Impact of insurance on hospital course and readmission after resection of benign meningioma

Abstract

Introduction:

Surgical outcomes and healthcare utilization have been shown to vary based on patient insurance status. We analyzed whether patients’ insurance affects case urgency for and readmission after craniotomy for meningioma resection, using benign meningioma as a model system to minimize confounding from the disease-related characteristics of other neurosurgical pathologies.

Methods:

We analyzed 90-day readmission for patients who underwent resection of a benign meningioma in the Nationwide Readmission Database from 2014–2015.

Results:

A total of 9783 meningioma patients with private insurance (46%), Medicare (39%), Medicaid (10%), self-pay (2%), or another scheme (3%) were analyzed. 72% of all cases were elective; with 78% of cases in privately insured patients being elective compared to 71% of Medicare (p>0.05), 59% of Medicaid patients (OR 2.3, p<0.001), and 49% of self-pay patients (OR 3.4, p<0.001).

Medicare (OR 1.5, p = 0.002) and Medicaid (OR 1.4, p = 0.035) were both associated with higher likelihood of 90-day readmission compared to private insurance. In comparison, 30-day analyses did not unveil this discrepancy between Medicaid and privately insured, highlighting the merit for longer-term outcomes analyses in value-based care. Patients readmitted within 30 days versus those with later readmissions possessed different characteristics.

Conclusions:

Compared to patients with private insurance coverage, Medicaid and self-pay patients were significantly more likely to undergo non-elective resection of benign meningioma. Medicaid and Medicare insurance were associated with a higher likelihood of 90-day readmission; only Medicare was significant at 30 days. Both 30 and 90-day outcomes merit consideration given differences in readmitted populations.

Introduction

Socioeconomic disparities drive differences in post-operative outcomes across a variety of procedures [1–5]. Insurance type can be affected by multiple factors: state of residence, age, income, and certain chronic co-morbidities, amongst others, and demonstrates high collinearity with race and education attainment [6,3,7]. Certain insurance types limit access to preventative care, which may lead to a greater frequency of acute hospital presentations that result in non-elective surgical procedures [8]. National and state policy can also influence the type of insurance a patient carries, which in turn influences patient access to care. The nature of a patient’s insurance coverage has been implicated in post-operative morbidity during hospital stay and after discharge [9,2,10].

Meningioma is the most common intracranial tumor and is regularly diagnosed amongst an adult population more reflective of the general population in health comorbidities [11]. The primary treatment modality for those necessitating intervention is operative resection, which can be curative in many patients [12,11]. We focused on benign meningioma to minimize confounding from the disease-related characteristics of other neurosurgical pathologies. We examined patients undergoing resection of an intracranial meningioma as a model system for investigating the impact of patient insurance coverage on access to healthcare, degree of emergent or elective status on presentation for surgery, and rate of readmission after surgery for benign meningioma.

Methods

Data source

We analyzed cases within the Nationwide Readmissions Database (NRD) from 2014–2015. The NRD contains data from approximately half of all hospital discharges in the United States. The NRDs have previously been utilized to study hospital readmissions [13–15]. The Partners Institutional Review Board has exempted analyses of deidentified nationwide databases from review, and obtaining individual patient consent is not required for these analyses.

Inclusion and Exclusion Criteria

Inclusion criteria included an International Classification of Diseases 9 (ICD-9) code corresponding to benign meningioma (225.2); ICD-9 procedural code for craniotomy for tumor resection (01.2); age over 18 years; and discharge between January and September. The discharge month range was selected so 90- day readmissions could be calculated in total. Malignant meningiomas were not included due to the more likely administration of radiation or chemotherapy for aggressive tumors, which serve as possible confounding factors and are not captured in this dataset.

These criteria yielded a case load of 9907 cases. Cases were excluded for death during index hospitalization (n=121) and a post-operative length of stay of 0 days (n=3), suggesting an aborted or miscoded procedure, yielding a final cohort of 9783 cases.

