Impact of Celecoxib Restrictions in Medicare Beneficiaries with Arthritis

n managerial n

Impact of Celecoxib Restrictions in Medicare Beneficiaries With Arthritis

Anthony M. Louder, PhD, RPh; Ashish V. Joshi, PhD; Amy T. Ball, PharmD; Joseph C. Cappelleri, PhD; Michael C. Deminski, MS, RPh; and Robert J. Sanchez, PhD

onsteroidal anti-inflammatory drugs (NSAIDs) are widely used in the treatment of osteoarthritis (OA) and rheuma- Objective: To compare the incidence of serious toid arthritis (RA).1-3 The most common adverse events ex- gastrointestinal (GI) complications and associated N medical costs in a population with either osteoar- perienced by NSAID users are gastrointestinal (GI) related. Adverse thritis (OA) or rheumatoid arthritis (RA) enrolled GI events can range from very minor dyspepsia to life-threatening GI in Medicare plans with celecoxib formulary restrictions versus plans without such restrictions. 4,5 bleeding or perforation. The major risk factors for GI complications Methods: This study was a retrospective cohort related to NSAID use include history of GI bleed, concomitant use analysis of Medicare members in plans with and of anticoagulants, use of 2 or more NSAIDs, history of peptic ulcer, without celecoxib restrictions. Members diagnosed with OA or RA were identified and followed high-dose NSAID, age >60 years, severe illness, Helicobacter pylori in- for 1 year. fection, and concomitant© use Managed of corticosteroids. Care &6 Lanas et al indicated Results: The restricted group had higher levels of Healthcare Communications, LLC nonselective nonsteroidal anti-inflammatory drug that individuals over the age of 75 years are considered to be at high use (51% vs 40%; P <.001), and celecoxib use was risk and their risk equals that of persons with a history of peptic ulcer double in the unrestricted group (16% vs 8%; P disease.7 It has been estimated that more than 90% of all NSAIDs <.001). The incidence of a serious GI complication was slightly higher in the restricted group (5.4% prescribed are for individuals over the age of 65 years.6 It is well docu- vs 4.6%; P <.001). The adjusted mean serious GI mented that the elderly experience significant and serious GI compli- complication–related cost for the restricted group was more than 15 times higher than that for the 7,8 cations with NSAIDs. nonrestricted group ($1559 [95% confidence inter- Celecoxib, a selective cyclo-oxygenase (COX-2) inhibitor NSAID, has val (CI) $1341-$1811] vs $101 [95% CI $87-$117]); adjusted mean arthritis-related medical costs been shown to be as effective as traditional nonselective NSAIDs (nsN- were $5733 per year (95% CI $5097-$6448) for the SAIDs),9 although the product labeling for all NSAIDs carries the same restricted group and $3170 (95% CI $2816-$3569) for the unrestricted group. GI and cardiovascular warning. The majority of traditional nsNSAIDs Conclusions: The restricted group had significant- have generic equivalents available, while there are no COX-2 generic ly less use of celecoxib, indicating that restriction equivalents. Many health plans, in an effort to control costs, have imple- was effective at reducing celecoxib utilization. Although limitations exist when comparing mented utilization management techniques such as tier placement, prior populations from different health plans, and the authorization, or step therapy to decrease utilization of COX-2 agents.10-12 underlying causes of serious GI complications are multifactorial, the restricted group had a higher In 2004, Briesacher et al found that members in a 3-tier pharmacy benefit incidence of serious GI complications and higher design with arthritis and serious GI comorbidity were less likely to use costs related to serious GI complications and a COX-2 agent compared with those with a 1-tier pharmacy benefit.13 arthritis. (Am J Manag Care. 2011;17(7):503-512) Johnson et al evaluated the impact of NSAID formulary restriction on medical costs.14 They showed that the more restrictive formularies led to higher hospitalizations for OA and RA members and higher total healthcare costs for RA members.14 Although these approaches have the ability to lower pharmacy costs by decreasing utilization, very little has been published to show the impact of the restrictions on medical outcomes. Since 2007, some Medicare Advantage and Prescription Drug Plans have used various formulary management strategies to restrict the use of celecoxib. Humana health plans have used a step-therapy In this article policy with prior authorization Take-Away Points / p504 For author information and disclosures, see end of text. www.ajmc.com restriction. That is, in order to Full text and PDF get celecoxib covered through

