Disparities based on insurance status in the American health care system are well established. However, to the authors' knowledge, this is the first study to evaluate variables that may explain differences based on payer type in the outcomes after surgery for spinal metastases.
Data from the Nationwide Inpatient Sample (2005-2008) were retrospectively extracted. Patients ages 18 to 64 years who underwent surgery for spinal metastases were included. Multivariate logistic regression was performed to calculate the adjusted odds of in-hospital death and the development of a complication for Medicaid recipients and for those without insurance compared with privately insured patients. All analyses were adjusted for differences in patient age, gender, primary tumor histology, socioeconomic status, hospital bed size, and hospital teaching status.
A total of 2157 hospital admissions were evaluated. The adjusted odds of in-hospital death were significantly higher for Medicaid recipients (crude rate: 6.5%; odds ratio [OR], 1.79; 95% confidence interval [95% CI], 1.11-2.88 [P = .02]) and uninsured patients (crude rate: 7.7%; OR, 2.15; 95% CI, 1.04-4.46 [P = .04]) compared with privately insured patients (crude rate: 3.8%). Complication rates were also significantly higher for Medicaid recipients (OR, 1.34; 95% CI, 1.04-1.72 [P = .02]). However, after also adjusting for acuity of presentation, the odds of in-hospital death were not significantly different for Medicaid (OR, 1.38; 95% CI, 0.86-2.21 [P = .18]) or uninsured patients (OR, 1.86; 95% CI, 0.90-3.83 [P = .09]); in addition, complication rates did not appear to differ significantly.
The impact of insurance status on the health of Americans is a question of the utmost importance to national policy makers. In 2008, there were 46.3 million uninsured Americans and an additional 87.4 million Americans had government-sponsored insurance.1 The Patient Protection and Affordable Care Act (PPACA) was passed in 2010 to decrease the number of Americans without health insurance.2 However, although PPACA may decrease the number of uninsured Americans, the number of patients who are underinsured (defined as having insurance coverage that lacks benefits, limits access to care, or has prohibitive out-of-pocket costs) may continue to increase.2-4 In addition, the number of Americans with government-sponsored insurance will increase, but some have expressed concern about the quality of care provided to Medicaid recipients, partially because of low physician reimbursmenet.5
A growing body of literature suggests that Americans with suboptimal insurance are at risk for poor medical and surgical outcomes.1, 5-31 Disparities based on insurance status are well documented in oncology patients: restricted access to cancer screening,11 poor postoperative outcomes after tumor resection,19 limited ability to enroll in clinical trials,8 and inferior survival12, 13, 24 for patients with unfavorable insurance have been described for those with many different tumor types. However, to the best of our knowledge, investigation into outcomes based on payer type specifically for patients with metastatic disease has been limited. Because patients with metastases typically have relatively poor general health and have had substantial contact with the American health care system throughout the course of their illness, and because of the extensive spinal fusion operation that is often performed, those undergoing surgery for spinal metastases may be particularly affected by insurance status. However, to our knowledge, the impact of payer type on the postoperative outcomes after surgery for spinal metastases has not been previously evaluated.
The relation between insurance status and surgical outcomes is complex,1 but 3 variables have been described that may partially explain the disparities based on insurance for surgical patients19: 1) restricted access to high-quality care; 2) higher acuity of presentation; and 3) poorer preoperative health. The goal of this study is to describe any variability based on payer type that may exist in the outcomes after surgery for spinal metastases, and to determine the degree to which such disparities may be attributable to differences in comorbidities, acuity of presentation, and the performance of surgery at a high-volume hospital. To that end, we report what to our knowledge is the first study comparing in-hospital mortality, postoperative complications, length of hospital stay, total hospital charges, and discharge disposition for Medicaid recipients and for uninsured patients undergoing surgery for spinal metastases compared with those with private insurance.