Primary and Secondary Outcomes

The primary outcome was 90-day readmission rate. Secondary outcomes included 30-day readmission rate, the elective nature of operations, and total hospital charge for the index hospitalization. For patients with multiple readmissions within this timeframe, only the first readmission was considered.

Co-Variates

Variables extracted from the NRDs included patient age and sex; insurance (private, Medicaid, Medicare, or other); chronic co-morbidities based on Charlson Comorbidity Index variables included in the NRD are (diabetes mellitus [DM], congestive heart failure [CHF], chronic lung disease, acquired immune deficiency syndrome [AIDS], and lymphoma) as well as hypertension given previously observed associations between hypertension and post-operative outcomes; patient income quartile (relative to geographic location); hospital bed size (relative to geographic location and hospital type); hospital teaching status; hospital location (including large and small metropolises [defined by a population threshold of over 1 million for large metropolises and less than 1 million for small metropolises], compared to an unincorporated rural region, micropolitan locations, and non-urban areas); benign meningioma operative volume at a given hospital; and case urgency. Dual-eligible patients who were encoded to be on Medicaid but met Medicare criteria (at least 65 years old, ICD-9 diagnosis of end-stage renal disease or stage 5 chronic kidney disease [ICD9 585.5 or 585.6], or ICD-9 diagnosis of amyotrophic lateral sclerosis [ICD9 335.20]) were categorized as Medicare; these represented 4.9% of patients listed with Medicaid as their primary insurer. Operative volume was divided into quartiles by institution based on the full range of operative volumes.

Age was analyzed as a linear variable; all other variables were analyzed as categorical. We categorize insurance as private, Medicare, Medicaid, and other (including other government schemes and uninsured or self-pay patients), and use private insurance as the reference group for all analyses.

Missing Data

Insurance status was missing for 0.1% of cases. Hospital charges, income, and case urgency were missing in 4.3%, 1.9%, and 0.3% of cases. There were no missing data for age, sex, co-morbidities, and hospital features. Missing data were imputed through multiple imputation using the “amelia” package within The R Project for Statistical Computing 3.5.2 [16].

A total of 10 imputed datasets, versions of the dataset with missing values mathematically estimated based on other variables, were creating using demographic information, co-morbidities, estimated income quartile by zip code, hospital features, case urgency, insurer, and the outcomes of 90-day readmission, length of stay, and total hospital charge. 30-day readmissions were not included in the imputation model due to collinearity with 90-day readmissions. We included outcomes in our imputation to improve the specification of our model, increasing the accuracy and quantity of data available for the model to use when generating estimates of missing values [17].

Statistical Analysis

Descriptive statistics were initially conducted utilizing the chi-square test for categorical variables and the ANOVA test for comparison of multiple means. Multivariate linear and logistic regression were utilized to generated adjusted odds ratios for outcomes of interest. Odds ratios are presented with 95% confidence intervals (95% CIs). Interaction terms were applied to regression models to gauge the effect of age on relationships between insurance and outcomes.

Unless otherwise noted, multivariate analyses included adjustment for demographic variables (age, gender, estimated income quartile by zip code), select comorbidities (congestive heart failure, chronic lung disease, depression, diabetes mellitus, hypertension, obesity), and unweighted hospital volume within the NRD.

The R Project for Statistical Computing 3.5.2 was utilized for all statistical tests. GraphPad Prism version 7 (San Diego, CA) was utilized to generate figures. A probability value of less than 0.05 was used as the cut-off for statistical significance for descriptive statistics and regression main effects. A p value cut-off of 0.25 was utilized as the cut-off for statistical significance for interaction terms.

Results

Patient Demographics

9,783 patients (median age 59 years, IQR 49–69 years; 69% female) were included in analysis (Table 1). Patients were covered by private insurance (46%), Medicare (39%), Medicaid (10%), self-pay (2%), or other insurance types (3%). Insurance type was unknown for 13 cases (0.1%). Though Medicaid patients were more likely to be in the lowest quartile of income, Medicaid patients in the first or second quartiles across the entire cohort were younger than the average Medicaid patient in the cohort. Relative to patients with private insurance, patients on Medicaid or Medicare had a higher prevalence of co-morbidities in general (Table 1).

