Pharmacoeconomic Evaluation of Gastrointestinal Tract Events During Treatment With Risedronate or Alendronate: A Retrospective Cohort Study

August 1, 2004
Sunanda Kane, MD, MSPH

Natalie N. Borisov, PhD

Diana Brixner, RPh, PhD

Supplements and Featured Publications, Gastrointestinal Events and Medical Costs Associated With Bisphosphonate Therapy for Osteoporosis, Volume 10, Issue 7

Objective: To assess the cost of treating gastrointestinal(GI) tract events associated with alendronateand risedronate treatments.

Methods: GI-related direct medical costs associatedwith alendronate or risedronate use wereobtained from a large administrative medical andpharmaceutical claims database for the period fromNovember 1, 2000, to February 28, 2002. Direct GI-relatedcosts were evaluated in the initial 4 monthsof newly started bisphosphonate therapy in patientsaged 65 years and older with no history of GI eventsin the 6-month pretreatment period.

Results: A total of 4259 women and men wereincluded in the analysis. There were no differencesbetween treatment groups (alendronate, n = 3636;risedronate, n = 623) in age, gender, or duration ofbisphosphonate treatment. Fewer patients treatedwith risedronate (21 of 623; 3.4%) experienced GItract events compared with patients treated with alendronate(196 of 3636; 5.4%) during the first 4 monthsof bisphosphonate treatment. Both inpatient and outpatientvisits were less common in the risedronategroup than in the alendronate group (inpatient, 0.8%vs 1.4%; outpatient, 2.6% vs 4.0%, respectively). Theestimated average annual GI-related cost for patientstreated with alendronate was approximately $72 000per 1000 patients compared with $26 000 per 1000patients treated with risedronate.

Conclusion: In a patient population 65 years orolder initiating bisphosphonate therapy for osteoporosis,risedronate was associated with markedlylower GI-related diagnoses, medical resource utilization,and direct medical costs compared withalendronate. The substantially higher cost of managingand treating alendronate-related GI eventsshould be considered when choosing among bisphosphonatetherapies.

(Am J Manag Care. 2004;10:S216-S226)

Alendronate and risedronate are the 2most commonly prescribed oral bisphosphonatesfor the treatment ofpostmenopausal and glucocorticoid-inducedosteoporosis. Both agents have demonstratedgastrointestinal (GI) tract adverse eventprofiles similar to placebo in controlled trials.1,2 However, clinical trials for alendronateexcluded patients with major upper GI tractdisease (eg, peptic ulcers or dyspepsia),whereas trials for risedronate did not.3-5

Although the GI tract adverse event profilesof alendronate and risedronate havebeen well documented in the clinical literature,there is a relative dearth of data on theburden of these events from a payer perspective.There have been no head-to-headstudies designed to evaluate differences inthe cost of GI tract adverse events associatedwith alendronate and risedronate. Oneobservational study reported significantlyhigher GI-related medical costs in patientstreated with alendronate than in population-basedcontrols.6 A better understanding ofthe GI-related costs with different bisphosphonatesis important to managed careorganizations (MCOs) and other healthcareproviders for making optimal use of theirresources.

The purpose of this study was to estimateand compare medical costs of GI-relatedevents associated with the use of alendronateand risedronate in patients enrolledin the Protocare Sciences ProprietaryManaged Care Database. Direct medicalcosts of GI-related events were evaluated inthe 4-month treatment period following theinitiation of bisphosphonate therapy forosteoporosis in a sample of patients aged 65years and older who had newly started bisphosphonate therapy and had no reportedGI events in a 6-month pretreatment period.


Data. A retrospective cohort study wasconducted using Protocare Sciences integratedadministrative medical and pharmaceuticalclaims data. The Protocare Sciencesproprietary database is composed of claimsand eligibility records for more than 10 millionlives covered under various public(Medicaid) and private benefit plans since1991. The private benefit plan informationincludes data on medical services and pharmacyclaims provided through commercialhealth maintenance organizations (HMOs),preferred provider organizations (PPOs),Medicare plans, and other indemnity products.Patients were tracked longitudinally,and approximately 76% of the members inthe private benefit plan database belongedto commercial HMOs or PPOs. The remainingmembership was enrolled in Medicareplans (Medicare Risk and Medicare Supplementproducts) and a variety of specialtyand administrative services—only products,including group life, dental, disabilityincome, workers' compensation, and pharmacymanagement services.

