Insurance Disparities in the Outcomes of Spinal Cord Stimulation Surgery

Insurance Disparities in the Outcomes of SpinalCord Stimulation SurgeryKevin T. Huang, BA*; Matthew A. Hazzard, MD*; Ranjith Babu, MS*;Beatrice Ugiliweneza, PhD†; Peter M. Grossi, MD*; Billy K. Huh, MD, PhD‡;Lance A. Roy, MD‡; Chirag Patil, MD†; Maxwell Boakye, MD§,¶;Shivanand P. Lad, MD, PhD*

Objectives: The Affordable Care Act aims to expand health insurance and to help narrow existing health care disparities.Medicaid patients have previously been noted to be at an increased risk for impaired access to health care, delayed medicaltreatment, and the receipt of substandard care. Conversely, those with commercial insurance may be subject to overtreatment. Thegoal of this study was to evaluate how Medicaid versus commercial insurance status affects outcomes following spinal cordstimulation (SCS) surgery.

Materials and Methods: A retrospective cohort study of 13,774 patients underwent either percutaneous or paddle permanent SCSimplantation, selected from the Thomson Reuter’s MarketScan database between 2000 and 2009. Patients were characterized by ageat initial procedure, gender, baseline comorbidity burden, procedure-associated diagnosis code, follow-up, and type of insurance(Medicaid vs. commercial insurance). Outcome measures included probability of reoperation, timing and type of reoperation, pres-ence of postoperative complications (immediate, 30 days, and 90 days), and overall utilization of health resources postoperatively.Multivariate analysis was performed comparing the relative effect of insurance status on outcomes following initial surgery.

Results: Medicaid patients had greater healthcare resource utilization as measured by medications prescribed, emergencydepartment visits, and length of stay; however, commercially insured patients had significantly higher overall costs ($110,908 vs.$64,644, p < 0.0001). Commercial and Medicaid patients did not significantly differ in their complication rates during the indexhospitalization or at 30 days or 90 days postoperatively. The group were also not significantly different in their two-year reopera-tion rates (7.32% vs. 5.06%, p = 0.0513).

Conclusions: There are substantial insurance disparities that affect healthcare utilization and overall cost following SCS. Efforts fornational healthcare reform should examine system factors that will reduce socioeconomic disparities in outcomes following SCS.

Keywords: Disparities, insurance coverage, spinal cord stimulation, socioeconomics

Conflict of Interest: The authors report no conflict of interest.

INTRODUCTION

Spinal cord stimulation (SCS) is an increasingly widely employedstrategy in the management of several chronic pain conditions,including failed back surgery syndrome, complex regional pain syn-drome, and peripheral neuropathy (1).

Socioeconomic factors, in particular insurance disparities, havelong been known to impart a significant impact on patients’ access tocare. Currently, the largest health coverage program in the UnitedStates, Medicaid, serves approximately 67 million individuals (2).Eligibility varies for each state, but in general Medicaid provideshealth coverage for lower-income people, the elderly, families andchildren, and individuals with disabilities (3). Nearly 61% of adultMedicaid enrollees have a chronic or disabling condition (4,5).Roughly half of Medicaid beneficiaries under age 65 are racial andethnic minorities who may experience barriers to care, a greaterincidence of chronic disease, lower quality of care and higher mor-tality than the general population (6). While these patients may beserved by similar healthcare providers as those patients insuredcommercially, not all physicians in the United States accept Medicaid

Address correspondence to: Shivanand P. Lad, MD, PhD, Division ofNeurosurgery/Department of Surgery, Duke University Medical Center, Box3807, Durham, NC 27710, USA. Email: [email protected]

* Division of Neurosurgery, Department of Surgery, Duke University MedicalCenter, Durham, NC, USA;

† Department of Neurosurgery, Center for Neurosurgical Outcomes Research,Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, LosAngeles, CA, USA;

‡ Department of Anesthesia, Divison of Pain, Duke University Medical Center,Durham, NC, USA

§ Department of Neurosurgery, University of Louisville, Louisville, KY, USA; and¶ Department of Neurosurgery, Robley Rex VA Medical Center, Louisville, KY, USA

For more information on author guidelines, an explanation of our peer reviewprocess, and conflict of interest informed consent policies, please go to http://www.wiley.com/bw/submit.asp?ref=1094-7159&site=1Financial support: None to discloseDisclaimer: The content of this paper does not represent the views of the Depart-ment of Veterans Affairs or the United States Government.