MATERIALS AND METHODS
Data were retrospectively extracted from the Nationwide Inpatient Sample (NIS; Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project) from 2005 through 2008. As the largest all-payer national inpatient database, the NIS contains approximately 8 million discharges from more than 1000 hospitals. The NIS is a 20% stratified sample of all nonfederal hospitals in the United States, in which stratification is based on hospital characteristics (geographic region, urban or rural location, teaching status, ownership, and size). The NIS has been used extensively to evaluate the outcomes of patients undergoing spine surgery.32
Patients were included if they had spinal metastases and underwent surgical decompression with or without fusion. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes of 198.3, 198.4, and 198.5 were used to identify patients with a secondary malignant spine tumor. The ICD-9-CM procedure codes were used to extract patients who had undergone spinal decompression (03.01, 03.09, 03.4, and 03.53) or fusion (81.00-81.08, 81.61). To decrease the likelihood of misclassification, patients were only included if spine surgery was one of the first 3 coded procedures.
The expected primary payer is encoded in the NIS as 1) Medicare, 2) Medicaid, 3) private insurance (including health maintenance organizations), 4) self-pay, 5) no charge, and 6) other (including Workers' Compensation, Civilian Health and Medical Program of the Uniformed Services [CHAMPUS], Civilian Health and Medical Program of the Department of Veterans Affairs [CHAMPVA], Title V, and other government programs). To restrict the study to adults at the greatest risk for lack of insurance coverage, only patients ages 18 to 64 years were evaluated.10, 11, 19 Because they are primarily patients with baseline disability or end-stage renal disease, those with an expected primary payer of Medicare or other were excluded.19 For consistency with prior reports,19 payer status was categorized as 1) private, 2) Medicaid, or 3) uninsured (self-pay or no charge).
Patient and Hospital Characteristics
Predictor variables extracted were patient age, gender, socioeconomic status, comorbid disease, primary tumor histology, visceral metastases, myelopathy, hospital volume, admission type, hospital bed size, and hospital teaching status. Age was evaluated in 3 categories: 1) 18 to 35 years, 2) 36 to 50 years, and 3) 51 to 64 years. Although data regarding the individual patient's socioeconomic status is not included in the NIS, the median household income for the patient's ZIP code is reported as 1) < $39,000, 2) $39,000 to $47,999, 3) $48,000 to 62,999, and 4) > $63,000. The independent effect of the comorbidities defined by Elixhauser et al. were evaluated33; however, paralysis, other neurological deficits, metastatic disease, lymphoma, and primary tumor were not examined because of their potential association with spinal metastases. Primary tumor histology was categorized as 1) lung (162, V10.11); 2) breast (174, V10.3); 3) prostate (185, V10.46); 4) renal cell cancer (189.0, V10.52); and 5) other, unknown, or unspecified. Visceral (nonspinal) metastases were determined using the ICD-9-CM series 197 and 198 (198.0, 198.1, 198.2, 198.6, 198.7, and 198.8), and myelopathy was extracted using 336.3, 336.8, and 336.9. The NIS encodes admission type as 1) elective or 2) nonelective.
Because 100% of discharges from sampled hospitals are included in the database, the NIS can be used to determine the hospital volume of a procedure using the unique identification number for that hospital. The number of operations performed at that hospital for spinal metastases over the period for all patients (regardless of age or payer status) was assessed; for consistency with prior reports,34 hospitals were divided based on volume into quintiles, such that each quintile had approximately 20% of the entire patient population. The lowest volume quintile was defined as very low volume, and subsequent quintiles were defined as low, medium, high, and very high volume.
The outcomes evaluated were in-hospital mortality, the development of at least 1 complication, length of hospital stay, total hospital charges, and discharge disposition. Complications were extracted using ICD-9-CM codes for postoperative neurological complications (997.00-997.09); pulmonary complications (518.5, 518.81, 518.84, and 997.3); venous thromboembolic events (415.11-415.19, 453.40-2, 453.8, and 453.9); cardiac complications (997.1 and 410); urinary and renal complications (584.5, 584.9, and 997.5); gastrointestinal complications (008.45, 560.1, and 997.4); infectious complications, including a wound infection (998.32, 998.51, 998.59, 998.6, 998.81, and 998.83), urinary tract infection (595.0, 595.9, and 599.0), meningitis (320), sepsis (038), or pneumonia (481, 482, and 486); and decubitus ulcers (707.01-09).