Study Population. A dataset of administrative,medical, and pharmaceutical recordswas obtained between June 1, 2000, andJune 30, 2002, for patients aged 65 years orolder. Patients were selected with a new prescriptionfor risedronate 5 mg daily or alendronate5 or 10 mg daily (35 or 70 mgweekly) between November 1, 2000, andFebruary 28, 2002 (a 16-month captureperiod). The date of the first alendronate orrisedronate prescription for each patientwas defined as the index date. Therefore, allpatients included in the cohort had 6months of observation before their indexdate (pretreatment period) and 4 monthsfollowing initiation of the therapy (treatmentperiod). A 4-month follow-up periodwas considered to be clinically sufficient todetect effects related to treatment, becauseGI adverse events typically occur early inbisphosphonate dosing.7

International Classification of Diseases,Ninth Revision, Clinical Modification (ICD-9-CM)

CurrentProcedural Terminology (CPT-4)

Patients who had a prescription for a bisphosphonateand/or any GI event and/or GI-relatedmedications during the 6-month pretreatmentperiod were excluded from thecohort to allow for an analysis of a "clean"GI-related history population. GI eventswere defined using GI-related primary diagnosis and/or primary codes(Appendices 1 and 2). GI medicationsincluded prescriptions for H2 antagonists,proton pump inhibitors (PPIs), and cytoprotectives(Appendix 3).


Patients who had at least 1 prescriptionfor risedronate 30 mg or alendronate 40 mg(doses typically used for Paget's disease) aswell as code of Paget's disease atany point during the study period wereexcluded from the study cohort. Patientswho switched between alendronate andrisedronate during the treatment periodswere excluded from the study as well toallow valid comparisons of GI-related medicalcosts attributable only to the specificbisphosphonate.





GI-related Medical Resource Utilization.Resource utilization was evaluated duringthe 4-month treatment period followinginitiation of the bisphosphonate therapy interms of GI-related outpatient and inpatientvisits and GI medications. Primary diagnosis codes and primary codes identified a priori were used to determineGI-related inpatient and outpatient visits.Only primary and codes were used for the analysis as a conservativeapproach for investigating GI-relatedresource utilization and its association withbisphosphonate use. This approach avoidscounting secondary and tertiary resourceutilization, thereby enhancing specificity.

GI-related inpatient visits consisted ofvisits to surgical suites, extended care facilities,emergency rooms, and inpatient hospitalstays. An inpatient hospital stay wasdefined as a single episode of care andcounted as a single visit, even if the hospitalstay was longer than 1 day. GI-related outpatientvisits included both outpatient hospitalstays and physician office visits.Patients with GI-related events could havemore than 1 inpatient and/or outpatient visitin the first 4 months after initiating bisphosphonatetherapy.

GI-related Direct Medical Costs. Fromthe database, we assessed GI-related directmedical costs using reported reimbursedamounts in 2002 US dollars. Direct medicalcosts were measured in the 4-month treatmentperiod following initiation of the bisphosphonatetherapy in terms of GI-related(1) "outpatient cost" defined as a cost ofhospital outpatient and physician office visits;(2) "inpatient cost" defined as a cost ofstay as a hospital inpatient, or in an emergencyroom, surgical suite, or extended carefacility; (3) "GI medication cost" defined asa cost of prescription for H2 antagonists,PPIs, and cytoprotectives; and (4) "other GI-relatedcosts."

All costs associated with GI-related outpatientcare as defined above included testsand procedures in outpatient locations.Likewise, all costs associated with GI-relatedinpatient care included tests and proceduresperformed in the inpatient carelocations. For example, GI-related surgerywas a part of inpatient cost, and GI-relatedlaboratory tests were a part of outpatientcost. GI medication cost included pharmacydispensing fees. Cost of over-the-counter(OTC) GI medications was not includedbecause the database does not track OTCuse. Other costs included the GI-relatedcosts of skilled nursing facilities, nursinghomes, and ambulances.

A conservative approach was employed toaddress cost outliers. A GI-related cost ofless than $5 was excluded from the analysis,because this cost was likely the result of capitatedreimbursement. Including such costsin the analysis would introduce bias to costestimates, because capitated reimbursementdoes not represent a true service cost to anMCO. Patients with a total cost of more than$2000 (a midpoint of the reported averagetotal GI-related medical costs to an MCO:$1368-$27808) were examined further andwere checked against the secondary diagnosis.If the secondary diagnosis indicated asevere condition that was not GI related (eg,heart failure or diabetes), the reimbursementfor the entire service claim wasexcluded from the total cost of this patient.


Statistical Analysis. All cost andresource utilization comparisons betweenalendronate and risedronate patients wereperformed using the 2-sample, 2-sided test.A 0.05 level of significance was used todetermine statistically significant differencesbetween bisphosphonate treatments.