Neuromodulation: Technology at the Neural Interface

Received: January 2, 2013 Revised: February 14, 2013 Accepted: March 5, 2013

(onlinelibrary.wiley.com) DOI: 10.1111/ner.12059

428

www.neuromodulationjournal.com Neuromodulation 2013; 16: 428–435© 2013 International Neuromodulation Society

(7). Compared with those with commercial insurance, those withMedicaid are known to have lower rates of cancer screening, highermortality and longer inpatient stays following surgery, and are lesslikely to receive parenteral analgesics in the emergency department(ED) (8–10).

Given the known insurance disparities in surgical outcomes atlarge, along with the increasingly large socioeconomic role of SCS inthe treatment of chronic pain (11), we sought to examine the rela-tionship between a patient’s primary payer status and outcomesafter SCS surgery. Based on prior research that has demonstratedthat those with public insurance receive necessary operations at alower rate, we hypothesized that patients with Medicaid insurancewould have fewer reoperations after SCS surgery. Furthermore, wehypothesized that Medicaid insurance would be associated with anumber of poor postoperative health measurements such asnumber of complications, length of hospitalization, outpatienthealthcare utilization, and healthcare costs. In order to test thishypothesis, we designed an observational, national-level, longitudi-nal analysis examining the outcomes of patients with commercial orMedicaid insurance who underwent SCS surgery.

METHODS AND MATERIALSData Source

In order to compare patients with different insurance types at thenational level, we used the Commercial Claims and Encounters andMedicaid datasets of Thomson Reuter’s MarketScan® data base (12).The MarketScan data bases are constructed from paid claims fromover 100 unique payers across the United States and capturepatient-level data on inpatient and outpatient service usage for over70 million unique patients. The data base is maintained explicitly forresearch purposes, and all patient data are de-identified and com-pliant with the Health Insurance Privacy and Portability Act of 1996.

Inclusion and Exclusion CriteriaWe queried the inpatient admission tables of those datasets for all

hospitalizations in which a primary procedure of SCS implantationwas performed between January 1, 2000 and December 31, 2009.

In order to identify patients who had undergone SCS, wescreened the inpatient and outpatient tables for cases includingeither a percutaneous lead system (Current Procedural Terminology[CPT]-4 codes 63650 and 63685 concurrently) or a paddle leadsystem (CPT-4 codes: 63655 and 63685 concurrently). For eachpatient, the first occurring case satisfying the said conditions wasconsidered the index procedure. We only included patients 18 yearsor older at the time of the index procedure. If the end enrollmentdate for a patient was available, we used that to calculate follow-uptime. If no end enrollment date was available, we used December31, 2009 as the last date to calculate follow-up time.

Independent VariableThe independent variable for our analyses was insurance status

(commercial vs. Medicaid) during the period from 2000 to 2009.

Explanatory VariablesThe explanatory variables were measured at the date of the index

procedure. These included age, gender, year of index procedure,number of available preoperative follow-up days, number of avail-

able postoperative follow-up days, place of initial surgery (inpatientvs. outpatient placement), type of initial surgery (paddle vs. percu-taneous leads), comorbidities, International Classification of Dis-eases, Ninth Revision (ICD-9) diagnosis codes associated with theprocedure, and insurance type (commercial or Medicaid). Comor-bidities were measured with the Charlson index score computedusing Deyo’s adaptation to ICD-9-CM codes (13,14).

We summarized continuous variables such as age, preoperativefollow-up days, postoperative follow-up days, cumulative hospitaldays, and count variables such as cumulative services, cumulativeprescriptions using means and standard deviations. Categoricalvariables such as year of index procedure, Charlson index (catego-ries: 0, 1, 2, 3+), place of initial surgery, gender, procedure-associateddiagnosis code, presence of reoperation, and presence of complica-tions were summarized using counts and percentages.

Dependent VariablesThe outcome variables of interest were reoperation, complica-

tions, healthcare resource use (inpatient hospital days, total clinicvisits, total ED visits, and number of prescribed medications in theoutpatient setting), and healthcare resource costs.