Table 1.

Analysis of demographic and clinical characteristics demonstrate that compared to privately insured patients, Medicaid patients tend to have more comorbidities and are more likely to be in a lower income bracket. Medicare patients tend to be older and have more comorbidities than privately insured patients as well.

	Total (n=9783)		Private (n=3231)		Medicaid (n=973)		Medicare (n=2437)		Self-Pay (n=234)		Other (n=315)		Unknown (n=13)		p
Age	Median	IQR	Median	IQR	Median	IQR	Median	IQR	Median	IQR	Median	IQR	Median	IQR
Years	59	14	52	11	48	12	70	10	50	11	52	12	52	12	< 0.001
Sex	N	%	N	%	N	%	N	%	N	%	N	%	N	%
Female	6731	68.8	3231	72.1	730	71.3	2410	64.8	157	67.1	196	62.2	7	53.8	< 0.001
Co-morbidity	N	%	N	%	N	%	N	%	N	%	N	%	N	%
AIDS*	5	0.1	1	0.0	1	0.1	3	0.1	0	0.0	0	0.0	0	0.0	0.715
Hypertension	5135	52.5	1788	39.9	417	42.9	2651	70.3	114	48.7	158	50.2	7	53.8	< 0.001
Diabetes Mellitus	1975	20.2	612	13.7	180	18.5	1094	29.0	38	16.2	48	15.2	3	23.1	< 0.001
CHF**	235	2.4	30	0.7	18	1.8	181	4.8	4	1.7	2	0.6	0	0.0	< 0.001
Chronic Pulmonary	1182	12.1	444	9.9	125	12.8	561	14.9	29	12.4	22	7.0	1	7.7	< 0.001
Liver Failure	133	1.4	55	1.2	14	1.4	58	1.5	2	0.9	4	1.3	0	0.0	0.738
Lymphoma	40	0.4	14	0.3	4	0.4	22	0.6	0	0.0	0	0.0	0	0.0	0.192
Renal Disease	403	4.1	66	1.5	24	2.5	292	7.7	10	4.3	0	0.0	0	0.0	0.36
Income	N	%	N	%	N	%	N	%	N	%	N	%	N	%
Lowest Quartile	2090	21.4	690	15.4	343	35.3	870	23.1	86	36.8	95	30.2	6	46.2
Second Quartile	2293	23.4	962	21.5	246	25.3	946	25.1	68	29.1	68	21.6	3	23.1	< 0.001
Third Quartile	2475	25.3	1200	26.8	201	20.7	958	25.4	39	16.7	75	23.8	2	15.4
Highest Quartile	2731	27.9	1556	34.7	159	16.3	928	24.6	33	14.1	53	16.8	2	15.4
Unknown	194	2.0	71	1.6	24	2.5	67	1.8	8	3.4	24	7.6	0	0.0

^*Acquired Immune Deficiency Syndrome

^**Congestive Heart Failure

Treating Hospital Location and Characteristics

We investigated whether the proportion of patients with a particular insurance type varied across hospitals located in urban versus rural environments. Patients included in this study were treated at hospitals in small (30%) or large metropolises (70%), as divided across threshold population of 1 million; a small minority were treated in suburban or rural centers (total <0.1%). Treatment occurred predominantly at hospitals with large bed-size, defined relative to the population of the geographic region in which the hospital is located and relative to other hospitals of the same type in that region (rural, urban nonteaching, and urban teaching) (77%). In both small and large metropolises, the majority of patients underwent resection at teaching hospitals (84 and 93%, respectively). Selection of a teaching hospital by patients were comparable across insurance type; with 91% of private insurance and Medicaid, 88% of Medicare, and 90% of self-pay patients treated at teaching hospitals.

