly DW JrPerra JHSansur CABerven SH: Morbidity and mortality in the surgical treatment of six hundred five pediatric patients with isthmic or dysplastic spondylolisthesis. Spine (Phila Pa 1976) 36:3083122011

    • Search Google Scholar
    • Export Citation
  • 10

    Lee MJKonodi MACizik AMBransford RJBellabarba CChapman JR: Risk factors for medical complication after spine surgery: a multivariate analysis of 1,591 patients. Spine J 12:1972062012

    • Search Google Scholar
    • Export Citation
  • 11

    Lovecchio FHsu WKSmith TRCybulski GKim BKim JYS: Predictors of thirty-day readmission after anterior cervical fusion. Spine (Phila Pa 1976) 39:1271332014

    • Search Google Scholar
    • Export Citation
  • 12

    Martin BIDeyo RAMirza SKTurner JAComstock BAHollingworth W: Expenditures and health status among adults with back and neck problems. JAMA 299:6566642008

    • Search Google Scholar
    • Export Citation
  • 13

    Nandyala SVMarquez-Lara AFineberg SJSingh RSingh K: Incidence and risk factors for perioperative visual loss after spinal fusion. Spine J 14:186618722014

    • Search Google Scholar
    • Export Citation
  • 14

    Nasser RYadla SMaltenfort MGHarrop JSAnderson DGVaccaro AR: Complications in spine surgery. J Neurosurg Spine 13:1441572010

    • Search Google Scholar
    • Export Citation
  • 15

    Patil CGSarmiento JMUgiliweneza BMukherjee DNuño MLiu JC: Interspinous device versus laminectomy for lumbar spinal stenosis: a comparative effectiveness study. Spine J 14:148414922014

    • Search Google Scholar
    • Export Citation
  • 16

    Pugely AJMartin CTGao YMendoza-Lattes SA: Outpatient surgery reduces short-term complications in lumbar discectomy: an analysis of 4310 patients from the ACS-NSQ-IP database. Spine (Phila Pa 1976) 38:2642712013

    • Search Google Scholar
    • Export Citation
  • 17

    Smith JSFu KMGPolly DW JrSansur CABerven SHBroadstone PA: 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. Spine (Phila Pa 1976) 35:214021492010

    • Search Google Scholar
    • Export Citation
  • 18

    Smith JSSansur CADonaldson WF IIIPerra JHMudiyam RChoma TJ: Short-term morbidity and mortality associated with correction of thoracolumbar fixed sagittal plane deformity: a report from the Scoliosis Research Society Morbidity and Mortality Committee. Spine (Phila Pa 1976) 36:9589642011

    • Search Google Scholar
    • Export Citation
  • 19

    Veeravagu ACole TJiang BRatliff JK: Revision rates and complication incidence in single- and multilevel anterior cervical discectomy and fusion procedures: an administrative database study. Spine J 14:112511312014

    • Search Google Scholar
    • Export Citation
  • 20

    Veeravagu ACole TSJiang BRatliff JKGidwani RA: The use of bone morphogenetic protein in thoracolumbar spine procedures: analysis of the MarketScan longitudinal database. Spine J 14:292929372014

    • Search Google Scholar
    • Export Citation
  • 21

    Yadla SMalone JCampbell PGMaltenfort MGHarrop JSSharan AD: Early complications in spine surgery and relation to preoperative diagnosis: a single-center prospective study. J Neurosurg Spine 13:3603662010

    • Search Google Scholar
    • Export Citation
  • 22

    Yadla SMalone JCampbell PGMaltenfort MGSharan ADHarrop JS: Preoperative diagnosis and early complications in thoracolumbar spine surgery: a single center prospective study. J Spinal Disord Tech 24:E16E202011

    • Search Google Scholar
    • Export Citation
  • 23

    Yadla SMalone JCampbell PGNasser RMaltenfort MGHarrop JS: Incidence of early complications in cervical spine surgery and relation to preoperative diagnosis: a single-center prospective study. J Spinal Disord Tech 24:50542011

    • Search Google Scholar
    • Export Citation

If the inline PDF is not rendering correctly, you can download the PDF file here.

Article Information

Correspondence Correspondence John Ratliff, Department of Neurosurgery, Stanford University Medical Center, 300 Pasteur Dr., R291 MC 5327, Stanford, CA 94305–5327. email: jratliff@stanford.edu.

Dr. Veeravagu and Mr. Cole contributed equally to this work.

INCLUDE WHEN CITING Published online June 12, 2015; DOI: 10.3171/2014.12.SPINE14659.

DISCLOSURE This work was supported by a research grant from the Orthopedic Research and Education Foundation (Grant OREF 112–209), the Walsh Foundation, and the Stanford University Medical Scholars Program. CPT codes are a registered trademark of the American Medical Association.

© AANS, except where prohibited by US copyright law.