Reoperations were measured by scanning all postoperative inpa-tient and outpatient encounters for a given patient for the presenceof an ICD-9-CM or CPT code corresponding to percutaneous lead orpaddle lead system placement. Complications were defined as anypostoperative renal, cardiac, neurological, deep vein thrombosis/pulmonary embolism, pulmonary, wound, or infection complica-tion. The incidence of complications was evaluated during the indexprocedure hospitalization and for all inpatient or outpatientencounters within 30 days and 90 days after surgery. Healthcareresource use and costs were totaled from any subsequent inpatientstays, outpatient clinic visits, total ED visits, and total recorded out-patient medication prescriptions. All costs were inflated to 2009 USdollars using the medical component of the consumer price index,which can be accessed through the website of the US Bureau ofLabor Statistics (15). Healthcare resources use and costs were evalu-ated at one and two years after the index surgery as well as at thelast available record of the patient. The comparison of the two insur-ance types was performed on cohorts of patients who had at leasttwo years postoperative data and on all patients.

Statistical AnalysisTo compare insurance groups, we used nonparametric rank sum

test for continuous variables and the chi-square test for categoricalvariables. Outcomes that showed a significant statistical differencefor the univariate comparison were then compared using multivari-ate analyses to adjust the difference. Multivariate models comprisedprocedure group (percutaneous or paddle leads), age, gender,Charlson index, and ICD-9 diagnosis code as covariates in additionto the insurance group (Medicaid vs. commercial). Cumulative daysof inpatient hospitalization were analyzed with log-linear models.Cumulative services and cumulative prescriptions were analyzedwith negative binomial models; costs were analyzed with linearmodels on log-transformed variables; reoperation and complicationwere analyzed with logistic regression. All tests were two sided andwere considered statistically significant if the p value was less than0.05. Because we examined a large number of outcome variables,and some could become significant just by random chance, weadjusted the p values to control for the false discovery rate in eachtable using the Benjamini and Hochberg’s approach (16,17). Weused the software SAS 9.3 (SAS Institute, Inc., Cary, NC, USA) for datapreparation and data analysis.

429DISPARITIES IN SPINAL STIMULATION


RESULTS

In total, 11,314 patients met inclusion criteria for this study(Fig. 1). Of those, the large majority were commercially insured(86.49%). Our two groups were similar in terms of age (p = 0.7781),but patients in the Medicaid group had longer average follow-uptimes (849 vs. 588 days, p < 0.0002), were more likely to be female(61.83% vs. 61.25%, p = 0.0017), and were more likely to undergopaddle lead placement as their index procedure (41.62% vs. 32.50%,p < 0.0002). Our Medicaid group had a higher baseline comorbidityburden, with 11.71% having a Charlson index of 2 or greater, com-pared with only 5.80% of the commercial insurance group (p <0.0002). Moreover, associated diagnosis codes differed significantlybetween our groups, with Medicaid patients more likely to have adiagnosis of postlaminectomy syndrome (ICD-9: 722.83, 39.59% vs.36.48%, p = 0.0299) and the commercial patients more likely to havea diagnosis of low back pain/syndrome (ICD-9: 724.2, 26.89% vs.21.73%, p < 0.0002) (Table 1).

When we analyzed these patients for short-term outcomes, wefound that there were no significant differences in complicationsbetween Medicaid and commercially insured patients during theindex hospitalization (0.36% vs. 0.26%, p = 0.5521) or during the30-day (1.37% vs. 1.64%, p = 0.4251) or 90-day (2.22% vs. 2.88%, p =0.1219) postoperative periods (Table 2). Commercially insured

patients had higher costs for their index procedure ($34,013 vs.$13,423, p < 0.0001).

In order to examine long-term differences between insurancegroups, we restricted our patient sample to include only those withat least two years of available follow-up data (and thus eliminatedpotential biases from differences in follow-up). This restricted oursample to 3456 patients and their baseline characteristics are pre-sented in Table 3. Differences between our two insurance groupspersisted in this subset of patients, with Medicaid patients continu-ing to have a higher proportion of females and higher comorbidityscores, as well as being more likely to receive paddle lead placementand more likely to have a diagnosis of postlaminectomy syndrome.

When we examined long-term outcomes, we found that at thetwo-year postoperative mark there were no significant differencesin reoperation rates (commercial vs. Medicaid = 5.06% vs. 7.32%, p =0.0513) (Table 4). In univariate analysis, Medicaid patients weremore likely to receive reoperations, and in particular paddle leadsystem placements (2.05% vs. 3.74%). However, these results werenot significant in multivariate analysis after correcting for baselinedifferences between the two groups (p = 0.0600). For the overallstudy period, we found that Medicaid patients were more likely toreceive reoperations with a paddle lead system (aOR: 1.593, 95% CI:1.161–2.186, p value 0.0055), although overall reoperation rates didnot significantly differ.

