Concerned about rising prevalence and costs ofdiabetes among its employees, Pitney Bowes Increcently revamped its drug benefit design to synergizewith ongoing efforts in its disease managementand patient education programs. Specifically, basedon a predictive model showing that low medicationadherence was linked to subsequent increases inhealthcare costs in patients with diabetes, the companyshifted all diabetes drugs and devices from tier2 or 3 formulary status to tier 1. The rationale wasthat reducing patient out-of-pocket costs would eliminatefinancial barriers to preventive care, and therebyincrease adherence, reduce costly complications,and slow the overall rate of rising healthcare costs.This single change in pharmaceutical benefit designimmediately made critical brand-name drugs availableto most Pitney Bowes employees and their covereddependents for 10% coinsurance, the samecoinsurance level as for generic drugs, versus the previouscost share of 25% to 50%. After 2 to 3 years,preliminary results in plan participants with diabetesindicate that medication possession rates haveincreased significantly, use of fixed-combinationdrugs has increased (possibly related to easier adherence),average total pharmacy costs have decreasedby 7%, and emergency department visits havedecreased by 26%. Hospital admission rates,although increasing slightly, remain below the demographicallyadjusted Medstat benchmark. Overalldirect healthcare costs per plan participant with diabetesdecreased by 6%. In addition, the rate ofincrease in overall per-plan-participant health costsat Pitney Bowes has slowed markedly, with net per-plan-participant costs in 2003 at about $4000 peryear versus $6500 for the industry benchmark. Thisrecent moderation in overall corporate health costsmay be related to these strategic changes in drugbenefit design for diabetes, asthma, and hypertensionand also to ongoing enhancements in the company'sdisease management and wellness programs.
(Am J Manag Care. 2005;11:S170-
Direct medical costs for diabetes andrelated complications are notoriouslyhigh. The Centers for Disease Controland Prevention estimates the cost at $92 billionper year,1 and health insurance expendituresfor the individual with diabetes aretriple those of the average consumer.2 Fromthe US employers' perspective, the burden ofdiabetes extends even further to include the$40-billion annual cost for indirect expendituresbecause of disability, work loss, andpremature mortality.1,3,4 In fact, disease-relatedwork absences and disabilityaccount for about one third of the total costper employee with diabetes.5
Fortunately, investing in aggressive diabetescontrol not only improves blood glucoselevels but reduces medical complications andcosts6-8 and may also boost productivity andlower absenteeism.3,9,10 Although optimizeduse of medications is essential to improveddiabetes control,9 the trend in pharmaceuticalbenefit design has been to pass higherportions of drug costs to members to slow thegrowth of the healthcare budget. But employersmust balance this need for employeecost-sharing and "demand management"with the desire to increase access to optimalcare for costly conditions like diabetes.11 Ifnot managed judiciously, the high copaysand coinsurance may become a financial barrierto proper diabetes care. The short-termsavings may actually impede the opportunityfor longer-term health budget savings andproductivity gains.
At Pitney Bowes, the escalating cost ofhealthcare hit home most recently in theyear 2000, when per-employee claimsjumped by 13% versus an increase of just3% in the benchmark of similar companies(as measured by the Hewitt Health ValueIndex). Although the company's averageper-participant cost of approximately$3300 per year remained well below thebenchmark average, the double-digitincrease was cause for alarm. Some of theincrease was attributed to managed carecompanies cutting back on utilization management programs and some to an averageemployee age slightly above the benchmark,thus increasing prevalence ofchronic medical conditions. Perhaps relatedto this demographics issue was theadditional fact that more extremely high-cost(>$100 000) cases of end-stage long-termdisease were being treated. But thesewere only guesses about the real sources ofthe cost spike.
As described in this article, the company'sconcerns led it to tap its own considerableinformation resources to identify itstrue cost drivers for several long-term diseases.After evaluating these drivers, managementthen altered its prescription drugplan in an attempt to boost plan participant(ie, employee and covered dependents)access to medications required for the treatmentof key chronic conditions such as asthma,diabetes, and hypertension. Key findingsrelated to diabetes are presented in thisreport because the details of employer-initiatedefforts may be relevant to any healthcaremanager dealing with rising costs forlong-term diseases.
The Opportunity at Pitney Bowes
Pitney Bowes, a Fortune 500 companyproviding integrated mail and documentmanagement systems services and solutions,has 35 000 employees worldwide and a revenueof $5 billion per year. Within the UnitedStates, the company's 23 000 employees are58% male with an average age of 41 years andan average length of service of 8.1 years.Twenty-five percent of employees are locatedin the New York, New Jersey, andConnecticut tristate area, with wide dispersalacross the remaining states.