Baseline Characteristics of the PatientPopulation and GI Events During Followup.A total of 4259 women and men aged 65years and older fulfilled the inclusion criteriaand were included in the analysis. Ofthese patients, 623 (15%) received risedronatetreatment and 3636 (85%) receivedalendronate treatment, a reflection of themarket share of these drugs at the time ofthe study. There were no differencesbetween the risedronate and alendronatepatient cohorts in age, gender, or exposureto bisphosphonates (Table 1). Each cohortwas comprised of 94% women, and the meanage was approximately 75 years in bothgroups. An average overall exposure timetaking each bisphosphonate therapy wasapproximately 80 days. No difference inNSAID use between cohorts in either the 6-month pretreatment period or 4-monthtreatment period was observed.

Patients treated with alendronate daily(5 or 10 mg/day) and alendronate weekly(35 or 70 mg/week) were comparable interms of GI-related medical resource utilizationand associated direct medical cost(Table 2). Although the total medical costwas lower for patients treated with alendronateweekly, the difference in costbetween daily and weekly alendronate wasnot statistically significant. Therefore, dailyand weekly doses of alendronate were combinedfor the analyses.



The incidence of specific GI events duringthe 4-month follow-up period among alendronate- and risedronate-treated patientsis enumerated in Table 3. Overall, alower proportion of patients taking risedronateexperienced GI-related eventsduring the 4-month follow-up period(3.4% vs 5.4%, = .034). These frequenciesrepresent unique GI events (based on specificGI-related codes) experiencedby unique patients in each GI-related category(ie, GI bleeding, ulcers, GI symptoms,abdominal pain, and other). However, apatient may have experienced GI events indifferent categories and, thus, the samepatient may be present in multiple GI-relatedcategories.


The distribution of patients who experiencedGI events and/or GI medications, andGI-related outpatient and inpatient visitsduring the 4-month period following initiationof bisphosphonate therapy is presentedin Table 4. A higher proportion of alendronate-treated patients required GI-relatedoutpatient care compared with risedronate-treatedpatients (4.0% vs 2.6%, = .055)(Table 4).



GI-related Medical Resource Utilization.Medical resource utilization per 1000members per month associated with bisphosphonate-related GI events in the follow-up period is presented in Table 5.Risedronate-treated patients consistentlyutilized fewer medical resources (outpatientand inpatient visits) compared withalendronate-treated patients in the first 4months after initiating bisphosphonatetherapy. Risedronate-treated patients averagedjust over half the number of outpatientvisits incurred by alendronate-treatedpatients (8.8 vs 15.3 visits per 1000 membersper month, = .027). Risedronate-treatedpatients averaged fewer than halfthe number of inpatient visits of alendronate-treated patients (2.4 vs 6.0 visitsper 1000 members per month, = .016).



Direct Medical Cost of GI Events. Acomparison of the direct medical costs of GIevents and/or GI medications in patientsreceiving alendronate or risedronate is providedin Table 6. During the first 4 monthsafter initiation of bisphosphonate therapy,the total average GI-related direct medicalcost per 1000 members per month was significantlylower for risedronate-treatedpatients compared with alendronate-treatedpatients ($2164 vs $5996, < .001).Furthermore, when comparing the risedronatewith alendronate weekly dose only,the total average GI-related direct medicalcost per 1000 members per month remainssignificantly lower for risedronate-treatedpatients ($2164 vs $5639, = .001).



An average GI-related inpatient cost per1000 members per month for risedronate-treatedpatients was 6 times lower than theinpatient cost for alendronate-treatedpatients ($436 vs $2691, < .001). For outpatientcare, the average GI-related costper 1000 members per month for risedronate-treated patients was more than 2.5times lower than the corresponding costfor alendronate-treated patients ($860 vs$2223, = .009).


This study estimated the GI event rates,medical resource utilization, and directmedical costs of GI events in an elderly populationinitiating bisphosphonate therapy.An important observation is the differencesin clinical versus economic impact of managingthese events. GI-related medical costduring the first 4 months of treatment forpatients with osteoporosis taking alendronatewas nearly 3 times higher than thecost for patients taking risedronate ($5996vs $2164, per 1000 members per month).The cost of inpatient care appeared to be theprimary reason for the differential in totalcost observed between alendronate- andrisedronate-treated patients. This suggeststhat alendronate-treated patients may haveexperienced more severe GI events that ledto inpatient care. Inpatient visits–typicallythe most expensive component in overallmedical care–were 2.4 and 6 visits per 1000members per month for risedronate andalendronate, respectively. This 2.5-fold differencein inpatient visits translates to a 6-fold difference in inpatient cost ($436 vs$2691 per 1000 patients per month for risedronateand alendronate, respectively).These differences between GI-related costsfor risedronate versus alendronate representa substantial potential savings in overallcosts for an MCO. It is important for MCOsto look beyond the clinical differencesbetween treatments and also consider thefinancial impact to a plan and its members.