Headings

References

  • 1

    Bazaz RLee MJYoo JU: Incidence of dysphagia after anterior cervical spine surgery: a prospective study. Spine (Phila Pa 1976) 27:245324582002

    • Search Google Scholar
    • Export Citation
  • 2

    Campbell PGYadla SMalone JZussman BMaltenfort MGSharan AD: Early complications related to approach in cervical spine surgery: single-center prospective study. World Neurosurg 74:3633682010

    • Search Google Scholar
    • Export Citation
  • 3

    Chen CJSaulle DFu KMSmith JSShaffrey CI: Dysphagia following combined anterior-posterior cervical spine surgeries. J Neurosurg Spine 19:2792872013

    • Search Google Scholar
    • Export Citation
  • 4

    Chitale RCampbell PGYadla SWhitmore RGMaltenfort MGRatliff JK: International Classification of Disease Clinical Modification 9 modeling of a patient comorbidity score predicts incidence of perioperative complications in a nationwide inpatient sample assessment of complications in spine surgery. J Spinal Disord Tech 2012. [epub ahead of print]

    • Search Google Scholar
    • Export Citation
  • 5

    Cole TVeeravagu AJiang BRatliff JK: Usage of recombinant human bone morphogenetic protein in cervical spine procedures: analysis of the MarketScan longitudinal database. J Bone Joint Surg Am 96:140914162014

    • Search Google Scholar
    • Export Citation
  • 6

    Deyo RAHickam DDuckart JPPiedra M: Complications following surgery for lumbar stenosis in a veteran population. Spine (Phila Pa 1976) 38:169517022013

    • Search Google Scholar
    • Export Citation
  • 7

    Deyo RAMirza SKMartin BIKreuter WGoodman DCJarvik JG: Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in older adults. JAMA 303:125912652010

    • Search Google Scholar
    • Export Citation
  • 8

    Fineberg SJAhmadinia KOglesby MPatel AASingh K: Hospital outcomes and complications of anterior and posterior cervical fusion with bone morphogenetic protein. Spine (Phila Pa 1976) 38:130413092013

    • Search Google Scholar
    • Export Citation
  • 9

    Fu KMGSmith JSPolly DW JrPerra JHSansur CABerven SH: Morbidity and mortality in the surgical treatment of six hundred five pediatric patients with isthmic or dysplastic spondylolisthesis. Spine (Phila Pa 1976) 36:3083122011

    • Search Google Scholar
    • Export Citation
  • 10

    Lee MJKonodi MACizik AMBransford RJBellabarba CChapman JR: Risk factors for medical complication after spine surgery: a multivariate analysis of 1,591 patients. Spine J 12:1972062012

    • Search Google Scholar
    • Export Citation
  • 11

    Lovecchio FHsu WKSmith TRCybulski GKim BKim JYS: Predictors of thirty-day readmission after anterior cervical fusion. Spine (Phila Pa 1976) 39:1271332014

    • Search Google Scholar
    • Export Citation
  • 12

    Martin BIDeyo RAMirza SKTurner JAComstock BAHollingworth W: Expenditures and health status among adults with back and neck problems. JAMA 299:6566642008

    • Search Google Scholar
    • Export Citation
  • 13

    Nandyala SVMarquez-Lara AFineberg SJSingh RSingh K: Incidence and risk factors for perioperative visual loss after spinal fusion. Spine J 14:186618722014

    • Search Google Scholar
    • Export Citation
  • 14

    Nasser RYadla SMaltenfort MGHarrop JSAnderson DGVaccaro AR: Complications in spine surgery. J Neurosurg Spine 13:1441572010

    • Search Google Scholar
    • Export Citation
  • 15

    Patil CGSarmiento JMUgiliweneza BMukherjee DNuño MLiu JC: Interspinous device versus laminectomy for lumbar spinal stenosis: a comparative effectiveness study. Spine J 14:148414922014

    • Search Google Scholar
    • Export Citation
  • 16

    Pugely AJMartin CTGao YMendoza-Lattes SA: Outpatient surgery reduces short-term complications in lumbar discectomy: an analysis of 4310 patients from the ACS-NSQ-IP database. Spine (Phila Pa 1976) 38:2642712013

    • Search Google Scholar
    • Export Citation
  • 17

    Smith JSFu KMGPolly DW JrSansur CABerven SHBroadstone PA: 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. Spine (Phila Pa 1976) 35:214021492010

    • Search Google Scholar
    • Export Citation
  • 18

    Smith JSSansur CADonaldson WF IIIPerra JHMudiyam RChoma TJ: Short-term morbidity and mortality associated with correction of thoracolumbar fixed sagittal plane deformity: a report from the Scoliosis Research Society Morbidity and Mortality Committee. Spine (Phila Pa 1976) 36:9589642011

    • Search Google Scholar
    • Export Citation
  • 19

    Veeravagu ACole TJiang BRatliff JK: Revision rates and complication incidence in single- and multilevel anterior cervical discectomy and fusion procedures: an administrative database study. Spine J 14:112511312014

    • Search Google Scholar
    • Export Citation
  • 20

    Veeravagu ACole TSJiang BRatliff JKGidwani RA: The use of bone morphogenetic protein in thoracolumbar spine procedures: analysis of the MarketScan longitudinal database. Spine J 14:292929372014

    • Search Google Scholar
    • Export Citation
  • 21

    Yadla SMalone JCampbell PGMaltenfort MGHarrop JSSharan AD: Early complications in spine surgery and relation to preoperative diagnosis: a single-center prospective study. J Neurosurg Spine 13:3603662010

    • Search Google Scholar
    • Export Citation
  • 22

    Yadla SMalone JCampbell PGMaltenfort MGSharan ADHarrop JS: Preoperative diagnosis and early complications in thoracolumbar spine surgery: a single center prospective study. J Spinal Disord Tech 24:E16E202011

    • Search Google Scholar
    • Export Citation
  • 23

    Yadla SMalone JCampbell PGNasser RMaltenfort MGHarrop JS: Incidence of early complications in cervical spine surgery and relation to preoperative diagnosis: a single-center prospective study. J Spinal Disord Tech 24:50542011

    • Search Google Scholar
    • Export Citation

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 277 264 51
PDF Downloads 207 187 8
EPUB Downloads 0 0 0

PubMed

Google Scholar