The company has an integrated health,wellness, and disability strategy with databasesproviding timely feedback on each.For example, cost data are available on allplan participant medical and pharmacyexpenditures. Also available for scrutiny areemployee disability rates, absence data,worker compensation costs, and demographicinformation along with employeerisk factors and behaviors, productivityinformation, and selected survey results. Alldata are available in aggregated, deidentifiedformat that is fully compliant with theHealth Insurance Portability and AccountabilityAct of 1996 (HIPAA).
The medical benefits at Pitney Bowes areavailable through self-insured or fully-insuredplans and share a common benefitdesign. About 80% of plan participants opt fora self-insured plan. There are 46 local andnational health maintenance organizationcarriers and 4 national preferred providerorganizations. In all of the self-funded plansand a few of the others, the drug benefits areprovided by a carve-out pharmacy benefitmanager. This coverage of approximately90% of all employees under 1 common pharmaceuticalplan provides a potentially powerfulsingle point of entry for studying—andfor leveraging—long-term disease outcomesin the Pitney Bowes population.
Typically, disease management programsare the logical solution for measuring andmanipulating health in areas such as asthma,cardiovascular disease, and diabetes.Although disease management is part of theintegrated effort of long-term disease care atPitney Bowes—in fact, such programsremain a condition for bidding on companybusiness—the presence of close to 50 separatehealth plan vendors precluded any easyopportunity to have an impact on diabetescare with a single employer-driven programchange. The company had already introducedan Internet-based health portal foremployees and an opt-in voluntary diseasemanagement program, and modified theirexisting wellness program to encouragemore long-term disease awareness and treatment.These existing disease managementand patient education/wellness programshelped form a broad base of a sensible benefitspackage aimed at improving care andlimiting total costs. Still, company managementwas looking for a new catalyst tochange health behaviors and limit costs.Specifically, they wanted to get ahead of thecost curve by quickly shifting more plan participantswith expensive long-term diseasesinto prevention compliance.
In this situation, the drug benefit designsuggested itself as a logical tool for employer-driven population management. Thepotential value of this common denominatorof employee health became even more strikingafter the company's health benefits managers realized, as discussed next, that lowmedication adherence was linked to high-costclaims.
The Predictive Model:What Causes High-cost Claims?
Shortly after the 13% cost surge of 2000,the company commissioned an analysis todetermine what population-based factorscaused plan participants to migrate from"normal-cost" ($400-$700 per year) to "high-cost"status (>$10 000 per year). A consultingcompany called Medical Scientists Inc(recently acquired by LifeMasters SupportedSelfCare, Inc) worked with Pitney Bowes todevelop a hybrid artificial intelligence programthat defined the end point in question—in this case, the transition to high costs—andthen identified the population-level variablesassociated with that end point. The programwas unique in the way that it scoured thedatabase for any employee variable linked toincreased costs—not just the preconceivedvariables such as age, concomitant disease, orhemoglobin A1C level. Notably, the predictivemodel also considered the total cost of carefor employees, including not only direct medicalcosts, such as medical claims, pharmacy,and behavioral health, but also indirect costsrelated to absenteeism and disability. Thedatabase and cost assumptions for the predictivemodeling were set up and are stillmaintained by Medstat, a division of theThomson Corporation.
The model results quickly confirmed therelationship between long-term conditionsand future high total healthcare costs. A keyactionable finding was that illness burdenand costs were driven by a lack of pharmaceuticaladherence. In diabetes, those planparticipants with 9 or fewer 30-day prescriptionfills for their diabetic medications weremost likely to transition into the high-costgroup. In other words, the patients with diabeteswho were refilling their insulin andoral medications only two thirds of the timeor less were likely to become the costliest.Similar findings were identified for asthmaand hypertension. Again, lack of adherenceto treatment has been known for decades tocontribute to diabetes complications.12,13 Butreading about nonadherence in a dry journalarticle and seeing a low possession ratelinked directly to next-year per-patient costsin a printout from your own health plandatabase are 2 quite different experiences.As described next, this stark evidence ofnonadherence as a cost driver led to questioningof fundamental assumptions aboutcost sharing, price elasticity, and drug accessibilityin the Pitney Bowes prescriptionbenefit plan.
Redesigning the Pharmacy Benefitsfor Diabetes
The drug plan at Pitney Bowes is not dissimilarfrom that used by many other largecorporations. The 2 main options are theRegular Rx Plan and the "buy-up" Extra RxPlan with a slightly lower coinsurance andcopays, and an annual out-of-pocket maximumof $500 (Table 1). About 1 of every 4employees elects the Extra Rx Plan. Theplan is built on a traditional 3-tier formularybut has several consumer-centric elements,including limited prior authorization (only 6drugs require it, all because of safety concerns)and no policies calling for mandatorygenerics, step therapy, or therapeutic substitution.Overall, the plan achieves a 75%/25%cost sharing with employees and is consideredto be performing well in terms of standardpharmacy benefit measures, with an11% rate of mail service prescriptions, anoverall 48% generic drug utilization, and ageneric drug fill efficiency rate of 91%.
