1. Top of page
  2. Abstract


To assess the impact of patient out-of-pocket (OOP) expenditures on adherence and persistence with biologics in patients with rheumatoid arthritis (RA).


An inception cohort of RA patients with pharmacy claims for etanercept or adalimumab during 2002–2004 was selected from an insurance claims database of self-insured employer health plans (n = 2,285) in the US. Adherence was defined as medication possession ratio (MPR): the proportion of the 365 followup days covered by days supply. Persistence was determined using a survival analysis of therapy discontinuation during followup. Patient OOP cost was measured as the patient's coinsurance and copayments per week of therapy, and as the proportion of the total medication charges paid by the patient. Multivariate linear regression models of MPR and proportional hazards models of persistence were used to estimate the impact of cost, adjusting for insurance type and demographic and clinical variables.


Mean ± SD OOP expenditures averaged $7.84 ± $14.15 per week. Most patients (92%) paid less than $20 OOP for therapy/week. The mean ± SD MPR was 0.52 ± 0.31. Adherence significantly decreased with increased weekly OOP (coeff = −0.0035, P < 0.0001) and with a higher proportion of therapy costs paid by patients (coeff = −0.8794, P < 0.0001), translating into ∼1 week of therapy lost per $5.50 increase in weekly OOP. Patients whose weekly cost exceeded $50 were more likely to discontinue than patients with lower costs (hazard ratio 1.58, P < 0.001).


Most patients pay less than $20/week for biologics, but a small number have high OOP expenses, associated with lower medication compliance. The adverse impact of high OOP costs on adherence, persistence, and outcomes must be considered when making decisions about increasing copayments.


  1. Top of page
  2. Abstract

Evidence has been accumulating that cost sharing affects utilization, adherence, and persistence with medications among commercially insured populations (1–7). Comparative, quasi-experimental studies have investigated the effects of benefit design changes (e.g., addition of a third, nonpreferred brand-name tier) or copayment increases on utilization of pharmaceuticals.

In 2 controlled studies in commercially insured populations, Fairman et al and Motheral and Fairman compared pharmaceutical and medical utilization after implementation of a third tier (nonformulary medication) in the benefit plan (2, 3). The intervention groups demonstrated lower utilization of third-tier medications and lower growth in prescription claims relative to comparison groups whose employers retained the 2-tier benefit structure. Medication continuation rates did not significantly differ between the intervention versus comparison groups for most of the drugs studied. In another pre-post, controlled study, Motheral and Henderson found that increased copayment for brand drugs was associated with lower utilization of brand drugs but not lower overall utilization in the intervention group, and with increased brand claims in the control (unchanged copayment) group (4). Huskamp et al showed that employees facing a major increase in cost sharing were more likely than a control group to switch to a lower-tier drug or to discontinue medication altogether (5). A more moderate change in benefit design was associated with switching but not discontinuation. Brixner et al studied persistence and compliance with medication for 5 chronic disease categories among members of an integrated health care network (6). Discontinuation rates were significantly higher in 4 of these categories for patients experiencing a copayment increase compared with patients with no benefit change. However, compliance (as measured by medication possession ratio [MPR]) decreased in only 1 of the conditions studied.

A retrospective study by Goldman et al predicted the effects of doubling copayments on utilization in 8 therapeutic classes of drugs (1). Substantial reductions were observed for all classes, although use by patients with chronic illness tended to decrease less for drugs that treated their condition. Landsman et al noted similar behavior across 9 medication classes (7). The direct relationship between prescription drug cost sharing and medical outcomes is relatively unstudied; most studies of claims data use proxies for health status such as health care utilization and spending. Increased drug cost sharing may have adverse consequences in terms of health care utilization in patients with chronic illness. Goldman et al systematically reviewed the literature on associations among cost sharing, the use of prescribed drugs, the use of other health services, and health outcomes (8). The relationship was unambiguous in studies focusing on patients with congestive heart failure, lipid disorders, diabetes, and schizophrenia; greater patient cost sharing for drugs was associated with greater use of inpatient and emergency medical services. However, when the study population was not restricted to patients with chronic illness, elderly patients, or poor patients, increased copayments were not associated with increased health care utilization.