Length of hospital stay and total hospital charges, which exclude professional fees, are encoded directly in the NIS. Discharge disposition is encoded as 1) routine (to home), 2) transfer to a short-term hospital, 3) other transfer (including to a skilled nursing facility or to intermediate care), 4) home health care, 5) against medical advice, 6) died, and 7) unknown. A nonroutine discharge was defined any disposition other than the first.
Descriptive statistics were conducted for demographic variables, which were compared using the chi-square test. Multivariate logistic regression models were constructed for categorical variables and multivariate linear regression was performed to analyze length of hospital stay and total hospital charges. Because of the positive skew of length of stay and charges, logarithmic transformation was performed. Statistical analyses were performed using STATA 11 software (StataCorp, College Station, Tex). All analyses were performed accounting for the complex survey sampling technique of the NIS (using survey commands), which incorporates hospital clustering and the sampling weight of each discharge. A P value < .05 was accepted as statistically significant.
Demographics of the Study Population
A total of 2157 admissions were analyzed. The expected primary payer was private insurance in 77.4% (n = 1670), Medicaid in 17.2% (n = 370), and self-pay or no charge in 5.4% (n = 117) of admissions. The demographics of the study population are compared by payer type in Table 1. Patient age, number of comorbidities, the percentage of patients with visceral metastases, and hospital teaching status did not differ significantly by insurance status. A higher percentage of Medicaid recipients and uninsured patients had a nonelective admission or myelopathy at presentation; hospital volume also differed significantly by payer status, because a lower percentage of uninsured patients were admitted to a high-volume or very high-volume hospital (Figure 1).
Table 1. Comparison of the Characteristics of the Patients Undergoing Surgery for Spinal Metastases in the United States by Insurance Status
Approximately 4.5% of patients died during the hospitalization and 31.1% developed at least 1 complication. After adjusting for patient age, gender, primary tumor histology, socioeconomic status, hospital teaching status, and hospital bed size, both Medicaid and uninsured patients had a significantly higher adjusted odds of in-hospital death compared with patients with private insurance; likewise, Medicaid recipients had a higher adjusted odds of developing a postoperative complication (Table 2). However, after also adjusting for admission type (elective vs nonelective), visceral metastases, and myelopathy, neither in-hospital mortality nor postoperative complications were found to differ significantly by insurance status (Table 3).
Table 2. Evaluation of How Acuity of Presentation, Comorbidities, and Hospital Volume Impact Differences by Insurance Status in the Adjusted Odds of In-Hospital Mortality and Developing a Complication After Surgery for Spinal Metastases
Private (n = 1670)
Medicaid (n = 370)
Uninsured (n = 117)
95% CI indicates 95% confidence interval; OR, odds ratio.
The initial model is adjusted for differences in patient age, gender, primary tumor histology, medium household income of the patient's ZIP code, hospital teaching status, and hospital bed size.
Statistically significant difference.
Acuity of presentation includes admission type (elective vs nonelective), visceral metastases, and myelopathy.
Initial model with acuity of presentationc and comorbidities
Adjusted OR (95% CI)
1.17 (0.90-1.54) P = .24
1.05 (0.66-1.65) P = .85
Initial model with acuity of presentation,c comorbidities, and hospital volume
Adjusted OR (95% CI)
1.18 (0.90-1.54) P = .23
1.05 (0.66-1.65) P = 85
Table 3. Outcomes After Surgery for Spinal Metastases by Payer Statusa
Nonroutine Hospital Discharge
95% CI indicates 95% confidence interval; OR, odds ratio.
All analyses are adjusted for differences in patient age, gender, comorbid disease, the percentage of patients with myelopathy, the percentage of patients with visceral metastases, primary tumor histology, medium household income of the patient's ZIP code, hospital teaching status, and hospital bed size.