VOL. 17, NO. 7 n THE AMERICAN JOURNAL OF MANAGED CARE n 503 n managerial n

Data Source Take-Away Points The restricted celecoxib group was This study compared the incidence of serious gastrointestinal (GI) complications and associated GI- and arthritis-related medical costs between Medicare plans with and without identified from Humana’s Medicare formulary restriction policies for celecoxib. population using medical, pharmacy, n Plans with a formulary restriction policy had a significantly higher incidence of serious and eligibility data. The unrestricted GI complications. n Plans with a formulary restriction policy had significantly higher serious GI-related medi- celecoxib group was identified using the cal costs. MarketScan Medicare Supplemental and n Plans with a formulary restriction policy had significantly higher arthritis-related medical Coordination of Benefits Database from costs. Thomson Reuters. Unrestricted status was based on participating in a plan that did not have a prior authorization/step their prescription benefit, Humana Medicare members must therapy requirement for celecoxib. Thomson Medstat associ- have tried and failed with 2 prescription-strength nsNSAIDs ates assigned the unrestricted status specific to each member on the preferred drug list in the previous 6 months or must for each calendar year. To assign this indication, members were satisfy at least 1 of the following secondary criteria: currently first grouped by drug plan and year. Details on celecoxib cov- on a prescription-strength GI medication, anticoagulant or erage by plan and year were determined using Fingertip For- antiplatelet therapy, bisphosphonate, chronic oral corticoste- mulary (Fingertip Formulary LLC, Glen Rock, New Jersey), roid, or antineoplastic agent. If neither of these criteria are Epocrates (Epocrates Inc, San Mateo, California), or the plans’ satisfied, the member’s provider could file a prior authorization Web sites. When information was not available, the propor- request. tion of celecoxib fills relative to all drug fills was used as a proxy The purpose of this study was to compare the incidence for restriction. Members enrolled in plans with <1.5% fills for of serious GI complications and associated medical costs in celecoxib were tagged as restricted. Only members enrolled in a an elderly Medicare population with OA and/or RA enrolled plan with an unrestricted status during their entire study period in plans with restrictions on celecoxib versus plans without were considered for inclusion. restrictions on celecoxib. Study Period The study period was January 1, 2006, to December 31, METHODS 2008. Members who met all of the following criteria were in- Study Design cluded in the study: medical claim with an ICD-9-CM diag- This retrospective study used a cohort design to identify nosis code in any of the first 3 diagnosis fields for OA (715.x) and track members with arthritis who were covered under a or RA (714.x or v82.1) between January 1, 2007, and Decem- Medicare Advantage or Medicare Supplemental plan. Two ber 31, 2007. The service date of the first claim was defined as groups were identified for comparison: (1) restricted celecoxib the index date. Those members who met the inclusion crite- coverage and (2) unrestricted celecoxib coverage. The study rion were observed for 12 months prior to their index date and design was approved by an institutional review board. 12 months after their index date.

Study Population Outcome Measures All members with coverage through a Medicare Advan- Pharmacy and medical claims with service dates that oc- tage or Medicare Supplemental health benefit plan who were curred during the individual subjects’ 24-month observation continuously enrolled during an observation period 365 days period were used to identify the outcomes of interest for the prior to and after the index date and who were between 65 and study. The composite primary end point of a serious GI com- 90 years of age on the index date were included in the study. plication was defined as a GI bleed or perforation (Table 1). Members who met either of the following 2 criteria were ex- Serious GI complications were further identified as upper or cluded from the study: (1) diagnosis of familial adenomatous lower in nature. The relevant codes were selected based on polyposis during the preindex or follow-up period (defined as published literature and expert opinion. The time until a seri- International Classification of Diseases, Ninth Revision, Clinical ous GI complication was calculated as the number of days be- Modification [ICD-9-CM] code 211.3 in any diagnosis field on tween the index date and service date on the medical claim. a medical claim) or (2) diagnosis of ankylosing spondylitis Serious GI complication–related total medical per mem- during the preindex or follow-up period (defined as ICD-9- ber per year (PMPY) costs during the follow-up period in- CM code 720.0 in any diagnosis field on a medical claim). cluded the total outpatient and inpatient medical costs where

504 n www.ajmc.com n JULY 2011 Impact of Celecoxib Restrictions

n Table 1. Diagnostic Codes for Serious Gastrointestinal Complications Description ICD-9-CM Code(s) Gastric ulcera 531.xx Duodenal ulcera 532.xx Peptic ulcera 533.xx Gastrojejunal ulcer 534.xx Gastritis and duodenitis with hemorrhagea 535.01, 535.51, 535.61 Other disorders of stomach and duodenum 537.83, 537.84 Lower gastrointestinal bleedingb 562.12, 562.13, 562.02, 562.03, 569.3, 569.8x Unspecified gastrointestinal bleeding 578.0, 578.1, 578.9 ICD-9-CM indicates International Classification of Diseases, Ninth Revision, Clinical Modification. aUpper serious gastrointestinal complication. bLower serious gastrointestinal complication.