MarketScan Data N = nearly 150 million patients

Patients 18 years or older who underwentpercutaneous lead system or paddle lead system

N = 13,774

All patients (N = 11,314): patients 18 years or older whounderwent percutaneous lead system or paddle lead

system with commercial or Medicaid insurance

Patients with at least two years postoperativeenrollment (N = 3456): patients 18 years or older

who underwent percutaneous lead system or paddlelead system with commercial or Medicaid insurance

Drop Medicare N = 2460

Drop patients with less than two yearspostoperative continuous enrollment

N = 7858

Commercial N = 9786

MedicaidN = 1528

Commercial N = 2787

MedicaidN = 669

Figure 1. Study design and inclusion and exclusion criteria.

430

HUANG ET AL.


Despite the relative similarities in reoperation rates, we foundvastly different health resource utilization patterns. Both for thestudy period overall and at the two-year follow-up mark, SCSpatients with Medicaid had accrued significantly more inpatienthospital days (2-year est. ratio: 1.615, 95% CI: 1.312–1.989, p <0.0001), more trips to the ED (2-year est. ratio: 4.572, 95% CI:3.842–5.442, p < 0.0001), more clinic visits (2-year est. ratio: 1.422,95% CI 1.337–1.513, p < 0.0001), and more use of prescriptionmedications (2-year est. ratio: 1.319, 95% CI 1.168–1.489, p <0.0001), even after adjusting for baseline differences in age,gender, comorbidities, and initial procedure type. Similar to indexhospitalization costs, we found that overall two-year costs inthe Medicaid group were significantly lower ($64,644 vs. $110,908,p < 0.0001)

DISCUSSION

In accordance with the Affordable Care Act, there is expected tobe a large expansion in the number of patients enrolled in Medicaid,with an estimated 16 million new enrollees starting in 2014 (18).With the increasing number of Medicaid enrollees, it is important tounderstand the potential effects of insurance status on outcomesfollowing SCS, particularly given the many known interactionsbetween chronic pain management and socioeconomic status(19,20). As we find in this study, Medicaid patients undergoing SCSare much more likely to need more days of hospitalization, morevisits to the ED, and more prescription medications following theirsurgeries, even after adjusting for differences in diagnosis code,baseline comorbidities and demographics, and despite having

Table 1. Baseline characteristics, all patients.

Insurance typeTotal(N = 11314)

Commercial(N = 9786)

Medicaid(N = 1528)

p Value

CharacteristicsAge, mean (SD) 49.1 (10.0) 49.0 (9.6) 50.0 (12.3) 0.7781Preoperative follow-up days, mean (SD) 994 (828) 926 (790) 1432 (926) <0.0002Postoperative follow-up days, mean (SD) 623 (618) 588 (587) 849 (751) <0.0002Gender: female, N (%) 6996 (61.83) 5994 (61.25) 1002 (65.58) 0.0022Charlson index, N (%) <0.0002

0 10300 (91.04) 9005 (92.02) 1295 (84.75)1 268 (2.37) 214 (2.19) 54 (3.53)2 101 (0.89) 79 (0.81) 22 (1.44)3+ 645 (5.70) 488 (4.99) 157 (10.27)

Place of initial surgery, N (%) 0.1420Inpatient 1095 (9.68) 964 (9.85) 131 (8.57)Outpatient 10219 (90.32) 8822 (90.15) 1397 (91.43)

Type of initial surgery, N (%) <0.0002Percutaneous lead 7498 (66.27) 6606 (67.50) 892 (58.38)Paddle lead 3816 (33.73) 3180 (32.50) 636 (41.62)

Diagnoses, N (%)Postlaminectomy syndrome (722.83) 4175 (36.90) 3570 (36.48) 605 (39.59) 0.0299Thoracic or lumbosacral neuritis

or radiculitis, unspecified (724.4)3742 (33.07) 3217 (32.87) 525 (34.36) 0.2762

Lumbago (724.2) 2963 (26.19) 2631 (26.89) 332 (21.73) <0.0002None of the above 3468 (30.65) 3034 (31.00) 434 (28.40) 0.0554

Note: All p values are adjusted for the false discovery rate. Values significant at the p < 0.05 level are bolded.

Table 2. Short-term outcomes.