We categorized treatment centers by volume of benign meningioma resections into quartiles. While treatment centers within the highest volume quartile constituted 24% of treatment sites, 69% of all patients received treatment at one of these sites (Figure 1). 75% of patients in the highest income quartile were treated at a high-volume center, compared to 70%, 67%, and 62% in the third, second, and lowest income quartiles (χ², p<0.001). Consistent with the income trends, patient insurance was also associated with treatment at a high-volume center: 74% of patients on private insurance received treatment at a high-volume center, compared to Medicaid (63%), Medicare (65%), and self-pay patients (65%) (χ ², p<0.001).

Figure 1.

The majority of patients were operatively treated at a high-volume center (HVC), defined as the top quartile of centers in terms of operative benign meningioma volume. Patients with private insurance were more likely to receive treatment at one of these centers.

Hospital Course

We hypothesized that the clinical course of patients, including their case urgency and post-operative length of stay, associated with their insurance status. The majority of cases overall were elective (72%) rather than urgent or emergent. Patients with private insurance were more likely to have an elective case designation (78%) compared to those with Medicaid (59%), Medicare (71%), or self-pay (49%) (Table 2).

Table 2.

Medicaid, self-pay, and other insurance are associated with non-elective designation.

	Non-elective Designation*
	%	OR	95% CI	p
Private	78.1	ref	-	-
Medicaid	59.0	2.27	1.95–2.64	<0.001
Medicare	71.0	1.10	0.97–1.25	0.15
Self-Pay	48.9	3.41	2.60–4.48	<0.001
Other	69.5	1.46	1.13–1.88	0.004

^*excluding 30 cases with missing data on case urgency

On multivariable analysis, Medicaid and self-pay were associated with non-elective case designation compared to patients with private insurance (OR 2.27 (95%CI 1.94–2.63) and 3.42 (95%CI 2.60–4.49), both p<0.001). Within this multivariable model, male sex (OR 1.23, p<0.001) and treatment in non-high volume centers (OR 1.66, 1.90, and 2.38 for the 3^rd, 2^nd, and lowest quartiles, respectively, all p<0.001) were also associated with non-elective case designation.

The median length of stay (LOS) for the cohort was 4 days (IQR 3–8 days) and demonstrated a positive skew (mean = 5.8 and 12.3 days for elective and non-elective cases, respectively). Multivariate linear regression demonstrated that Medicaid and Medicare patients both stayed 1.6 days longer than patients on private insurance, respectively, after adjustment for patient demographics, co-morbidities, case urgency, and hospital volume by quartile (both p<0.001).

Readmission

The overall 90-day readmission rate was 5.8%. This rate was 4.4% for privately insured patients, 5.8% for patients on Medicaid, 7.5% for patients on Medicare, and 5.6% for self-pay patients (Table 3). Compared to private insurance, Medicaid insurance (OR 1.40, 95%CI 1.02–1.91, p=0.035) and Medicare insurance (OR 1.47, 95%CI 1.216–1.88, p=0.002) were associated with 90-day readmission, while self-pay and other insurances were not, on multivariable analysis (Figure 2, Panel 1). In contrast, overall 30-day readmission rate was 2.3% for privately insured patients, 3.1% for patients on Medicaid, 3.5% for patients on Medicare, and 1.7% for self-pay patients. Medicare insurance was also associated with 30-day readmissions compared to private insurance (OR 1.44, 95%CI 1.03–2.02, p=0.034) (Figure 2, Panel 2).

Table 3.

Descriptive statistics on primary and secondary outcomes are presented.

	Private	Medicaid	Medicare	Self-Pay	Other
90-day readmission	4.4%	5.8%	7.5%	5.6%	6.0%
30-day readmission	2.3%	3.1%	3.5%	1.7%	3.5%
Total index hospital charge (median)	$95,943	$121,703	$112,926	$119,136	$90,813

Figure 2.