n Table 2. Gastroprotective/Gastrotoxic Medications Therapeutic Category Generic Name Nonselective nonsteroidal anti-inflammatory druga Aspirin, choline magnesium, diclofenac, diflunisal, etodolac, fenoprofen, flurbiprofen, ibuprofen, indomethacin, ketoprofen, ketorolac, meclofenamate, meloxicam, nabumetone, naproxen, oxaprozin, piroxicam, salsalate, sulindac, tolmetin sodium Oral glucocorticosteroida Betamethasone, budesonide, cortisone, dexamethasone, hydrocortisone, methylprednisolone, paramethasone, prednisolone, prednisone, triamcinolone Oral bisphosphonatea Alendronate, etidronate, ibandronate, risedronate, tiludronate Antineoplastica Abarelix, anastrozole, bevacizumab, bicalutamide, carboplatin, carmustine, cisplatin, cladribine, clofarabine, cyclophosphamide, cytarabine, docetaxel, erlotinib, etoposide, irinotecan, letrozole, etc Anticoagulant usea Dicumarol, warfarin Antiplatelet therapya Abciximab, anagrelide, dipyridamole, aspirin/dipyridamole, cilostazol, clopido- grel, eptifibatide, ticlopidine, tirofiban Heparin/heparinoida Ardeparin, dalteparin, danaparoid, enoxaparin, fondaparinux, heparin, tinzaparin Proton pump inhibitorb Dexlansoprazole, esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole

b Histamine H2-receptor antagonist Cimetidine, famotidine, nizatidine, ranitidine, ranitidine bismuth citrate Defense mechanism stimulatorb Sucralfate Prostaglandin analogueb Misoprostol Cyclooxygenase-II inhibitorc Celecoxib Nonbiologic DMARD usec Azathioprine, auranofin, aurothioglucose, cyclosporine, gold sodium thiomalate, hydroxychloroquine, leflunomide, methotrexate, sulfasalazine Biologic DMARD usec Abatacept, adalimumab, anakinra, etanercept, infliximab, rituximab DMARD indicates disease-modifying antirheumatic drug. aGastrotoxic medication. Gastrotoxic medications are those medications identified in the step-therapy policy as being associated with increased risk for gastrointestinal complications. bGastroprotective medication. cArthritis therapy. the primary or secondary diagnosis code was for a serious GI out-of-pocket costs were included in the calculation of serious complication. Arthritis-related total medical PMPY costs in GI- and arthritis-related medical costs. Gastroprotective and the follow-up period included the total outpatient and in- gastrotoxic medication utilization was evaluated using the patient medical costs where the primary or secondary diag- pharmacy claims for the member. Arthritis treatment utiliza- nosis code was for RA or OA. Both plan-paid and member tion was also evaluated (Table 2). An RxRisk-V score, which

VOL. 17, NO. 7 n THE AMERICAN JOURNAL OF MANAGED CARE n 505 n managerial n

n Table 3. Sample Size Attrition Member Characteristics Restricted Unrestricted Identified with a rheumatoid arthritis and/or osteoarthritis diagnosis between January 1, 2007, 204,329 114,988 and December 31, 2007 Excluded for not being between 65 and 90 years old on the index date 38,318 6373 Excluded for not being continuously eligible 12 months before and after the index date 87,376 7862 Excluded for having a familial adenomatous polyposis diagnosis 11,674 15,389 Excluded for having an ankylosing spondylitis diagnosis 216 239 Number of members remaining in each of the final study groups 86,011 86,107

is an enhancement of the Chronic Disease Score,15,16 was cal- period had $1 added to their costs. Models provided the ad- culated using pharmacy utilization during the 12 months prior justed estimates of mean cost for each of the groups. to the index date. The unweighted version used in this analy- SAS version 9.2 (SAS Institute Inc, Cary, North Caro- sis was calculated using a simple count of drugs in unique drug lina) was used for all statistical analyses. categories. The score can range from 0 to 27.17-19