Rheumatoid arthritis (RA) is a chronic inflammatory disorder characterized by progressive disability due to inflammation and the destruction of joints. The goals of RA therapy are to prevent or control joint damage, prevent loss of function, and decrease pain (9). Anti–tumor necrosis factor α (anti-TNFα) agents, including the soluble receptor protein etanercept and the anti-TNFα monoclonal antibodies adalimumab and infliximab, have been among the most recent improvements in RA treatment. Infliximab is administered intravenously, usually in a physician's office, while etanercept and adalimumab are formulated for subcutaneous injection and are generally dispensed at pharmacies for patients to self-administer weekly or biweekly. Anti-TNFα agents have been shown to play a key role in the treatment of RA by decreasing inflammation, improving functional status, and inhibiting progression of joint damage (9–13). Consequently, among patients with RA, anti-TNFα therapy has been associated with increases in employability (14), the likelihood of employment, and the number of hours worked per week (15), as well as with decreases in absenteeism (14, 16–19) and the need to seek less physically demanding work (19). As with therapies for other chronic illnesses, poor adherence or discontinuation of anti-TNFα therapy could cause patients to forgo these benefits.

Employers and payers have responded to the rising costs of health care by making several benefit design changes that shift costs to patients, including increasing deductibles, instituting multi-tier drug formularies, and, more recently, implementing coinsurance (20, 21). With coinsurance, patients pay a percentage of a drug's cost rather than a fixed copayment; the patient's payment responsibilities for higher-cost medications can be a substantial monthly amount (21). These cost-sharing strategies are likely to have a disproportionate impact on a small number of patients with chronic diseases, and those with limited incomes. In particular, when therapies have no cheaper alternatives, patients without financial resources may have to forego therapy or try to get along on suboptimal doses (22).

The current study focused on patients newly prescribed with the anti-TNFα agents dispensed at pharmacies (etanercept and adalimumab). Infliximab was not included in this study because it is not customarily dispensed by pharmacies. The study objective was to determine whether patients' out-of-pocket (OOP) costs of therapy affected their compliance with these medications. Two similar aspects of compliance, adherence and persistence, were analyzed. Adherence was defined as the proportion of all the days in a year during which a patient had medication available; patients who skipped doses or those who discontinued had lower adherence. Persistence considered how long patients maintained therapy and what factors contributed to drug discontinuation.


  1. Top of page
  2. Abstract

Design and data source.

A retrospective cohort study was conducted using commercial insurance claims from the MEDSTAT MarketScan database for the years 2002–2004. This database captures person-specific information on clinical utilization and expenditures for inpatient procedures, outpatient procedures, and prescription drugs from self-insured employer health plans. These data, drawn from approximately 45 large employers, include health care claims of employees and dependents, early retirees, COBRA beneficiaries, and Medicare-eligible retirees with employer-sponsored Medicare Supplemental plans. Additionally, details regarding features of insurance benefits such as OOP limits were captured from each plan's Summary Plan Description booklet. These data were available for 155 individual plans from the 2002–2004 period, which covered ∼50% of the patients in our study. Recent publications using these data have reported studies of the effects of copayments on adherence to statins (23) and disease-modifying agents taken for multiple sclerosis (24).

Study population.

An inception cohort of patients diagnosed with RA who had newly initiated etanercept or adalimumab treatment was identified. Patients were required to have 1 inpatient or 2 outpatient diagnoses of RA (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 714.xx) at some time during 2002–2004, 6 months of insurance eligibility prior to index (i.e., their first etanercept or adalimumab prescription), and 12 months of insurance eligibility post index. This study only included patients who received subcutaneous etanercept or adalimumab at pharmacies. National Drug Codes (NDCs) were used to identify etanercept and adalimumab prescriptions. Patients who received infused medications (infliximab) and etanercept and adalimumab dispensed at a physician's office were not included because measurements of adherence behavior for physician-administered and self-administered medications are too dissimilar. Patients who initiated etanercept or adalimumab and switched to infliximab during the study followup period were excluded (n = 26). The demographics of the patients meeting the selection criteria (n = 2,285) are shown in Table 1.