Length of Hospital Stay and Total Hospital Charges
The median length of hospital stay was 9 days (interquartile range [IQR], 6-15 days); the length of hospital stay was significantly longer for Medicaid and uninsured patients compared with those with private insurance (Table 4). The median total hospital charge was $97,968 (IQR, $56,540-$166,344), and hospital charges were significantly higher for Medicaid enrollees, but were not different for uninsured patients, compared with those with private insurance (Table 4). However, after also adjusting for length of stay, total charges were found to be significantly lower for uninsured patients (-6.11%; 95% confidence interval [95% CI], −11.54 to −0.67 [P = .03]), but not significantly different for Medicaid recipients (-2.04%; 95% CI, −5.60 to 1.51 [P = .26]), compared with those with private insurance.
Table 4. Efficiency of Care After Surgery for Spinal Metastases by Insurance Statusa
Length of Hospital Stay, Days
Total Hospital Charges, US$
95% CI indicates 95% confidence interval; IQR, interquartile range.
All analyses are adjusted for differences in patient age, gender, comorbid disease, primary tumor histology, the percentage of patients with myelopathy, the percentage of patients with visceral metastases, medium household income of the patient's ZIP code, hospital volume of surgery for spinal metastases, hospital teaching status, and hospital bed size.
Approximately 57.7% of patients in the study population had a nonroutine hospital discharge. The adjusted odds of a nonroutine hospital discharge were not significantly different for Medicaid recipients, but were significantly lower for uninsured patients compared with those with private insurance (Table 3).
Spinal metastases are an important cause of morbidity and mortality, and approximately 5% to 10% of patients with a malignancy will develop metastatic epidural spinal cord compression.35-43 Radiotherapy, stereotactic radiosurgery, percutaneous spinal procedures (vertebroplasty or kyphoplasty), or surgical decompression with or without fusion may be the optimal treatment depending on the primary tumor type, prognosis, and spinal stability.39, 44-54 Indications for surgery include rapid neurological deterioration; neural compression because of spinal deformity, osseous, or tumor pathology; tumor histology that is resistant to radiotherapy; and disease recurrence after maximal radiotherapy.40, 44, 55-57 The decision to pursue surgery rather than another treatment modality should not vary by insurance status.
In the current study, 2157 patients from across the United States were evaluated to investigate the impact of payer status on outcomes after surgery for patients with spinal metastases. Patient gender and socioeconomic status differed significantly by insurance status, but these variations are consistent with the known demographics of patients with unfavorable payer types in the United States. Medicaid and uninsured patients were significantly more likely to have a nonelective hospital admission and present with myelopathy. Hospital volume differed significantly by payer type: a lower percentage of uninsured patients were treated at high-volume and very high-volume hospitals. After adjusting for demographic and hospital variables, both Medicaid enrollees and uninsured patients had significantly higher adjusted odds of in-hospital death. Likewise, Medicaid patients had significantly higher adjusted odds of developing a postoperative complication. However, after also adjusting for acuity of presentation, in-hospital morality and complications did not differ significantly by insurance status. Although a nonroutine hospital discharge was significantly less likely for uninsured patients, this may reflect restricted access to acute rehabilitation facilities.
Acuity of Presentation
In this study, 3 variables were used to indicate acuity of presentation: 1) hospital admission type (elective vs nonelective); 2) the percentage of patients who also presented with visceral (nonspinal) metastases; and 3) the percentage of patients who presented with myelopathy (which was used to indicate a greater severity of metastatic spine disease because there is no specific ICD-9-CM code for spinal cord compression). Hospital admission type differed significantly by payer status, with a significantly higher percentage of privately insured patients having an elective admission. Likewise, the percentage of patients who presented with myelopathy differed significantly by payer status, with the lowest percentage noted among those with private insurance; moreover, a lower percentage of patients with private insurance presented with visceral metastases. The higher adjusted odds of in-hospital death for Medicaid and uninsured patients after surgery for spinal metastases did not persist after adjusting for these variables. This suggests that acuity of presentation, rather than differential treatment by individual providers, may be primarily responsible for differences by payer type observed in the outcomes after surgery for patients with spinal metastases.