Statistical Analysis RESULTS Bivariate descriptive analyses were conducted on all study A total of 172,118 members met the inclusion criteria, variables. The t test or Wilcoxon rank-sum methods were uti- with 86,011 members in the restricted group and 86,107 in lized for comparing continuous variables, and c2 tests were uti- the unrestricted group. Table 3 shows the results of the appli- lized for the categorical variables. The baseline characteristics cation of the inclusion/exclusion criteria for the groups. Table (demographic, clinical, and drug utilization patterns) were 4 shows the baseline characteristics for the groups. The ma- compared between the celecoxib-restricted and celecoxib- jority of members in both groups were female, with approxi- unrestricted groups. Statistical significance was set at P <.05. mately 33% of both groups being male. Large differences were Cox proportional hazard models were used to compare the seen in the geographic regions between the restricted and unrestricted and unrestricted groups on the risk of serious GI restricted groups. The majority of members in the restricted complications. Time until a serious GI complication was the group were located in the South and Midwest regions (70% dependent (outcome) variable used in the model. The inde- and 23%, respectively). A more even distribution was seen pendent (predictor) variables used in the model were age, sex, in the unrestricted group between the Midwest, South, and region, OA/RA, use of medications or clinical conditions that West regions (29%, 37%, and 28%, respectively). The mean increase risk of GI complications, and use of gastroprotective age for each of the groups was similar (75 years), although medications. Members were censored if they did not have a found to be statistically significantly different (P <0.01) due to serious GI complication by the end of their follow-up period. the large sample sizes of the groups. The restricted and unre- Collinearity among the covariates was assessed, and based on stricted groups showed an equal distribution of members with the results there were no significant correlations between the a diagnosis of OA (88%), RA (5.3%), and OA with RA (6%). covariates. Time-dependent covariates were included in the The unrestricted group had more members with a prior history models to assess the proportional hazards assumption. of a GI bleed (4.4% vs 2.1%). There were slightly more mem- Generalized linear models (GLMs) using a gamma distribu- bers with a history of H pylori infection in the restricted group tion as their probability distribution and log-link as their link (0.6% vs 0.4%). The unrestricted group had a slightly higher function were used to compare the medical costs of serious GI RxRisk-V score (5.9 vs 3.8). complications and arthritis for the groups. The usefulness of With respect to the utilization of gastroprotective medica- the GLMs for modeling the medical costs is their ability to tions, more members in the unrestricted group had a claim(s) handle nonconstant variance (heteroscedasticity) and avoid for a proton pump inhibitor (PPI) (37% vs 32%; P <.01), having to retransform log-transformed costs (a common strat- whereas more members in the restricted group had a claim(s) 19 egy for handling highly skewed medical cost data). The co- for a histamine H2-receptor antagonist (H2RA) (13% vs 8%; variates used in the models were the same as those used in the P <.01). Compared with the unrestricted group, the restricted proportional hazard model. Members with zero costs (related group had more claims for drugs recognized as gastrotoxic: to serious GI complications or arthritis) during the follow-up nsNSAIDs (51% vs 40%) and oral glucocorticosteroids (33%

506 n www.ajmc.com n JULY 2011 Impact of Celecoxib Restrictions

n Table 4. Baseline Demographicsa Restricted (n = 86,011) Unrestricted (n = 86,107) Characteristic No. % No. % Sexb Male 28,237 32.8 28,817 33.5 Geographic regionb Northeast 946 1. 1 5235 6.1 Midwest 20,089 23.4 24,855 28.9 South 59,863 69.6 31,705 36.8 West 5113 5.9 24,312 28.2 Age, y, mean (SD)b 75.4 (6.4) 76.3 (6.2) OA diagnosis 76,332 88.7 76,206 88.5 RA diagnosis 4594 5.3 4530 5.3 OA and RA diagnosisb 5085 5.9 5371 6.2 Peptic ulcer disease diagnosis 1806 2.1 1775 2.1 Helicobacter pylori infection diagnosisb 477 0.6 305 0.4 History of GI bleedb 1827 2.1 3800 4.4 RxRisk-V score, mean (SD)b 3.8 (2.1) 5.9 (2.7) Proton pump inhibitor useb 27,796 32.3 32,368 37.6

b Histamine H2-receptor antagonist use 11,898 13.8 7181 8.3 Misoprostol use 301 0.3 303 0.4 Sucralfate use 1508 1. 8 1626 1. 9 Nonselective nonsteroidal anti-inflammatory drug useb 43,899 51.0 34,715 40.3 Oral glucocorticosteroid useb 28,201 32.8 25,181 29.2 Oral bisphosphonate useb 15,387 1 7. 9 18,687 21.7 Antineoplastic useb 3955 4.6 3771 4.4 Antiplatelet useb 10,159 11. 8 10,964 12.7 Anticoagulant useb 10,871 12.6 14,644 1 7. 0 Heparin/heparinoid usec 4287 5.0 4561 5.3 Celecoxib useb 6593 7. 7 13,458 15.6 Nonbiologic DMARD useb 4122 4.8 7025 8.2 Biologic DMARD useb 375 0.4 1332 1. 5 DMARD indicates disease-modifying antirheumatic drug; GI, gastrointestinal; OA, osteoarthritis; RA, rheumatoid arthritis. aValues are numbers and percentages except where indicated. bP <.01. cP <.05. vs 29%). More utilization of bisphosphonates, antiplatelets, unrestricted group (5.4% vs 4.6%; P <.01 [Table 5]). This anticoagulants, and heparin was seen in the unrestricted translates into a higher relative risk (1.17 [95% confidence group. As expected, fewer members used celecoxib in the re- interval (CI) 1.12-1.21]) in the restricted group, although stricted group compared with the unrestricted group (8% vs the absolute difference in risk was small (0.8%). While we 16%). The unrestricted group also utilized more nonbiologic agree that the clinical relevance of this finding is debatable and biologic disease-modifying antirheumatic drugs. due to the small difference, the difference was statistically The members in the restricted group had a slightly high- significant, and given the prevalence of arthritis in the Medi- er risk for a serious GI complication compared with the care population, it could translate into significant outcome