Medicaid(N = 1528)

P Value

Complications, N (%), aOR (95%CI)Immediate index hospitalization 39 (0.34) 35 (0.36) 4 (0.26) 0.552130 days 159 (1.41) 134 (1.37) 25 (1.64) 0.425190 days 261 (2.31) 217 (2.22) 44 (2.88) 0.1219

Index procedure, mean (SD), est. ratio (95%CI)Costs 31232 (28127) 34013 (25895)

Reference13423 (34602)0.194 (0.178–0.212)

<0.0001

Note: All p values are for univariate analysis and adjusted for the false discovery rate. Values significant at the p < 0.05 level are bolded.



equivalent rates of postoperative complications. Meanwhile, theyare not more likely to receive repeat operations. In total, we take thisto be evidence that Medicaid patients undergoing SCS may havepoorer clinical outcomes from SCS that necessitate greater health-care follow-up.

As mentioned previously, not all healthcare providers in theUnited States currently accept Medicaid patients. This may bebecause reimbursement for Medicaid patients and procedures per-formed thereon has traditionally been significantly lower than forcommercially insured patients. Conceivably, patients insured bycommercial carriers may be billed at a higher rate than Medicaidpatients, perhaps in attempt to make up for disparate reimburse-ment rates or to cover the cost of caring for the un- or under-insured. This billing difference may account for the cost differencesnoted in this study but unfortunately is only speculative as there iscurrently no way to accurately assess this from the utilized database.

Socioeconomic status and insurance disparities are well known toaffect outcomes in chronic pain management. For example, Fuenteset al. studied a series of 2070 patients presenting to a multidisci-plinary pain center for a variety of diagnoses and found that thosefrom lower socioeconomic status neighborhoods had more pain-related disability and worse pain characteristics even after adjustingfor racial and age differences (19). Likewise, socioeconomic disad-vantage has been associated with less effective coping mechanismswith pain (21) and also increased risk for pain in a number of differ-ent modalities (22–24). There are numerous proposed mechanismsfor these disparities, including differences in coping strategies, racialand ethnic differences, occupational factors, and educational dis-parities (20).

These factors likely all contribute in part to disparities in painoutcomes and likely contributed to the findings in the current studyas well. Moreover, there were undoubtedly many additional factors

associated with insurance disparities that affected outcomes follow-ing SCS but were not directly examined in this study. In particular, ithas been previously noted that patients of lower socioeconomicstatus have higher levels of health utilization overall (25). Thus, ourfindings likely represent one aspect of a broader trend in medicalcare, and are indicative of how many social and demographicfactors shape patients’ treatment and outcomes.

It is important to note that there are a number of limitations toour study. As was previously discussed, we were limited by the avail-able data in our ability to characterize individual patients’ clinicalpresentation. Thus we are unable to comment on the individualsurgical indications for any given patient and also unable tomeasure patient reported outcomes such as pain severity and itscorrelation with healthcare utilization. In addition, our data sufferfrom potential coding inaccuracies and reporting bias, but as thedata are derived from claims data these should be minimized.

Practitioners who implement SCS to treat failed back surgery syn-drome or chronic back and/or extremity pain may not uniformly be“specialists” in the field. This was a criticism of a study of SCS out-comes in a worker’s compensation setting by Turner et al. in whichSCS did not demonstrate greater effectiveness vs. alternative treat-ments after 6 months in this select population (26,27). In light of thispossibility, choice of percutaneous vs. paddle lead implantationmay be biased by operator comfort and skill and may potentiallyaffect outcomes. Paddle lead electrodes may be selected by practi-tioners who believe there is less chance of lead migration over timeand more opportunity to cover low back pain over percutaneousleads, but this is speculative. Unfortunately, the utilized data basedoes not identify the type of practitioner who performed the SCSprocedure. It is conceivable that Medicaid patients may be morelikely to be referred to and treated at tertiary centers where there aremore practitioners who may preferentially perform paddle lead

Table 3. Patients with at least two years of available postoperative data.



