This study represents the first analysis using the National Inpatient Sample (NIS) database to examine the association between race and adverse outcomes in patients undergoing meningioma surgery. The results support the assertion that race is an important factor associated with increased incidence of complications in patients with intracranial meningioma who undergo surgical resection.
Meningiomas are slow-growing tumors of the meninges, often arising from intracranial or spinal dural surfaces.1 They are the most common primary brain tumor, accounting for 37.1% of all central nervous system tumors in adults, with an incidence of 8.33 per 100,000 person-years.2 The World Health Organization categorizes meningiomas into three separate grades: Grade I: benign meningiomas (80.6%), Grade II: atypical meningiomas (17.6%) and Grade III: malignant meningiomas (2.1%).2 Presenting symptoms are associated with tumor location and mass effect on adjacent nervous system structures. Patients with skull base meningiomas most commonly present with headaches, ocular deficits, cerebellar deficits and temporal deficits.3 Asymptomatic patients are often managed via observation and routine surveillance, while those with actively growing, symptomatic tumors typically undergo surgery.4
“Studies have found racial disparities in head and neck cancer patients.“
Studies have found racial disparities in head and neck cancer patients.5 Black patients have been shown to present with head and neck cancers at later stages, with more frequent distant disease, and to have lower five-year survival rates compared to white patients.5 Although research of racial disparities in patients with meningiomas is still sparse, studies using Surveillance, Epidemiology, and End Results (SEER) and the National Cancer Database (NCDB) found significantly increased odds of death in black patients with atypical meningiomas and malignant meningiomas compared to their white counterparts.6-8
While previous studies have analyzed racial disparities in survival following meningioma surgery using the NCDB and SEER, none, to our knowledge, have used the National Inpatient Sample (NIS) to conduct analyses of racial differences on outcomes such as complications, length of stay (LOS) and total charges.
This retrospective analysis used discharge data from the NIS, Healthcare Cost and Utilization Project (HCUP) and Agency for Healthcare Research and Quality from 2012 to 2014.9 The NIS represents 20% of all discharges from non-federal community hospitals, not including long-term acute care hospitals and rehabilitation centers. Through a stratified sampling design, the NIS accounts for an estimated 35 million hospitalizations nationally each year. The NIS represents 97% of the United States population. It contains unweighted data from more than 7 million discharges per year as it is the largest all-payer inpatient database in the United States.9 The NIS database is a powerful database that can be used to reach statistical power and significance even with relatively uncommon disorders. Therefore, it has been similarly used in numerous other retrospective studies spanning multiple fields of research.10-14 All information compiled used de-identified data. This study was categorized as exempt by the Rutgers New Jersey Medical School Institutional Review Board, Newark, New Jersey, based on their standing policy.
The NIS database was queried for patients undergoing meningioma surgery between the years 2012 to 2014 using ICD-9 codes. Patients were included if they had a meningioma diagnosis and had an excision procedure of the meninges or other brain tissue. This was performed similarly to a previous NIS database study analyzing meningioma surgery.15 Meningioma diagnoses included the ICD-9 diagnostic codes for malignant (192.1), benign (225.1) or uncertain behavior (237.6) neoplasm of the cerebral meninges, and surgical procedure involved the procedural codes for tissue excision of the cerebral meninges (01.51) or other excision of brain tissue (01.59). Meningiomas arising from the spinal canal were excluded from this analysis. Those under the age of 18 years old, missing demographic data or missing hospital characteristic data were also excluded.
The outcome measures in this study included in-hospital mortality, postoperative complications (cardiac, pulmonary, neurological, urinary/renal, infection, operative and all complications), prolonged LOS and high total charges. Prolonged LOS was defined as those patients with a LOS greater than the 90th percentile for the sample, and high total charges were defined with the same methodology. Postoperative complications were indicated using the ICD-9 diagnostic codes derived from a study by Shen and colleagues16 (see Table 1).
The primary exposure variable was race/ethnicity (white, black, Hispanic, Asian/Pacific Islander, Native American and other), which was analyzed for its effects on the aforementioned outcomes. Although the “Native American” and “other” categories were retained in the race variable to properly adjust for race in this study, their statistical outcomes were not shown due to small sample size and lack of granularity, respectively. To adjust for differences in patient and hospital characteristics, several confounding covariables were included in the analysis: age, gender (male, female), payer (private, Medicare, Medicaid, uninsured, other), median income quartile by zip code (1st – 25th, 26th – 50th, 51st – 75th, 76th – 100th), hospital bed size (small, medium, large), hospital teaching status (urban teaching, urban non-teaching, rural) and hospital region (Northeast, Midwest, South, West). The Elixhauser Comorbidity Index score was also included in the regression to adjust for any difference in comorbidities between the groups.17 Cases with missing data for any of the aforementioned covariates were removed from the analysis.