In examination of the specific events,there was a higher proportion of GI bleedingin patients with alendronate, which mayhave led to increased medical resource utilizationin all categories (inpatient and outpatientcare). Ulcers and GI symptoms bothtrended higher for patients taking alendronate,with esophageal reflux occurringonly among patients taking alendronate. Thesmall numbers preclude a thorough evaluationof specific GI event categories. Overall,GI events showed a statistically significantdifference between alendronate- and risedronate-treated patients. When taking intoaccount prophylaxis or treatment with GImedications, the observed GI event rate wasstill higher for patients treated with alendronate.As expected, fewer patients receivingGI medications had a GI event.

Although both daily and weekly doses ofalendronate have demonstrated GI tractadverse event profiles similar to placebo incontrolled trials,9-11 postmarketing studies ofalendronate daily have found an associationwith GI tract adverse events,7,12-15 suggestinga safer GI tract adverse event profile foralendronate weekly. In contrast, statisticallysignificant differences in GI-related medicalresource utilization and associated directmedical costs between patients taking alendronateweekly versus alendronate dailydosing were not observed.

Our findings have important implicationsfor managed care systems and nationalhealthcare programs. Osteoporosis is aprevalent disease in the elderly, and healthcaresystems will be under increasing pressureto treat this disorder as the proportionof elderly patients continues to rise. For atypical MCO, the estimated average GI-relatedcost per 1000 alendronate-treatedpatients per year would be $71 952 comparedwith $25 968 per 1000 risedronate-treatedpatients per year. This is a savings ofalmost $46 000 per year for every 1000patients who would initiate risedronate therapyinstead of alendronate therapy. If only50% of patients started osteoporosis therapywith risedronate instead of alendronatewithin a given plan, the savings could besubstantial.


There are several limitations in conductingthis type of investigation. The approachof using a large administrative database doesnot allow for individual chart review to validateGI-related codes, althoughthe identification of GI-related prescriptionsor procedures can provide internal validationof the diagnosis. It does not capture theuse of OTC GI medications either, thus theburden of GI medication use and its costmay be underestimated. Some GI medicationscaptured in the database may havebeen prescribed for prophylaxis rather thantreatment, adding to the overall cost of themedication. The potential bias of this type ofmedication prescribing or use would notnecessarily favor one treatment over theother, because there were no apparent differencesbetween the 2 groups in the cost ofprescription GI medications. We includedany prescription GI medication use as a partof total GI-related costs since any bisphosphonate-related GI medication use constitutesan additional economic burden for anMCO, regardless of the indication. Also, thedatabase does not allow distinguishingbetween inpatient medical costs that correspondto primary or secondary diagnosis;therefore, inpatient cost might be slightlyoverestimated. At the same time, we excludedvery high GI-related inpatient cost whenthe secondary diagnosis indicated a severecondition that was not GI related for patientswith total direct medical costs of morethan $2000.

In spite of these limitations, the use of anintegrated claims database provides anopportunity to observe treatment patternsthat occur in real-world practice. Such databaseshave the additional advantage ofincluding a large number of patients andincorporating many types of health plansand physician specialties, permitting a morecomplete picture of GI-related events, prescribinghabits, and costs once a drug is inuse across large populations.



Sensitivity Analysis. In this study weapplied the following restrictions to the analysis:(1) only primary and codes were included in the analysis to defineGI-related resource utilization and associatedmedical costs, and (2) the reimbursement forthe entire service claim was excluded fromthe total cost of a patient with a total cost ofmore than $2000 if the secondary diagnosisindicated a severe condition that was not GIrelated (eg, heart failure, diabetes). To assesswhether these restrictions introduced anybias in the results of this study, we comparedthe GI-related medical cost of risedronate-treatedpatients to the cost of alendronate-treatedpatients without applying theserestrictions. Adding secondary GI-relateddiagnoses and procedures to the definition ofa GI event should contribute to higher GI-relatedtotal direct medical costs for bothrisedronate- and alendronate-treated patients.Including upper bound outliers in the analysis,however, should have an impact only on thetotal medical cost of the alendronate cohortcaused by nonexistent outliers in the risedronatecohort.