Medicaid(N = 669)

p Value

CharacteristicsAge, mean (SD) 49.7 (9.7) 49.5 (8.9) 50.3 (12.4) 0.4470Preoperative follow-up days, mean (SD) 901 (726) 839 (685) 1162 (718) <0.0002Postoperative follow-up days, mean (SD) 1374 (858) 1337 (563) 1527 (646) <0.0002Gender: female, N (%) 2150 (62.21) 1690 (60.64) 460 (68.76)Charlson index, N (%) <0.0002

0 3193 (92.39) 2634 (94.51) 559 (83.56)1 57 (1.65) 33 (1.18) 24 (3.59)2 28 (0.81) 18 (0.65) 10 (1.49)3+ 178 (5.15) 102 (3.66) 76 (11.36)

Place of initial surgery, N (%) 0.0096Inpatient 339 (9.81) 292 (10.48) 47 (7.03)Outpatient 3117 (90.19) 2495 (89.52) 622 (92.97)

Type of initial surgery, N (%) <0.0002Percutaneous lead 2465 (71.33) 2037 (73.09) 428 (63.98)Paddle lead 991 (28.67) 750 (26.91) 241 (36.02)

Diagnoses, N (%)Postlaminectomy syndrome (722.83) 1277 (36.95) 998 (35.81) 279 (41.70) 0.0080Thoracic or lumbosacral neuritis or radiculitis, unspecified (724.4) 1159 (33.54) 930 (33.37) 229 (34.23) 0.6719Lumbago (724.2) 876 (25.35) 729 (26.16) 147 (21.97) 0.0312None of the above 1059 (30.64) 884 (31.72) 175 (26.16) 0.0080

Note: All p values are adjusted for the false discovery rate. Values significant at the p < 0.05 level are bolded.

432

HUANG ET AL.


Table 4. Long-term outcomes.



Medicaid(N = 669)

p Value

Reoperation, N (%), aOR (95% CI)Two-year reoperation

All-type 190 (5.50) 141 (5.06) 49 (7.32) 0.0513*reference 1.432 (1.011–2.030)

Percutaneous lead system 116 (3.36) 90 (3.23) 26 (3.89) 0.4227Paddle lead system 82 (2.37) 57 (2.05) 25 (3.74) 0.0600*

reference 1.636 (0.996–2.687)Overall reoperation

All-type 536 (15.51) 417 (14.96) 119 (17.79) 0.0796Percutaneous lead system 348 (10.07) 282 (10.12) 66 (9.87) 0.8452Paddle lead system 220 (6.37) 160 (5.74) 60 (8.97) 0.0055

reference 1.593 (1.160–2.189)Postoperative healthcare resources use, mean (SD), est. ratio (95% CI)

Two-year inpatient useCumulative days 3 (10) 3 (9) 4 (13) <0.0001

reference 1.615 (1.312–1.989)Cumulative costs 14497 (38721) 14756 (39706) 13414 (34320) 0.7359

Overall inpatient useCumulative days 7 (18) 5 (12) 12 (31) <0.0001

reference 2.267 (1.910–2.690)Cumulative costs 25711 (56370) 24693 (53105) 29954 (68198) 0.0058

reference 1.883 (1.220–2.908)Two-year outpatient use

Cumulative services 187 (167) 172 (138) 254 (245) <0.0001reference 1.422 (1.337–1.513)

Cumulative costs 49760 (48871) 54743 (41965) 28998 (66879) <0.0001reference 0.392 (0.357–0.427)

Overall outpatient useCumulative services 341 (357) 294 (245) 535 (602) <0.0001

reference 1.761 (1.649–1.880)Cumulative costs 69815 (64953) 75433 (60045) 46412 (78123) <0.0001

reference 0.465 (0.425–0.509)Two-year outpatient ED use

Cumulative services 8 (35) 5 (14) 23 (74) <0.0001reference 4.572 (3.842–5.442)

Cumulative costs 1016 (4242) 1003 (4448) 1072 (3249) <0.0001reference 2.436 (1.825–3.252)

Overall outpatient ED useCumulative services 15 (52) 9 (22) 43 (106) <0.0001


reference 2.454 (1.830–3.292)Two-year medication use

Cumulative prescriptions 101 (92) 95 (82) 128 (121) <0.0001reference 1.319 (1.168–1.489)

Cumulative costs 11030 (14827) 11399 (15102) 9492 (13527) 0.0006reference 0.554 (0.399–0.768)

Overall medication useCumulative prescriptions 189 (197) 176 (172) 243 (271) <0.0001


reference 0.488 (0.348–0.685)Overall costs: index hospitalization and postoperative healthcare resources use, mean (SD), est. ratio (95% CI)

Two-year cumulative 101952 (90127) 110908 (80894) 64644 (113982) <0.0001reference 0.422 (0.390–0.457)

Overall cumulative 142983 (121473) 151866 (112361) 105978 (148202) <0.0001reference 0.490 (0.453–0.531)

Note: All p values are adjusted for the false discovery rate. Bolded values were significant in both univariate and multivariate analysis. Multivariate p value isdisplayed. The multivariate model adjusted for age, gender, Charlson index, type of index procedure (paddle vs. percutaneous), and procedure diagnosis code.*Values with a single asterisk were significant in univariate analysis but not in multivariate analysis (multivariate p value displayed).



implantation, possibly accounting for the differences notedbetween the groups in this study. Despite these limitations, ourstudy did have a number of strengths, including a large sample size(N = 11,314), relatively long length of follow-up (mean = 618 days),and ability to control for a number of important covariates.