In January of 2002, however, the companyradically modified the pharmacy planfor participants with diabetes, asthma, orhypertension, allowing them to pay forbrand-name medications at the tier 1 rates(eg, 10% coinsurance) and thus reducing apotential barrier to care for long-term disease(Figure 1). These changes were initiateddespite the inevitable financial lossbecause of lower coinsurance paymentsand copays as well as foregone supplierrebates, an annual hit estimated at $1million. Of course, the whole purpose inlimiting out-of-pocket payments was toeliminate a potential reason that employeesdiscontinue taking their medications andthereby to keep more of them on therapyand out of the hospital or emergencydepartment (ED). So such a change couldsave money too.
But will it? There is ample literature tosupport the concept that excessive copayscause suboptimal use of essential medications.11,14,15 One recent study, for example,found that doubling the copay for diabetesdrugs led to a 23% decrease in per-memberper-year drug days supplied.15 The AmericanDiabetes Association also warns explicitlyabout the short-sightedness of erecting costbarriers to diabetes medications (see Sidebar,"Tight Cost Controls May Be Barrier toDiabetes Management").12 Still, the actualreturn on investment from lowered copays isless well documented and would, anyway,vary considerably from plan to plan andcompany to company. Thus, the polestar forthis effort at Pitney Bowes remains the internalcompany evidence linking low prescriptionfill rates to high subsequent costs.
For diabetes, the major benefit designchange involved moving a number of tier 2and 3 drugs to tier 1. These included insulinproducts such as Humalog, Humulin,Lantus, Novolin, and NovoLog as well as oralagents such as Actos (pioglitazone), Amaryl(glimepiride), Avandia (rosiglitazone),Avandamet (rosiglitazone/metformin), GlucotrolXL (glipizide extended release),Glucovance (glyburide/metformin), Prandin(repaglinide), Precose (acarbose), andStarlix (nateglinide). No judgments weremade about the relative efficacies of theseagents, the weight of clinical evidence orvalue-based arguments supporting their use,or about the manufacturer. Additionally,any test strips that were in tier 2 or 3 (eg,Accu-Chek, OneTouch Ultra) were alsoshifted to tier 1. Although the company wasalso enhancing its diabetes disease managementand wellness efforts in parallel withthese pharmacy benefit changes—forexample, glucometers were supplied free ofcharge to employees with diabetes—the trulynovel element within this evolving integratedapproach was the new benefit design.
Results in Diabetes: Better Adherence,Lower Costs
For the typical plan participant with diabetes,the formulary change had the immediatedesired financial impact, with theaverage cost of a 30-day fill dropping by50%. Many patients were paying 80% lessthan their previous drug costs—ie, a 10%coinsurance payment rather than a 50%payment.
For the company overall, the preliminaryresults in the 2- to 3-year period after thechange have also been promising. As trackedby Caremark Inc, Pitney Bowes' pharmacybenefits manager, rates of adherence with allmedications that shifted tiers increased significantly.Perhaps most important, the percentageof members with suboptimaladherence with insulins decreased by fullytwo thirds. Also, the percentage of membersusing fixed-combination oral hypoglycemicsincreased from 9% to 22%—andthe increases in adherence rates wereparticularly high for these individuals takingcombination therapy. Finally, amonginsulin-dependent diabetic plan participants,the shift to newer brands of test stripsin tier 1 was associated with a doubling inthe usage rate of these test strips on glucometers(from 28% usage to 55% usage).
Naturally, the company's pharmacy costsfor the insulin, insulin stimulators, insulinsensitizers, and test strips increased duringthis time. But the surprise was that althoughthe company's total annual pharmacy costsper covered person showed a mild increase,pharmacy costs for those with diabetes actuallydecreased by 7%. This overall decrease inpharmacy costs for employees with diabeteswas thought to result from a reduction incomplications and the avoided need for othereven more expensive drugs. Further detailsof this shift will be explored in a subsequentanalysis. In fact, the most recent data showthat total per-member, per-month pharmacycost increases for all participants in thePitney Bowes active pharmacy benefits planhave remained relatively stable, with annualincreases in the low double-digits, over thepast 3 years (Figure 2) despite the extracompany spending to increase access tomedications for 3 major long-term diseases.