VOL. 17, NO. 7 n THE AMERICAN JOURNAL OF MANAGED CARE n 507 n managerial n

n Table 5. Serious Gastrointestinal Complicationsa Variable Restricted (n = 86,011) Unrestricted (n = 86,107) Cumulative incidence of serious GI complicationsb 5.4% 4.6% Time until serious GI complication, days, mean (SD) 167.3 (112.0) 167.3 (108.0) Cumulative incidence of serious GI complications classified as upperb 1.0% 0.7% Cumulative incidence of serious GI complications classified as lower 1.5% 1.5% GI indicates gastrointestinal. aRelative risk (95% confidence interval) for the restricted group was 1.17 (1.12-1.21); P <.01. bP <.01. and cost differences between the 2 groups, as examined in unrestricted groups, respectively. The unadjusted arthritis- subsequent sections. While no differences existed in lower related medical PMPY costs were higher for the restricted GI complications between the groups, the incidence of upper group ($1436 vs $764; P <.0001). Table 8 presents the re- GI complications was higher in the restricted group (1.0% sults from the generalized linear regression models for seri- vs 0.7%; P <.01). Both groups were similar with respect to ous GI complication–related and arthritis-related medical time until a serious GI complication occurred. The average costs after controlling for age, sex, geographic region, un- number of days until a serious GI event occurred was 167 derlying clinical conditions, and drug utilization. The ad- days from the index date. justed mean PMPY costs for serious GI complications were Cox proportional hazard models were also used to predict $1559 vs $101 (P <.0001) for the restricted and unrestrict- the risk of serious GI complications. After controlling for de- ed groups, respectively, across the entire arthritis patient mographic and clinical characteristics and medications that population. A second GLM model was used to compare the impact the risk for a serious GI complication, the restricted serious GI complication costs between the restricted and group still showed a higher risk for a serious GI complication unrestricted groups for the subset of arthritis members with (hazard ratio 1.69; 95% CI 1.60-1.78), confirming the results of a serious GI complication only. This model showed that the unadjusted results. Covariates associated with an increased among those with a serious GI complication, the adjusted risk were age, prior GI bleed, H pylori diagnosis, RxRisk score, serious GI complication PMPY costs were 4 times higher in and sucralfate use. Those associated with a decreased risk were the restricted group compared with the unrestricted group OA (vs OA and RA combined), and PPI, H2RA, glucocor- ($4025 vs $1028, P <.0001). The model for arthritis-related ticosteroid, oral bisphosphonate, antineoplastic, antiplatelet, medical PMPY costs showed that the adjusted costs for the anticoagulant, heparin, and celecoxib use (Table 6). A sen- restricted group were 80% higher than those for the unre- sitivity analysis was performed by excluding members with a stricted group. prior GI bleed. The results from the proportional hazards were similar (hazard ratio 1.63; 95% CI 1.54-1.73). A second Cox proportional hazards model evaluating any DISCUSSION gastroprotective therapy use and level of exposure was applied. This study compared the incidence of serious GI com- Gastroprotective therapy use was defined as a member using a plications and disease-related expenditures in 2 groups of PPI, H2RA, sucralfate, or misoprostol between the index date members covered under a Medicare Advantage or Medicare and the end of their follow-up period. Gastroprotective thera- Supplemental plan who were diagnosed with OA or RA. The py use was higher in the unrestricted group than the restricted groups differed in terms of whether or not the health plan that group (36% vs 28%; P <.0001). The level of exposure was the administered the drug coverage benefit had a restriction (ie, number of claims made by the member during the period. The step therapy edit) in place for celecoxib. Overall the groups results of the Cox regression model indicated that users of any were similar in terms of age and sex. However, large differ- versus no gastroprotective therapy (dichotomous predictor) ences were seen in the geographic regions represented in each had more than 2.2 times the risk for a serious GI complication of the groups. The analyses indicated that region did not have of nonusers (Table 7). However, the increase in the number an impact on serious GI event rates, but there was an impact of gastroprotective therapy refills was associated with a 20% on the costs associated with treating serious GI complications decrease in the risk for a serious GI complication. and arthritis. This could be due to variations in treatment pat- The unadjusted mean serious GI complication medi- terns and reimbursement rates between regions. Both groups cal costs PMPY were $146 and $29 for the restricted and were also similar with respect to the prevalence of OA, RA, or