We feel this study represents a crucial first step in investigatingsocioeconomic disparities in the outcomes following SCS and neu-romodulation at large. These outcome differences, and theobstacles to patients that they represent, are potentially of value topolicy makers, health services researchers, and clinicians seeking toimprove care for socially disadvantaged patients.

CONCLUSIONS

In a large, observational, national-level, longitudinal analysis, weexamined the outcomes of patients with commercial or Medicaidinsurance who underwent SCS surgery. We found no differences inthe 90-day complication rate or overall reoperation rate betweenMedicaid and commercially insured patients. However, Medicaidpatients had significantly greater postoperative healthcare utiliza-tion in the two years following surgery, even after adjusting forinitial differences in comorbidities, age, gender, ICD-9 diagnosiscode, and type of initial procedure. Further research is needed intopotential socioeconomic disparities in outcomes following SCSsurgery.

Authorship Statement

Mr. Huang prepared the manuscript with vital intellectual inputfrom Mr. Babu and Drs. Hazzard, Grossi, Huh, Roy, Patil, Boakye, andLad. Dr. Lad was responsible for initial project design. Dr. Ugiliiwen-eza was in change of statistical analysis. Drs. Lad and Boakye alsoprovided leadership and supervision over the project. All authorsapproved the final manuscript.

How to Cite this Article:Huang K.T., Hazzard M.A., Babu R., Ugiliweneza B., GrossiP.M., Huh B.K., Roy L.A., Patil C., Boakye M., Lad S.P. 2013.Insurance Disparities in the Outcomes of Spinal CordStimulation Surgery.Neuromodulation 2013; 16: 428–435

REFERENCES

1. Lee AW, Pilitsis JG. Spinal cord stimulation: indications and outcomes. NeurosurgFocus 2006;21:E3.

2. Health Management Associates estimate based on Congressional Budget Office.Budget and Economic Outlook, January 2008: Office of the Actuary National HealthStatistics Group, 2008; 2008.

3. Healthcare.gov. What is Medicaid? Accessed February 1, 2013.4. Kaiser Commision on Medicaid and the Uninsured. 2001 data. Accessed February 1,

2013.5. Kronick RH, Bella M, Gilmer TP, Somers SA. The Faces of Medicaid II: Recognizing the

Care Needs of People with Multiple Chronic Conditions October 2007.6. Smedley BD, Stith AY, Nelson AR. Institute of Medicine, Committee on Understand-

ing and Eliminating Racial and Ethnic Disparities in Health Care. Unequal treatment:confronting racial and ethnic disparities in health care. Washington, DC: NationalAcademies Press, 2003.

7. Center for Health Care Strategies, Inc. Practice Size Exploratory Project. Improvingthe quality and cost-effectiveness of publicly financed health care. 2008. AccessedFebrurary 1, 2013.

8. Rodriguez MA, Ward LM, Perez-Stable EJ. Breast and cervical cancer screening:impact of health insurance status, ethnicity, and nativity of Latinas. Ann Fam Med2005;3:235–241.

9. LaPar DJ, Bhamidipati CM, Mery CM et al. Primary payer status affects mortality formajor surgical operations. Ann Surg 2010;252:544–550; discussion 550–541.

10. Hostetler MA, Auinger P, Szilagyi PG. Parenteral analgesic and sedative use amongED patients in the United States: combined results from the National HospitalAmbulatory Medical Care Survey (NHAMCS) 1992-1997. Am J Emerg Med2002;20:83–87.

11. Lad SP, Kalanithi PS, Arrigo RT et al. A socioeconomic survey of spinal cord stimula-tion (SCS) surgery. Neuromodulation 2010;13:265–268; discussion 269.

12. Hansen LG, Chang S. Health research data for the real world: the thomson reutersmarketscan databases. The MarketScan Databases (White papers). January 2012.