Medical utilization and costs for plan participantswith diabetes also decreasedbetween 2001 and 2003. The rate of ED visitsdropped by 26% in absolute terms andfurther distanced itself from the benchmarkrate (Table 2). Although it cannot be provedin this setting, this sharp decrease in ED visitsis likely related to improved adherencewith the oral hypoglycemia medications.The hospitalization rate increased slightly inparticipants with diabetes, a potential resultof the aging of the workforce. But note thatthis rate also remained below the demographicallyadjusted benchmark rates derivedfrom the Medstat database. Overall,the per-patient cost of care for Pitney Bowesplan participants with diabetes decreased by6% from 2001 to 2003.
These reductions in per-participant costof care for diabetes (and for asthma, forwhich reductions were also seen) likely contributedto the encouraging overall trend innet per-employee medical direct costs. Asshown in Figure 3, the average annualincrease in employee health cost from 2000to 2003 was 8.1% versus composite annualincreases of 12% to 15% for benchmark companies.Especially in light of the relativelyshort period of time since the benefitchanges were implemented, this moderationin employee health cost increases isextremely promising. Based on these preliminaryfindings, the healthcare team at PitneyBowes continues to track results—includingthe impact of improved care on indirectcosts, another potential contributor to thecompany bottom line—and to consider similaraccess-driven benefit changes in additionallong-term disease categories.
Sharp increases in diabetes prevalence atPitney Bowes during the past few years—the rate is now 36 episodes per 1000employees, an increase of about 50% from2001—makes the findings just presentedall the more compelling. As more of theUS population is diagnosed with diabetesand as the workforce continues to age, thepressures on the costs of diabetes and cardiovascularcare will build. As indicatedin this preliminary project, health managersmay be able to improve care andlimit overall costs for diabetes by selectivelylowering barriers to appropriatepharmaceutical access. This simplechange in benefit design can deliver anadded spark to established disease managementand wellness approaches in long-termcare.
1. Centers for Disease Control and Prevention. Nationaldiabetes fact sheet. Available online at: www.cdc.gov/diabetes/pubs/estimates.htm. Accessed April 21, 2005.
2. Peele PB, Lave JR, Songer TJ. Diabetes in employer-sponsoredhealth insurance. 2002;25:1964-1968.
3. Ng YC, Jacobs P, Johnson JA. Productivity lossesassociated with diabetes in the US. 2001;24:257-261.
4. Mayfield JA, Deb P, Whitecotton L. Work disabilityand diabetes. 1999;22:1105-1109.
5. Ramsey S, Summers KH, Leong SA, Birnbaum HG,Kemner JE, Greenberg P. Productivity and medical costsof diabetes in a large employer population. 2002;25:23-29.
N Engl J Med.
6. The Diabetes Control and Complications TrialResearch Group. The effect of intensive treatment ofdiabetes on the development and progression of longtermcomplications in insulin-dependent diabetes mellitus.1993;329:977-986.
7. UK Prospective Diabetes Study (UKPDS) Group.Intensive blood-glucose control with sulphonylureas orinsulin compared with conventional treatment and riskof complications in patients with type 2 diabetes(UKPDS 33). 1998;352:837-853.
8. Gilmer TP, O'Connor PJ, Manning WG, Rush WA.The cost to health plans of poor glycemic control.1997;20:1847-1853.
9. Testa MA, Simonson DC. Health economic benefitsand quality of life during improved glycemic control inpatients with type 2 diabetes mellitus: a randomized, controlled,double-blind trial. 1998;280:1490-1496.
10. Rizzo JA, Abbott TA 3rd, Pashko S. Labour productivityeffects of prescribed medicines for chronically illworkers. 1996;5:249-265.
11. Karter AJ, Stevens MR, Herman WH, et al;Translating Research Into Action for Diabetes StudyGroup. Out-of-pocket costs and diabetes preventive services:the Translating Research Into Action for Diabetes(TRIAD) study. 2003;26:2294-2299.
12. American Diabetes Association. Third-party reimbursementfor diabetes care, self-management education,and supplies. 2005;28(suppl 1):S62-S63.
13. Pladevall M, Williams LK, Potts LA, Divine G, Xi H,Lafata JE. Clinical outcomes and adherence to medicationsmeasured by claims data in patients with diabetes.2004;27:2800-2805.
N Engl J Med.
14. Huskamp HA, Deverka PA, Epstein AM, Epstein RS,McGuigan KA, Frank RG. The effect of incentive-basedformularies on prescription-drug utilization and spending.2003;349:2224-2232.
15. Goldman DP, Joyce GF, Escarce JJ, et al. Pharmacybenefits and the use of drugs by the chronically ill.2004;291:2344-2350.