Patient demographics, hospital characteristics, Elixhauser Comorbidity Index, and complication rates were compared among the race cohorts using Fisher’s exact test and one-way analysis of variance when appropriate, using weighted proportions. Logistic regression analysis was used to assess the independent effect of race on the outcomes of interest in meningioma surgery patients, while adjusting for patient demographics, hospital characteristics and Elixhauser comorbidity score. Although Cox Proportional Hazards Model could have potentially been used as a powerful tool for investigating mortality, the NIS database does not provide “time to death” data, precluding this type of analysis. The alpha level of this study was set at p<0.05. All analyses were done using SPSS Statistics Version 25.0 (IBM Corporation, Armonk, NY). Survey procedures (CSLOGISTIC and CSTABULATE) along with discharge weights were used in SPSS to account for the stratified sampling design of the NIS.
Demographics of Meningioma Surgery Patients Stratified by Race
Between the years 2012 to 2014, 16,000,000 cases were available in the NIS database. 7,745 patients were identified as those with meningioma who had undergone a resection procedure. Following the exclusion of cases with missing data, patient demographics or hospital characteristics, 7,093 patient cases remained. In this group of meningioma surgery patients, there were 70.0% white, 12.4% black, 9.1% Hispanic, 3.7% Asian/Pacific Islander, 0.5% Native American and 4.3% “other” patients (see Table 2).
Age, payer, median income quartile, hospital region, hospital teaching status and mean Elixhauser Comorbidity Index all significantly differed among the race cohorts. Meningioma surgeries showed a female predominance (68.1%), but differences in gender composition among the race cohorts were not significant. Private insurance usage was significantly higher among Asians (51.1%) and whites (48.1%) compared to blacks (41.6%) and Hispanics (39.9%) (p<0.05). A similar trend was seen in Medicare usage, which was significantly higher in whites (39.9%) compared to blacks (30.5%), Asians (27.2%) and Hispanics (26.6%) (p<0.05). Medicaid and uninsured payer types were more commonly used among Hispanics (28.4%) and blacks (24.0%) but less often used among Asians (18.3%) and whites (9.3%) (p<0.05). Analysis of median income quartiles among the race cohorts of meningioma patients indicated Asian (48.1%) and white (28.1%) patients more frequently resided in the highest median income quartile zip codes compared to Hispanics (17.8%) and blacks (14.0%) (p<0.05). Race cohorts with a greater proportion in the lowest income quartile zip codes (0th to 25th percentile) included blacks (42.1%) and Hispanics (33.3%) (p<0.05). The mean Elixhauser Comorbidity Index was higher in blacks (2.481) and Asians/Pacific Islanders (2.626) compared to whites (1.451) and Hispanics (1.288) (p<0.05) (see Table 2).
Univariate Analysis of Complications, Total Charges, and Length of Stay Stratified by Race Complications
Chi-square analysis showed a significant difference among the race cohorts in the incidence of complications. There was no statistical significance in mortality. Black patients, compared to whites, had higher incidences of almost all types of complications, including operative (16.7% vs. 12.5%), pulmonary (11.8% vs. 7.2%), neurological (10.1% vs. 7.0%), urinary/renal (5.1% vs. 2.7%) and infectious complications (2.3% vs. 1.0%) (p<0.05) (see Table 3). Asians/Pacific Islanders and Hispanic patients had higher incidences of complications than whites, such as operative (21.1% vs. 12.5%), pulmonary (10.0% vs. 7.2%), neurological (11.1% vs. 7.0%) and infectious (1.9% vs. 1.0%) complications, but these patients had a similar incidence of urinary/renal complications (2.6% vs. 2.7%) (p<0.05). Hispanics and whites differed to a lesser degree, with slightly higher incidence of pulmonary (8.2% vs. 7.2%), neurological (7.5% vs. 7.0%), infectious (1.2% vs. 1.0%) and operative (15.3% vs. 12.5%) complications in Hispanics (p<0.05). In summary, blacks had the highest incidences of overall complication at 32.6%, while whites had an incidence of 24.0% (p<0.05) (see Table 3).
Total Charges & LOS
The average total charges and LOS for meningioma surgery were significantly higher in minority groups compared to whites. Asian/Pacific Islander had the highest charges ($180,795.45), followed by Hispanics ($160,355.63) and blacks ($136,522.56), while whites had the lowest charges ($118,495.50) (p<0.05). In terms of LOS, blacks had the highest (8.35 days), followed by Hispanics (7.20 days) and Asian/Pacific Islanders (7.34 days) compared to whites (5.92 days) (p<0.05) (see Table 3).
Multivariate Analysis of Adverse Outcomes Stratified by Race
Multivariate regression indicated increased odds of experiencing multiple complications in black patients, including pulmonary (OR [95%CI], p-value: 1.588 [1.201-3.021], p<0.05), urinary/renal (1.799 [1.217-2.657], p<0.05), infectious (1.287 [1.038-1.595], p<0.05) and operative complications (1.287 [1.038-1.595], p<0.05) compared to whites. Asians/Pacific Islanders were more likely than whites to have operative complications (1.578 [1.136-2.192], p<0.05) (see Table 4).