Panel 1: (A) Patients on Medicare demonstrate elevated unadjusted odds ratios of 90-day readmission relative to privately insured patients. These odds ratios are calculated without accounting for any possible confounding factors. (B) Patients on Medicaid and Medicare demonstrate elevated adjusted odds ratios for 90-day readmission, which account for demographic factors, comorbidities, and treating hospital characteristics that could confound the relationship between insurance status and readmission. Red bars indicate p < 0.05.

Panel 2: Patients on Medicare demonstrate higher (A) unadjusted and (B) adjusted odds ratios of 30-day readmission relative to privately insured patients. Unadjusted odds ratios are calculated without accounting for potential confounding variables, while adjusted odds ratios account for demographic factors, comorbidities, and treating hospital characteristics that could confound the relationship between insurance status and readmission. Red bars indicate p < 0.05.

Panel 3: Subgroup analyses of the association between insurance and 90-readmission at (A) high-volume centers and (B) non-high-volume centers are shown.

Panel 4: Subgroup analyses of the association between insurance and 90-readmission for elective cases (A) and non-elective cases (B) are shown.

These models were adjusted for patient demographics and comorbidities, as well as hospital characteristics.

We further compared whether patients readmitted within 90 days reflected a different population than those readmitted within 30 days. To do so, we compared the 289 patients with a first readmission between 30 and 90 days after their operations (“late readmission”) with the 279 patients in our cohort with a first readmission within 30 days (“early readmission”). Patients in the late readmission group were more likely to be older (p=0.049), have a non-elective case designation (p=0.024), and a longer index hospitalization length of stay (p=0.004) than the early readmission group. Late readmission patients are also more likely to be in the top quartile of income than early readmission patients (p=0.003). Sex and medical comorbidities were not related to the timing of readmission, though readmitted patients at both time points had an increased prevalence of comorbidities.

We conducted subgroup analyses on patients stratified by treatment at a high-volume center and by case urgency. Medicaid patients were more likely than patients with private insurance to be readmitted at 90 days if they were treated at a at a high-volume center, but this relationship did not hold for Medicaid patients treated at a non-high-volume center. Medicare patients in both groups did not have a significant association with 90-day readmission as compared to privately insured patients. Figure 2, Panel 3). Case urgency did not alter the relationship between Medicaid and 90-day readmission, but did so for Medicare patients: Medicare insurance was more strongly linked to 90-day readmissions for non-elective cases than elective cases (Figure 2, Panel 4). There was no significant direct association between either center volume or case urgency and readmission at the 30 and 90-day time points.

Age was found to inversely modify the effect of Medicaid and Medicare insurance on 90-day readmission in a statistically significant manner (p = 0.224, p = 0.196) (Supplementary Digital Content 1). The adjusted ORs for 90-day readmission for 25 and 60-year-old patients on Medicaid relative to privately insured patients of the same age are 2.22 and 1.22, respectively. This indicates that, while Medicaid patients in general have greater odds of 90-day readmission compared to patients with private insurance, this phenomenon is more pronounced among younger patients.

Charges

The median hospital charge was $104,900 (IQR 62,990 – 174,500). The highest median hospital charges were observed for Medicare patients, followed by self-pay patients (Table 3). The adjusted total charges for patients on Medicaid were $38,060 higher on average (p<0.001) and for patients on Medicare were $13,130 higher (p<0.001) than private insurance, after adjustment for age and medical co-morbidities.

Discussion

Readmissions present a major burden on the health care system and individual patients, and the causes underlying these readmissions likely reflect discrepancies in patients’ health and access to care that may extend beyond this hospital visit [18]. We evaluated the impact of insurance coverage on the hospital course and readmissions for patients undergoing craniotomy for benign meningioma [2,9,10]. Our data unveiled unique contributions from the analysis of 90-day outcomes compared to the immediate 30-day outcomes, highlighting the merit for longer-term outcomes analyses in value-based care. Furthermore, the patterns observed in this pathology may be extrapolatable to the broader surgical population.

Operative Acuity

We observe a predisposition for Medicaid and self-paying patients to undergo non-elective meningioma resection, which may reflect a tendency to seek care less frequently. This correlation may reflect more advanced tumor progression prior to presentation and echoes observations of delayed presentation in other operative pathologies [19,9,20]. Underinsured patients also face barriers in seeing primary care physicians, affording imaging, and finding in-network neurosurgeons.