13. Charlson M, Pompei P, Ales K, MacKenzie C. A new method of classifying prognosticcomorbidity in longitudinal studies: development and validation. J Chronic Dis1987;40:373–383.

14. Deyo R, Cherkin D, Ciol M. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–619.

15. Consumer Price Index (CPI). http://www.bls.gov/cpi/16. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and pow-

erful approach to multiple testing. Royal Statistical Society. 1995 (Series B57):289–300.

17. Osborne JA. Estimating the False Discovery rate using SAS and JMP. Paper pre-sented at: SAS Users Group 2006.

18. Letter to Nancy Pelosi on H.R. 4872, Reconciliation Act of 2010 (Final Health Care Leg-islation). Washington, DC: Congressional Budget Office, 2010.

19. Fuentes M, Hart-Johnson T, Green CR. The association among neighborhood socio-economic status, race and chronic pain in black and white older adults. J Natl MedAssoc 2007;99:1160–1169.

20. Poleshuck EL, Green CR. Socioeconomic disadvantage and pain. Pain 2008;136:235–238.

21. Roth RS, Geisser ME. Educational achievement and chronic pain disability: mediat-ing role of pain-related cognitions. Clin J Pain 2002;18:286–296.

22. Gran JT. The epidemiology of chronic generalized musculoskeletal pain. Best PractRes Clin Rheumatol 2003;17:547–561.

23. Heliovaara M, Makela M, Knekt P, Impivaara O, Aromaa A. Determinants of sciaticaand low-back pain. Spine (Phila Pa 1976) 1991;16:608–614.

24. Torrance N, Smith BH, Bennett MI, Lee AJ. The epidemiology of chronic pain ofpredominantly neuropathic origin. Results from a general population survey. J Pain2006;7:281–289.

25. Epstein AM, Stern RS, Weissman JS. Do the poor cost more? A multihospital study ofpatients’ socioeconomic status and use of hospital resources. N Engl J Med1990;322:1122–1128.

26. Wasan AD. Spinal cord stimulation in a worker’s compensation population: howdifficult it can be to interpret a clinical trial. Pain 2010;148:3–4.

27. Turner JA, Hollingworth W, Comstock BA, Deyo RA. Spinal cord stimulation for failedback surgery syndrome: outcomes in a worker’s compensation setting. Pain2010;148:14–25.

COMMENTS

Great and well written article. My request is further elaboration onwhy the authors noted a higher level of redo and paddle systemsplaced in the Medicaid population. Is there a correlation with redoand post-laminectomy patients and increase cost utilization in Medic-aid and commercial carriers? Thus the patients with more severechronic pain may require more care than those with less severe painconditions.

Dr. Philip Kim, MDBryn Mawr, PA, USA

***Although there is significant regional variation in the financial issuesfacing neuromodulation, this is a very interesting paper about anongoing important feature of our daily practice of neuromodulators.Thus I would recommend this to be the first of a group of similar papersreporting those economic aspects all around the world using thesharpness of detailed analysis used in this paper.

Ivano Dones, MDMilan, Italy

434

HUANG ET AL.


***This is a timely piece of work now that more patients from lowersocio-economic backgrounds will become eligible for Medicaid. Clini-cal outcome from SCS appears to be affected by many factors includ-ing the type of insurance carrier. The likely explanation being intrinsicto socio-economic class but could relate to the implanting physicianand equipment selected. It is encouraging that there is little differencein 30 day re-admission rates although at 2 year follow up there weresome differences in re-operation rates, in particular replacement ofpaddle leads, although it was not robust enough to be statisticallysignificant.

The other fascinating difference is the cumulative 2-year costs wherecommercially funded patients are nearly twice as costly as Medicaidfunded patients despite Medicaid patients having more re-admissionsand prescription usage. Again there are likely to be several reasons forthis, some of which the authors discuss.

This reviewer works in a unitary government funded healthcaresystem where more than 95% of SCS procedures are funded by andcarried out within this public system. It is of interest to see how aninequitable healthcare system as seen in USA can produce such varia-tion. Hopefully the UK’s planned national neuromodulation registrymay provide further insights in the future.

Many thanks to these authors for this paper causes us to ask manyimportant questions.

Simon Thomson, MBBSBasildon, Essex, United Kingdom

Comments not included in the Early View version of this paper.



Documents

Insurance Disparities in the Outcomes of Spinal Cord Stimulation Surgery