508 n www.ajmc.com n JULY 2011 Impact of Celecoxib Restrictions

n Table 6. Cox Proportional Hazard Results for All Serious Gastrointestinal Complicationsa 95% Hazard Ratio Variable Hazard Ratio Confidence Limits P Restricted (vs unrestricted) 1.689 1.604, 1.779 <.0001 Age 1. 015 1.011, 1.018 <.0001 Male (vs female) 1.032 0.986, 1.081 .1752 Midwest (vs Northeast) 0.926 0.821, 1.045 .2150 South (vs Northeast) 1.004 0.892, 1.130 .9514 West (vs Northeast) 0.872 0.769, 0.988 .0319 OA (vs OA and RA) 0.673 0.616, 0.735 <.0001 RA (vs OA and RA) 0.878 0.779, 0.988 .0315 Prior GI bleed 3.887 3.635, 4.157 <.0001 Helicobacter pylori diagnosis 1.956 1.572, 2.434 <.0001 RxRisk 1.172 1.161, 1.183 <.0001 Proton pump inhibitor use 0.849 0.807, 0.893 <.0001

Histamine H2-receptor antagonist use 0.779 0.716, 0.848 <.0001 Misoprostol use 1.167 0.752, 1.812 .4898 Sucralfate use 1.518 1.287, 1.791 <.0001 Nonselective nonsteroidal anti-inflammatory drug use 0.747 0.712, 0.785 <.0001 Oral glucocorticosteroid use 0.652 0.613, 0.693 <.0001 Oral bisphosphonate use 0.722 0.674, 0.773 <.0001 Antineoplastic use 0.823 0.730, 0.929 .0016 Antiplatelet use 0.781 0.727, 0.838 <.0001 Anticoagulant use 0.759 0.709, 0.812 <.0001 Heparin use 0.753 0.657, 0.864 <.0001 Celecoxib use 0.731 0.668, 0.799 <.0001 Nonbiologic DMARD use 0.708 0.630, 0.796 <.0001 Biologic DMARD use 0.703 0.529, 0.932 .0145 DMARD indicates disease-modifying antirheumatic drug; GI, gastrointestinal; OA, osteoarthritis; RA, rheumatoid arthritis. aTime-dependent covariates were used to test the proportional hazard assumption. None of the time-dependent covariates were significant, indicating that the proportional hazard assumption was met. both conditions. As would be expected, the level of celecoxib about factors influencing prescriber behavior. Future studies utilization was much lower in the restricted group. could focus on the outcomes observed in those individuals Several other studies in Medicaid populations have dem- who are directly impacted by a formulary restriction policy to onstrated that prior authorization programs have a significant determine what alternative therapies, if any, are used if the impact on reducing the utilization of COX-2 inhibitors with decision is made to avoid or bypass the step therapy or prior an expected increase in nsNSAID utilization.10,20,21 However, authorization process. to our knowledge, no study has examined the association be- Another key finding that may have an impact on the inci- tween restrictions and serious GI complications in a Medicare dence of GI complications was the difference in utilization of population. This association is important because for most PPIs. The unrestricted group had higher levels of PPI use. It health plans, age greater than 65 years is one of the risk fac- was confirmed that the restricted plan also has some formulary tors indicating that the restriction should not be enforced (ie, restrictions on PPIs, which might explain some of this differ- members should be able to receive celecoxib). In spite of hav- ence. Other studies have shown that PPIs are used more of- ing this risk factor, only 8% of the restricted group received ten with COX-2 inhibitors than nsNSAIDs.22 Although this celecoxib, which is cause for concern and raises questions relationship was not directly analyzed in this study, it could

VOL. 17, NO. 7 n THE AMERICAN JOURNAL OF MANAGED CARE n 509 n managerial n

n Table 7. Cox Proportional Hazard Results for All Serious Gastrointestinal Complications (Osteoarthritis and Rheumatoid Arthritis) 95% Confidence Limits Variable Hazard Ratio for Hazard Ratio P Restricted (vs Unrestricted) 1.824 1.731, 1.922 <.0001 Age 1. 015 1.012, 1.019 <.0001 Male (vs female) 1.024 0.978, 1.072 .3095 Midwest (vs Northeast) 0.926 0.820, 1.045 .2114 South (vs Northeast) 1. 014 0.901, 1.142 .8169 West (vs Northeast) 0.841 0.742, 0.954 .0070 OA (vs OA and RA) 0.673 0.616, 0.735 <.0001 RA (vs OA and RA) 0.881 0.782, 0.992 .0363 Prior GI bleed 3.972 3.715, 4.247 <.0001 Helicobacter pylori diagnosis 1.773 1.426, 2.206 <.0001 RxRisk 1.189 1.179, 1.200 <.0001 Gastroprotective agent use 2.219 2.078, 2.369 <.0001 Gastroprotective agent claim count 0.795 0.783, 0.806 <.0001 Nonselective nonsteroidal anti-inflammatory drug use 0.744 0.708, 0.781 <.0001 Oral glucocorticosteroid use 0.637 0.599, 0.677 <.0001 Oral bisphosphonate use 0.727 0.679, 0.778 <.0001 Antineoplastic use 0.824 0.730, 0.930 .0017 Antiplatelet use 0.779 0.725, 0.836 <.0001 Anticoagulant use 0.742 0.693, 0.795 <.0001 Heparin use 0.738 0.644, 0.846 <.0001 Celecoxib use 0.753 0.689, 0.824 <.0001 Nonbiologic DMARD use 0.705 0.627, 0.793 <.0001 Biologic DMARD use 0.708 0.534, 0.940 .0170 DMARD indicates disease-modifying antirheumatic drug; GI, gastrointestinal; OA, osteoarthritis; RA, rheumatoid arthritis.