The findings of the current study are consistent with those of several studies that have shown that patients with suboptimal insurance are less likely to present for care, and when they do ultimately present, the acuity is greater.15, 20 Uninsured women in the United States have been shown to present with higher stage breast cancer.9 Moreover, uninsured patients are more likely to present with perforated appendicitis15 and a ruptured abdominal aortic aneurysm.20 Those with unfavorable insurance may be afraid to present for care because of limited financial resources. Geographic (including transportation) barriers to care may be another important factor. Patients with suboptimal insurance, especially those without insurance, may be less likely to have a regular primary care physician or access to specialty consultation.11 In turn, patients may be less likely to receive adequate screening for cancer and those patients with a history of malignancy may be less likely to undergo appropriate follow-up and evaluation for recurrence. Those with unfavorable insurance may have to receive their care in the emergency department, where treatment of subacute or chronic conditions may be suboptimal.
Access to Care
Americans without health insurance face substantial barriers to care, and studies have shown that uninsured patients are less likely to undergo surgery by high-volume providers.21, 58 Although provider volume is an imperfect proxy for health care quality, many studies have found that increased provider volume is associated with superior surgical outcomes.59-61 In this study, a lower percentage of uninsured patients were treated at high-volume or very high-volume hospitals. However, the addition of hospital volume had little effect on multivariate regression models evaluating in-hospital mortality and postoperative complications. This suggests that other variables may be primarily responsible for the differences in outcomes by insurance status noted after surgery for spinal metastases.
Unfavorable payer type has been shown to limit access to primary care services.11 As a result, patients with suboptimal insurance may be less likely to receive health maintenance services, less likely to receive follow-up and referral to specialists for chronic conditions, and to have difficulty paying for medications.4, 11 However, in the current study, there was no significant difference noted in the number of comorbidities by insurance type. Moreover, the addition of comorbidities (using the classification system of Elixhauser et al.33) did not decrease the adjusted odds of in-hospital mortality or postoperative complications for patients with unfavorable insurance. Thus, other factors may be primarily responsible for the differences in outcomes by payer type for patients undergoing surgery for spinal metastases.
Advantages and Limitations to the Study Design
The NIS has several unique features that make it well suited to investigate the impact of insurance status in the United States, and this database has been used to evaluate disparities based on payer type for patients with other conditions.6, 14, 17-19 The NIS was used in the current study because it is the largest all-payer national database. Thus, the NIS provides a broader perspective than regional or statewide databases, and, in contrast to single-payer databases, allows the impact of many different payer types, including self-pay and no charge, to be examined.
However, there are also limitations to the current study. The NIS uses the same category for all patients who have private insurance; however, some private insurance plans may have had inadequate coverage of certain services or excessive copays, and therefore these patients may actually have been underinsured. However, this classification likely only underestimates disparities based on insurance status by potentially including underinsured patients with the private insurance group.19 In addition, there may be variations in Medicaid eligibility and coverage by state that could not be accounted for. The NIS has limited information on patients at the time of presentation and thus it was not possible to adjust for the specific chief symptom, neurological examination, speed of neurologic deterioration, imaging findings (including spinal location or stability), or overall prognosis. The database only includes in-hospital data, and long-term neurological function and mortality could not be assessed. There may also be miscoded data in the NIS, but there is no reason to suspect that errors would preferentially affect patients with specific types of insurance.
In this nationwide study based on an administrative database, disparities based on insurance status in the outcomes after surgery for patients with spinal metastases were found to be largely attributable to a higher acuity of presentation for care among Medicaid recipients and uninsured patients. This suggests that decreasing financial barriers and improving access to care (including cancer screening, primary care, and specialists) for those with suboptimal insurance may be important. However, future research, particularly with data collected in a prospective fashion, will be needed to evaluate whether PPACA decreases barriers to care or improves outcomes after cancer surgery, including surgery for spinal metastases.
Dr. Dasenbrock is supported by the Doris Duke Clinical Research Fellowship for Medical Students at Johns Hopkins University.
CONFLICT OF INTEREST DISCLOSURES
Dr. Gokaslan has stock ownership in US Spine and Spinal Kinetics and has received research support from AO Spine North America, Integra LifeSciences, Johnson & Johnson, and DePuy Spine. He also has received fellowship support and a stipend as a board member and has received honorarium from AO Spine North America. Drs. Bydon and Witham are the recipients of research grants from Johnson & Johnson and DePuy Spine.