Location of Care

Patients on Medicaid and Medicare were less likely to receive care at a high-volume center than privately insured patients. This may reflect geographic discrepancies as well as the provider networks that different plans cover, though the opposite trend has been previously observed among patients arriving at the emergency department following a trauma [21]. Individual surgeon and center volume have been linked to patient outcome following surgery across multiple neurosurgical pathologies [22–26]. Medicare coverage is associated with 90-day readmissions only among patients treated at high-volume centers, while Medicaid coverage is not significantly associated with 90-day readmissions in either group. This suggests that hospital location may confound the associations between insurance and 90-day readmission or that our statistical power was too low for precise analysis of hospital location.

Readmission

Medicare is associated with higher likelihood for readmission at both 30 and 90 days, while an elevated risk for readmission in Medicaid patients was significant at only 90 days. This discrepancy between 90- and 30-day readmissions associations may reflect the underpowering of the study size or behavioral trends of different populations after surgery. Interestingly, patients with late readmission group tends to be older, more likely to present non-electively, and more likely to have longer hospital stays. While further analysis of temporal patterns of readmissions is warranted, these patterns suggest that the late readmission group may be enriched for patients who are sicker and have less access to medical care. The common 30-day readmission metric does not capture these patients, thus reducing its generalizability.

Patients on Medicaid insurance may be susceptible to readmission since their coverage may disincentivize close clinic follow-up, where post-operative concerns can be discussed [2,9,20,27–30]. Instead, these patients may present to the emergency department after concerns that could have been addressed at clinic progress. We identified age as a mediator for the relationship between Medicaid and readmission – younger Medicaid patients may lack social supports to help them make to appointments. The Medicare population is older and has more chronic co-morbidities, as verified in this cohort, which may partially drive their readmission risk in ways we were not able to control for with multivariable regression.

Efforts to engage Medicaid and Medicare patients in primary care soon after a hospital discharge have proven efficacy in reducing readmissions in these populations [31,32]. Neurosurgeons can identify patients at high risk for readmission prior to an operation and can work with support staff in clinics to aid these patients in communicating post-operative problems with the clinic rather than an emergency room. these patients in communicating post-operative problems with the clinic rather than an emergency room.

Hospital Charges

We found higher hospital charges for all non-private insurance groups. Hospital charge does not necessarily equate the cost for the patient, which is calculated after negotiation with the insurer. However, it does suggest an increased complexity of a hospitalization that required more billed procedures, prescriptions, and orders. Given that Medicaid and Medicare patients both experienced longer length of stays than patients on private insurance, additional days in the hospital due to complications or discharge planning may partially drive this high charge.

Limitations

Our study has pertinent limitations related to the database we study and our methodology. Tumor-level variables such as size and location of the tumor influence operative outcomes but were not available for our analysis. Further, the extent of resection achieved, associated neurologic and functional status, and the receipt of post-operative radiation were not available. The NRD does not contain detailed geographic information, and health care quality and availability of insurance differ widely in different areas of the country. We are unable to assess the causes of readmission since this information is not explicitly coded into the NRD, which limited analysis of the preventability of readmissions and comparative differences between indications for readmission across different insurance groups. Previous analysis of readmissions following resection of benign meningioma have indicated that causes of readmission within 30 days do not vary significantly from causes of readmission beyond 30 days [33]. Unfortunately, we were unable to pursue a similar analysis within the context of this large patient sample.

Implications of Findings

This nationwide retrospective cohort analysis has implications for both clinical practice and health policy. Operationally, early identification of patients at increased risk for postoperative readmission may prompt increased communication with and screening of patients in the follow-up period to identify potentially emerging concerns before they fulminate to necessitate emergent assessment. Our findings also merit follow-up in studies that examine readmissions extending beyond the traditional 30-day window, including investigations that capture details regarding the causes of these readmissions. Our analysis was limited by the nature of the available dataset, but the identified patterns support the extension of readmission consideration to at least 90 days after surgery in future research to better inform healthcare policy. The reasons for the disparities in readmissions seen by 90 days merit further exploration. Interventions to reduce readmissions between the 30 and 90-day time points include closer follow-up and improved access to social services during this interval.