Table 1. Characteristics of rheumatoid arthritis patients taking etanercept or adalimumab, 2002–2004*
 Frequency (n = 2,285)
  • *

    Values are the number (percentage) or the mean ± SD. DMARDs = disease-modifying antirheumatic drugs; NSAIDs = nonsteroidal antiinflammatory drugs; COPD = chronic obstructive pulmonary disease; EPO = exclusive provider organization; HMO = health maintenance organization; POS = point of service; PPO = preferred provider organization; OOP = out-of-pocket; anti-TNFα = anti–tumor necrosis factor α.

  • Six months prior to the index date, which is the date that the patient initiated etanercept or adalimumab.

  • Data on the cap features were not available for 1,274 patients.

  • §

    The patient was charged a coinsurance payment for adalimumab or etanercept during the followup period.

Female sex1,707 (74.7)
  DMARDs1,094 (47.9)
  NSAIDs1,544 (67.6)
  Narcotic analgesics1,007 (44.1)
  Hospitalization102 (4.5)
  Cardiovascular disease516 (22.6)
  Mental disorders141 (6.2)
  Infection287 (12.6)
  COPD or asthma130 (5.7)
  Cancer64 (2.8)
  Diabetes171 (7.5)
  Osteoporosis or fracture229 (10.0)
  Joint deformity, synovitis, or cervical   myelopathy84 (3.7)
  Carpal tunnel syndrome48 (2.1)
  Vasculitis8 (0.4)
Type of insurance plan 
  Comprehensive679 (29.7)
  EPO2 (0.1)
  HMO194 (8.5)
  POS303 (13.3)
  PPO1000 (43.8)
  POS with capitation105 (4.6)
Cap on patient's OOP expenditures 
  Medical only638 (63.1)
  No caps232 (23.0)
  Prescriptions and medical1 (0.1)
  Prescriptions only140 (13.9)
  Northeast218 (9.5)
  North Central710 (31.1)
  South894 (39.1)
  West461 (20.2)
  Unknown2 (0.1)
Urban1,706 (74.7)
Patient is the insured employee rather than the dependent1,280 (56.0)
Patient paid coinsurance for anti-TNFα agent§257 (11.3)
Patient's weekly OOP for anti-TNFα agent above $5087 (3.8)
Age, years54.2 ± 12.2
Charlson comorbidity index score1.55 ± 1.03
Number of physician visits 6 months prior to index3.36 ± 3.01
Weekly OOP for anti-TNFα agent, dollars7.84 ± 14.15
Proportion of anti-TNFα cost paid by patient, %2.7 ± 4.9
Medication possession ratio0.52 ± 0.31

Measures of adherence and persistence.

Each prescription claim in the database includes the days supply of the medication, which was determined based on the quantity of medication dispensed and the dosing information on the prescription. The MPR (total days supply of the medication divided by total days of followup), a common method of computing adherence in claims analyses (25), was used to measure adherence. The MPR was computed for each patient during the year of followup after their first prescription for etanercept or adalimumab.

Persistence was measured during the 1-year followup period using a survival analysis in which a patient was considered to have discontinued the drug if at least 30 days elapsed after the patient's cumulative days supply ran out. The discontinuation date was set to the date of the last prescription plus the number of days supplied by that prescription. Patients who switched between adalimumab and etanercept were considered to be adherent and persistent as long as the combination of the 2 drugs met the criteria for days supply.

The reliability of the days supply data element is critical to our computations of adherence and persistence. Prescription claims for adalimumab and etanercept were reviewed in order to determine the reliability of the values in the days supply field. The charge on each claim was compared with the wholesale price of the drugs, which was approximately $328 for a week's supply during 2004 (26). Costs of claims with ≤3 days supply were reviewed in detail, and 28 claims, made on behalf of 16 patients, were deleted because the cost was less than $120. The days supply of an additional 184 claims, affecting 59 patients, was revised by using the price of the claim to determine the days supply.

Independent variables.

Patients' OOP costs for anti-TNFα therapy were measured in 2 ways. First, the weekly OOP cost for each patient was the sum of all etanercept and adalimumab copayments and coinsurance payments made during the followup period divided by the total days supplied, and multiplied by 7 days. Deductibles were not included in this measure.