High total charges were more likely experienced by blacks (1.870 [1.494-2.341], p<0.05) and Hispanics (1.418 [1.077-1.866], p<0.05) when compared to whites. Odds of prolonged LOS were significantly higher in blacks (1.910 [1.493-1.443], p<0.05) and Hispanics (1.529 [1.119-2.088], p = 0.008). There was no significant difference in mortality among all of the race cohorts (see Table 4).
Several qualitative studies have noted that patients who undergo meningioma surgery face barriers to high-quality healthcare, lack of information regarding postsurgical management and lack of psychosocial support groups, all of which may disproportionately affect minorities.18,19 Lack of access to primary care is also considered a barrier for black patients in receiving a specialty referral and high-quality care.20 Anzalone and colleagues further suggested that access to healthcare and neurodiagnostic imaging like magnetic resonance imaging may play a role in adverse outcomes in blacks, who were more likely to present in later stages with larger tumors.21
Race has been suggested as a predictor of mortality for meningioma patients in several studies of other patient databases. One SEER analysis of benign intracranial meningiomas by Cahill and colleagues found that black patients faced a 30% increased risk of death compared to whites, although no other minorities faced increased risk.22 Another SEER analysis of intracranial meningiomas by Bhambavani and colleagues indicated worse survival for black patients, but improved survival for Asian/Pacific Islander patients compared to whites.23
Similarly, Yang and colleagues found 10% increased odds of mortality in black patients with decreased odds of mortality by 21% in Asian/Pacific Islander patients and 31% in Hispanic patients in a National Cancer Database analysis of patients treated for meningioma.24 Generally, studies seemingly showed a common thread of increased mortality odds or lower survival in black patients, while Asian patients had improved survival.
Our findings through analysis of the NIS did not reveal any significant increase or decrease in odds of mortality in black and other minority patients compared to white patients. This may potentially be explained through the limited ability of aforementioned SEER studies to adjust for comorbidities in multivariate analysis. Our analysis showed black patients had a relatively high Elixhauser Comorbidity Index score compared to white patients (2.481 vs. 1.451, p<0.001), potentially confounding mortality outcome analysis. Our study utilized the Elixhauser Comorbidity score to effectively adjust for these differences in the multivariate analysis.17
Our study revealed several important findings in racial disparities of postoperative complication rates. Univariate chi-square analysis showed that black, Hispanic, and Asian/Pacific Islander patients all had higher rates of multiple complications compared to white patients, including cardiovascular, pulmonary, neurological, infectious and operative complications. In a Thomson Reuters Marketscan univariate analysis of postoperative outcomes in meningioma patients using Medicaid, Mukherjee and colleagues found 30-day complication rates of 26.7% and 19.1% in blacks vs. whites, respectively. This is similar to our findings of 32.6% black patients vs. 24.0% white patients with postoperative complications.25 After adjusting for several important confounders, multivariate analysis in our study showed nearly twice the odds of urinary/renal and infectious complications in black patients. Furthermore, pulmonary and operative complications were more likely in blacks compared to whites, and Asian/Pacific Islanders had nearly 60% higher odds of operative complications. However, Hispanics did not show significantly increased or decreased odds of mortality relative to whites. Mukherjee and colleagues found the odds of postoperative complications to be 32% higher in blacks than whites.25 Interestingly, we similarly found 46% increased odds of postoperative complication in black patients. Generally, minority groups had significantly higher rates of complications, especially evident in black patients.
Mukherjee and colleagues also found 32% increased odds of prolonged LOS and 85% higher odds of total charges for black patients in a multivariate analysis.25 Our findings also showed blacks having higher odds of these outcomes, specifically 91% increased odds of prolonged LOS and 87% increased odds of high total charges. Furthermore, our study showed odds of prolonged LOS and high total charges in Hispanics were higher by 53% and 42%, respectively. We defined “high” or “prolonged” as values greater than the 90th percentile for the sample, while Mukherjee may have used a slightly different methodology, likely accounting for the variations in the odds for black patients.
Although the NIS is a powerful tool for providing insight into clinical events and outcomes, there are several limitations of this study that we must acknowledge. Racial coding cannot be confirmed as completely accurate for participating institutions in the NIS, and race was the most often missing variable for discharge data, as not all contributing institutions provide race or ethnicity information. It must also be considered that NIS is a claims-based database that is primarily used for reimbursement purposes, which may introduce its own biases compared to other databases used for quality improvement analysis, such as the American College of Surgeons National Surgical Quality Improvement Program database.26 Furthermore, the NIS does not provide any information on the stage, grade or subtype of cancer, preventing any adjustment for these differences in multivariate regression. The NIS additionally does not contain follow-up data post-discharge, preventing us from capturing all postoperative complications. Finally, errors in ICD-9 coding by contributing institutions may affect the quality of NIS data used in this study.
Minorities were generally more likely to face adverse outcomes, particularly black patients who had higher odds of complications, prolonged LOS and high cost of care compared to white patients. Therefore, measures must be taken to improve the quality of care for minorities undergoing surgery for intracranial meningioma.