Additionally, the increased likelihood of patients with Medicaid insurance to require non-elective resection in particular supports the re-examination of insurance benefit structures to ensure adequate access to neurosurgical specialist care throughout the course of disease. Delays in care, including elective surgery for benign meningioma, for patients with Medicaid are likely multi-factorial, and therefore, may be amenable to interventions. Improving Medicaid benefits as well as referral pathways to specialists for treatable conditions may improve outcomes for patients with benign meningioma, among other conditions. Though patients with Medicaid may have more comprehensive coverage than some patients with private insurance, low reimbursement rates can create functional “underinsurance” due to a critical number of medical facilities not accepting Medicaid patients or de-prioritizing access [34,35]. Increasing the level of reimbursement offered to facilities caring for patients with Medicaid coverage may help to improve timely patient access to specialist appointments for both urgent and non-urgent conditions. Patients with Medicaid coverage are also more likely to be cared for in safety net hospitals with reduced resources and capacity, which may have poorer performance [36]. Each of these factors is a prospective target for policy intervention.

We observed that patients of lower socioeconomic status, those with pre-existing conditions, and older patients who may lack adequate social support networks were disproportionately likely to experience postoperative readmissions that may not be captured by traditional 30-day readmission metrics. We also found that younger Medicaid patients were more likely than older Medicaid patients to require readmission, and this may similarly reflect an unmet need for social support. These patients may benefit from intentional provider follow-up after their initial discharge and other disparity-targeted health policies to lower readmission rates.

We contribute additional context on the application of non-elective versus elective resection of benign meningioma as a measure of healthcare access for neurosurgical care. The American College of Surgeons has championed an effort to define “disparity-sensitive” measures of surgical access and outcomes based on insurance status and other factors [37,38]. Some of these measures, such as rates of urgent to elective surgery for a particular condition, can then be used assess access to surgical care for a patient population or at the hospital level and to measure improvement [39,40]. For example, perforated to non-perforated appendicitis rates have been used to measure the effects of the Affordable Care Act Medicaid expansion program on access to surgical care in the US [41]. As we have characterized the patient populations who undergo non-elective versus elective resection of benign meningioma within the context of a large national database, this metric may facilitate assessment of surgical outcomes and access in future studies.

Conclusion

Using craniotomy for benign meningioma as a model system to minimize confounding from other neurosurgical pathologies, we evaluated the relationship between insurance coverage, case acuity, and hospital readmissions. Compared to patients with private insurance coverage, Medicaid and self-pay patients were significantly more likely to undergo non-elective resection. This may be the result of a reduced tendency to seek care in response to inadequate insurance coverage. Additionally, patients with Medicare insurance were significantly more likely than patients with private insurance to be re-admitted within both 30 and 90 days. This association was significant for Medicaid patients only at 90 days. These relationships may be confounded by geographic location and treatment center volume. Trends towards older age, greater operative acuity, and longer hospital stays in the population of patients readmitted between 30 and 90 days suggest that this group may be enriched for sicker patients with reduced access to care. The higher hospital charges for all non-private insurance groups may also reflect this discrepancy.

Supplementary Material

11060_2020_3581_MOESM1_ESM

Supplementary Digital Content 1. Table. Age and insurance on 90-day readmissions. The interaction term (β₁) can be used to calculate how the odds ratio (β) changes with a patient’s age and is a figure adjusted for other patient demographics, income, major co-morbidities, and treatment hospital features.

Figure 3.

Characteristics of patients readmitted between 0–30 days after their operation and those readmitted between 30–90 days after their operation. If a patient experienced multiple readmissions, only the first readmission is reflected.