Second, the patient's share of the anti-TNFα therapy cost was computed by dividing the sum of the copayments, coinsurance payments, and deductibles by the sum of the amount paid for the drug by the patient and the insurer.

Statistical analyses.

Separate multivariate linear regression models of the MPR were estimated using the weekly OOP costs (model 1) and the patients' shares of therapy costs (model 2). Kaplan-Meier survival curves of the probability of persistence were generated to compare patients with varying levels of OOP costs for anti-TNFα therapy. Multivariate proportional hazards models of persistence were estimated in separate specifications using the weekly OOP costs (model 3), patients' shares of therapy costs (model 4), and a binary variable representing weekly OOP costs greater than $50 (model 5). In addition to 1 OOP variable, other variables were tested for inclusion in the models and were included if they had a significance level (α) of at least 0.1. Variables assessed included demographic characteristics (sex, US region, age, and urban versus rural residence [assessed using residence in metropolitan statistical areas]), type of insurance (health maintenance organization [HMO], comprehensive, or preferred provider organization), patient status (insured versus dependent), preexisting comorbidities assessed using ICD-9-CM codes (cardiovascular disease, mental disorders, infections, chronic obstructive pulmonary disease/asthma, cancer, diabetes, osteoporosis/fracture, joint deformity/synovitis/cervical myelopathy, carpal tunnel syndrome, and vasculitis), pre-period utilization variables, and Charlson comorbidity index score (27). Utilization variables assessed during the period 6 months prior to index were prescriptions for disease-modifying antirheumatic drugs (DMARDs), prescriptions for nonsteroidal antiinflammatory drugs (NSAIDs), prescriptions for narcotic analgesics, hospitalizations, and number of visits to a physician. The Charlson score was determined using diagnoses recorded during the full 18-month period, and due to its relationship with individual comorbidities it was not used in the same multivariate models as the individual comorbidities. Adherence was also compared for patients whose insurance plans had different caps on patient expenditures (medical only, no cap, pharmacy and medical, or pharmacy only), but this variable was not used in the multivariate models because it was only available for a portion of the patients.


  1. Top of page
  2. Abstract

Of the 85,812 patients with RA, 7,324 had at least 1 prescription for etanercept or adalimumab; of these, a final cohort of 2,285 had insurance eligibility 6 months prior to and 12 months post index. The study population was 75% female and had a mean ± SD age of 54 ± 12 years. The excluded population was also 75% female but was slightly older, with a mean ± SD age of 58 ± 13 years. Seventy-five percent initiated with etanercept and 25% initiated with adalimumab. Patients paid a mean ± SD of $7.84 ± $14.15 per week for their anti-TNFα therapy (Table 1). Most patients (92%) paid less than $20 OOP per week of therapy, and nearly 13% paid less than $1 per week (Table 2). The mean ± SD proportion of the total anti-TNFα cost paid by the patient was 2.7% ± 4.9%. The mean ± SD MPR was 0.52 ± 0.31 before adjusting for possible confounders.

Table 2. Adherence, out-of-pocket (OOP) costs, and patients' shares of expenditures during their first year of etanercept or adalimumab therapy, 2002–2004*
 Patients, n (%)MPR
  • *

    MPR = medication possession ratio; anti-TNFα = anti–tumor necrosis factor α.

  • Mean MPR (proportion of the 365 followup days covered by the patient's supply of etanercept or adalimumab) across patients.

  • N = 2,285 patients. Dollar amounts are the mean, across patients, of patients' copayments + coinsurance payments × (7/total days supplied) of etanercept or adalimumab.

  • §

    N = 2,283 patients.