n Table 8. Results of Regression Models Predicting Medical Costs Related to Serious Gastrointestinal Complica- tions and Arthritis Variable Restricted Unrestricted P Serious GI complication medical costs Adjusted annual cost per member, mean (95% CI) $1559 ($1341-$1811) $101 ($87-$117) .0001 Serious GI complication medical costs in those with complication Adjusted annual cost per member, mean (95% CI) $4025 ($2202-$7358) $1028 ($584-$1808) .0001 Arthritis-related medical costs Adjusted annual cost per member, mean (95% CI) $5733 ($5097-$6448) $3170 ($2816-$3569) .0001 CI indicates confidence interval; GI, gastrointestinal. be an underlying reason why PPI use was higher in the unre- tiveness of these agents could be an underlying factor for the stricted group. Interestingly, the unrestricted plan members lower arthritis-related medical costs in the unrestricted group. also utilized more of nonbiologic and biologic disease-modifying antirheumatic drugs, even though the incidence of RA in Limitations both groups was similar. The increased use and resulting effec- First, the study was observational, so it was unable to

510 n www.ajmc.com n JULY 2011 Impact of Celecoxib Restrictions identify a causal relationship between formulary restric- that they are employed by Pfizer, the maker of Celebrex. Dr Joshi and Mr De- minski also report holding stock in the company. Dr Ball reports no relation- tion policy and serious GI complications and treatment ship or financial interest with any entity that would pose a conflict of interest costs. Second, cumulative medication exposure was not ac- with the subject matter of this article. counted for when evaluating gastrotoxic/gastroprotective Authorship Information: Concept and design (AML, AVJ, ATB, JCC, medication exposure. There could be significant differences MCD, RJS); acquisition of data (AML, ATB); analysis and interpretation of data (AML, AVJ, ATB, JCC, MCD, RJS); drafting of the manuscript (AML, between the therapeutic categories included in the analy- JCC, RJS); critical revision of the manuscript for important intellectual con- sis in terms of the cumulative amount of exposure required tent (AVJ, ATB, JCC, MCD, RJS); statistical analysis (AML, JCC); obtaining funding (AVJ, ATB, RJS); administrative, technical, or logistic support (AVJ); to impact the risk for a GI complication. For example, 30 and supervision (AVJ). days of continuous NSAID therapy may put a member at Address correspondence to: Anthony M. Louder, PhD, RPh, Competitive higher risk than 30 days of continuous antiplatelet therapy; Health Analytics, 325 West Main St, WFP6W, Louisville, KY 40202. E-mail: [email protected]. however, based on presence of claims they are given equal weighting. Third, another clinical limitation is that other GI bleed risk factors were unaccounted for (eg, residing in a REFERENCES skilled nursing facility, where the likelihood of being seden- 1. Recommendations for the medical management of osteoarthritis tary may be higher and thus put the member at a higher risk of the hip and knee: 2000 update. American College of Rheumatology Subcommittee on Osteoarthritis Guidelines. Arthritis Rheum. 2000; for a serious GI complication). Fourth, the study was unable 43(9):1905-1915. to account for confounding by health plan benefit structure. 2. Hochberg MC, Altman RD, Brandt KD, et al. Guidelines for the medical management of osteoarthritis, P I: osteoarthritis of the hip. Ameri- The restricted group consisted of members enrolled in a can College of Rheumatology. Arthritis Rheum. 1995;38(11):1535-1540. Medicare Advantage plan, whereas the nonrestricted group 3. Bérard A, Solomon DH, Avorn J. Patterns of drug use in rheumatoid consisted of members enrolled in a Medicare Supplemen- arthritis. J Rheumatol. 2000;27(7):1648-1655. tal plan. The types of individuals enrolled in these types 4. Singh G. Recent considerations in nonsteroidal anti-inflammatory drug gastropathy. Am J Med. 1998;105(1B):31S-38S. of plans could differ in terms of benefit structure. Lastly, 5. Lanas A, Serrano P, Bajador, Esteva F, Benito R, Sáinz R. Evidence members enrolled in a capitated (at-risk) benefit were not of aspirin use in both upper and lower gastrointestinal perforation. excluded. Providers commonly submit encounter claims Gastroenterology. 1997;112(3):683-689. 