There were 10 patients in the alendronatecohort who had total GI-related medicalcosts of more than $2000; however, only 3alendronate-treated patients had secondaryconditions that were not GI related and werecontributing to the higher inpatient costs forthese patients. When these costs are addedback, the inpatient cost of alendronate-treatedpatients increased from $2691 to $3432per 1000 patients per month.


When both GI-related primary and secondarydiagnoses are included in the analysis,the total GI-related medical costincreased for both risedronate- and alendronate-treated patients. However, the significantlylower total average GI-relateddirect medical cost per 1000 members permonth for risedronate-treated patients comparedwith alendronate-treated patients wasstill evident ($7961 vs $29 410, = .036).


In a patient population 65 years of age orolder initiating bisphosphonate therapy forosteoporosis, risedronate was associatedwith markedly lower GI-related diagnoses,medical resource utilization, and directmedical costs compared with alendronate.

These data suggest that risedronate usehas the potential to reduce GI-relatedhealthcare resources use and associateddirect medical costs relative to alendronate.Physicians and managed care decision makersshould consider cost differences amongoral bisphosphonates, because they relate toclinically significant GI events, when determiningagents for inclusion onto managedcare formularies.


We thank Lisa Bosch of Procter &Gamble, Inc, for her assistance in preparingthe manuscript.

This study was funded by Procter & Gamble Pharmaceuticals, Mason, OH, and Aventis Pharmaceuticals, Bridgewater, NJ.

Arch Intern Med.

1. Bauer DC, Black D, Ensrud K, et al. Upper gastrointestinaltract safety profile of alendronate: the FractureIntervention Trial. 2000;160:517-525.

Mayo Clinic Proc.

2. Taggart H, Bolognese MA, Lindsay R, et al. Uppergastrointestinal tract safety of risedronate: a pooledanalysis of 9 clinical trials. 2002;77:262-270.

N Engl J Med.

3. Liberman UA, Weiss SR, Broll J, et al. Effect of oralalendronate on bone mineral density and the incidenceof fractures in postmenopausal osteoporosis. TheAlendronate Phase III Osteoporosis Treatment StudyGroup. 1995;333:1437-1443.


4. Black DM, Cummings SR, Karpf DB, et al. Randomisedtrial of effect of alendronate on risk of fracturein women with existing vertebral fractures. FractureIntervention Trial Research Group. 1996;348:1535-1541.


5. Cummings SR, Black DM, Thompson DE, et al. Effectof alendronate on risk of fracture in women with low bonedensity but without vertebral fractures: results from theFracture Intervention Trial. 1998;280:2077-2082.

6. Colby CJ, Levin TR, Boyko WL, Schein JR. Cost ofacid-related disorders associated with alendronate use forosteoporosis in a large MCO. Paper presented at: 21stAnnual Meeting of the American Society for Bone andMineral Research; September 30, 1999; St. Louis, Mo.

Am J Manag Care.

7. Ettinger B, Pressman A, Schein J. Clinic visits andhospital admissions for care of acid-related upper gastrointestinaldisorders in women using alendronate forosteoporosis. 1998;4:1377-1382.

Am J Med.

8. Levin TR, Schmittdiel JA, Kunz K, et al. Costs of acid-relateddisorders to a health maintenance organization. 1997;103:520-528.

Mayo Clin Proc.

9. Greenspan S, Field-Munves E, Tonino R, et al.Tolerability of once-weekly alendronate in patients withosteoporosis: a randomized, double-blind, placebo-controlledstudy. 2002;77:1044-1052.

Int J Clin Pract Suppl.

10. Watts N, Freedholm D, Daifotis A. The clinical tolerabilityprofile of alendronate. 1999;101:51-61.


11. Schnitzer T, Bone HG, Crepaldi G, et al. Therapeuticequivalence of alendronate 70 mg once-weeklyand alendronate 10 mg daily in the treatment of osteoporosis.Alendronate Once-Weekly Study Group. 2000;12:1-12.

N Engl J Med.

12. Liberman UI, Hirsch LJ. Esophagitis and alendronate. 1996;335:1069-1070.

Dig Dis Sci.

13. Graham DY. What the gastroenterologist shouldknow about the gastrointestinal safety profiles of bisphosphonates. 2002;47:1665-1678.

Pharmacoepidemiol Drug Saf.

14. Park BJ, Clouse J, Shatin D, Stergachis A. Incidenceof adverse oesophageal and gastric events in alendronateusers. 2000;9:371-376.

N Engl J Med.

15. de Groen PC, Lubbe DF, Hirsh LJ, et al. Esophagitisassociated with the use of alendronate. 1996;335:1016-1021.