6. Laine L, Connors LG, Reicin A. Serious lower gastrointestinal clini- only for their capitated members. The lack of reliable pay- cal events with nonselective NSAID or coxib use. Gastroenterology. ment information for these members may have affected the 2003;124(2):288-292. cost analysis. 7. Lanas A, Ferrandez A. Inappropriate prevention of NSAID-induced gastrointestinal events among long-term users in the elderly. Drugs Aging. 2007;24(2):121-131. 8. Griffin MR, Ray WA, Schaffner W. Nonsteroidal anti-inflammatory CONCLUSION drug use and death from peptic ulcer in elderly persons. Ann Intern Med. 1988;109(5):359-363. We found that that the Medicare population that be- 9. Smalley WE, Ray WA, Daughtery JR, Griffin MR. Nonsteroidal anti- longed to plans with celecoxib formulary restriction poli- inflammatory drugs and the incidence of hospitalizations for peptic ulcer disease in elderly persons. Am J Epidemiol. 1995;141(6):539-545. cies in place had significantly lower celecoxib utilization, 10. Cannon GW, Breedveld FC. Efficacy of cyclooxygenase-2-specific indicating that the restriction was effective at reducing uti- inhibitors. Am J Med. 2001;110(suppl 3A):6S-12S. lization. However, after statistically controlling for underly- 11. Fischer MA, Schneeweiss S, Avorn J, Solomon DH. Medicaid prior- authorization programs and the use of cyclooxygenase-2 inhibitors. N ing clinical risk factors, being in the restricted group was Engl J Med. 2004;351(21):2187-2194. associated with a higher risk for experiencing a serious GI 12. Smalley WE, Griffin MR, Fought RL, Sullivan L, Ray WA. Effect of a complication during the 1-year follow-up period. Being in prior-authorization requirement on the use of nonsteroidal anti-inflammatory drugs by Medicaid patients. N Engl J Med. 1995;332(24): the restricted group also was associated with higher serious 1612-1617. GI-related and arthritis-related medical costs. Our findings 13. Briesacher B, Kamal-Bahl S, Hochberg M, Orwig D, Kahler KH. Three-tiered copayment drug coverage and use of nonsteroidal anti- underscore the importance of evaluating formulary restric- inflammatory drugs. Arch Intern Med. 2004;164(15):1679-1684. tion policies in terms of both the costs related to adverse 14. Johnson TJ, Stahl-Moncada S. Medicaid prescription formulary re- clinical outcomes and the immediate intended benefits in strictions and arthritis treatment costs. Am J Public Health. 2008;98(7); 1300-1305. terms of drug cost savings produced. 15. Lanas A, Garcia-Rodriguez LA, Polo-Tomás M, et al. Time trends and impact of upper and lower gastrointestinal bleeding and perforation in clinical practice. Am J Gastroenterol. 2009;104(7):1633-1641. Author Affiliations: From Competitive Health Analytics, Inc (AML, ATB) Louisville, KY; Pfizer Inc (AVJ, JCC, RJS, MCD), New York, NY. 16. Sloan KL, Sales AE, Liu CF, et al. Construction and characteristics of the RxRisk-V: a VA-adapted pharmacy-based case mix instrument. Med Funding Source: This study was funded by Pfizer, Inc. Care. 2003;41(6):761-774. Author Disclosures: Dr Louder reports employment and stock ownership 17. Sales AE, Liu CF, Sloan KL, et al. Predicting costs of care using a with Humana, which has a utilization management restriction policy in place pharmacy-based measure risk adjustment in a veteran population. for celecoxib. Dr Joshi, Dr Cappelleri, Mr Deminski, and Dr Sanchez report Med Care. 2003;41(6):753-760.

VOL. 17, NO. 7 n THE AMERICAN JOURNAL OF MANAGED CARE n 511 n managerial n

18. Farley JF, Harley CR, Devine JW. A comparison of comorbidity and prescription drug use in a managed care Medicaid population. measurements to predict healthcare expenditures. Am J Manag Care. Clin Ther. 2004;26(9):1518-1532. 2006;12(2):110-119. 21. Carroll NV, Smith JC, Berringer RA, Oestreich GL. Evaluation of an 19. Blough DK, Ramsey SD. Using generalized linear models to assess automated system for prior authorization: a COX-2 inhibitor example. medical care costs. Health Serv Outcomes Res Methodol. 2000;1(2): Am J Manag Care. 2006;12(9):501-508. 185-202. 22. Johnell K, Fastbom J. Concomitant use of gastroprotective drugs 20. Hartung DM, Touchette DR, Ketchum KL, Haxby DG, Goldberg BW. among elderly NSAID/COX-2 selective inhibitor users: a nationwide Effects of a prior-authorization policy for celecoxib on medical service register-based study. Clin Drug Investig. 2008;28(11):687-695. n

512 n www.ajmc.com n JULY 2011