Weekly OOP cost level for anti-TNFα, dollars  
  0 to <1292 (12.8)0.56
  >1 to 2276 (12.1)0.58
  >2 to 3300 (13.1)0.61
  >3 to 4336 (14.7)0.54
  >4 to 5239 (10.5)0.54
  >5 to 10494 (21.6)0.45
  >10 to 20167 (7.3)0.51
  >20 to 3067 (2.9)0.42
  >30 to 4012 (0.5)0.55
  >40 to 5015 (0.7)0.38
  >50 to 6044 (1.9)0.40
  >60 to 7015 (0.7)0.29
  >70 to 808 (0.4)0.41
  >8020 (0.9)0.22
Patient's share of anti-TNFα expenditures, %§  
  0 to <1782 (34.25)0.57
  1 to <2797 (34.91)0.55
  2 to <3289 (12.66)0.42
  3 to <4144 (6.31)0.42
  4 to <547 (2.06)0.63
  5 to <10112 (4.91)0.45
  10 to <1517 (0.74)0.52
  15 to <2021 (0.92)0.34
  20 to <2551 (2.23)0.40
  25 to <308 (0.35)0.27
  30 to <354 (0.18)0.30
  35 to <405 (0.22)0.17
  40 to <455 (0.22)0.29
  45 to <501 (0.04)0.46


Table 2 indicates that without adjustment for possible confounders, higher weekly OOP costs appear to be associated with lower levels of adherence. The multivariate models of adherence confirmed that after adjusting for possible confounders, adherence was significantly lower when the weekly OOP cost or the patient's share of therapy cost was higher (Table 3). Converting MPR to therapy days (MPR × 365) shows how many days of adherence are lost when OOP costs rise. A week of therapy was lost with either an increase of $5.50 in weekly OOP cost or an increase of 2.2 percentage points in the share of anti-TNFα cost paid for by patients. The other variables that were significantly associated with lower adherence were female sex and HMO insurance. Variables associated with higher adherence were the use of DMARDs prior to anti-TNFα therapy initiation and residence in the northeastern US (Table 3).

Table 3. Multivariate linear regression models of effect of out-of-pocket (OOP) costs on adherence measured by medication possession ratio*
Independent variablesModel 1 (n = 2,285)Model 2 (n = 2,283)
Parameter estimatePParameter estimateP
  • *

    Anti-TNFα = anti–tumor necrosis factor α; DMARDs = disease-modifying antirheumatic drugs; HMO = health maintenance organization.

  • Uses patient's weekly OOP costs for anti-TNFα. Patients copayments + coinsurance payments × (7/total days supplied) of anti-TNFα agents.

  • Uses patient-paid proportion of anti-TNFα cost. Patients copayments + coinsurance payments + deductible/total charges for anti-TNFα agents.

Weekly OOP for anti-TNFα agent−0.0035< 0.0001
Proportion of anti-TNFα cost paid by patient−0.8794< 0.0001
Prescriptions for DMARDs 6 months prior0.03630.00420.03470.0064
Northeastern US region0.06380.00310.06590.0023
HMO insurance−0.0909< 0.0001−0.08710.0001


Persistence analyses, which used the same 1-year followup period as adherence analyses, showed similar results. Univariate Kaplan-Meier curves illustrated that persistence was markedly better for patients with OOP costs below $50 for weekly therapy compared with those who paid more than $50 for a week of therapy (Figure 1). The probability of persisting for 1 year was 32% for patients with OOP costs above $50/week compared with 57% for patients with OOP costs less than $50/week. The adjusted hazard ratio (HR; 1.579, P < 0.001) indicates that patients with OOP costs greater than $50/week were 58% more likely to stop therapy than those with lower OOP costs (Table 4, model 5). Also, patients were 8% more likely to stop therapy for every $10 increase in weekly OOP costs (HR 1.008, P < 0.001) (Table 4, model 3). An increase of 1 percentage point in the patient's share of the therapy cost increased the probability of therapy discontinuation by 2.7% (HR 1.027, P < 0.001) (Table 4, model 4). Other variables that led to worse persistence were a higher Charlson score and a previous prescription for a narcotic analgesic 6 months prior to therapy initiation (Table 4). Persistence was better in older patients.

thumbnail image

Figure 1. Univariate Kaplan-Meier curves illustrating persistence on etanercept/adalimumab for patients with out-of-pocket payments over versus under $50 per week. 95% CI = 95% confidence interval.

Download figure to PowerPoint

Table 4. Multivariate hazard ratios for therapy discontinuation estimated using proportional hazards regression*
 Model 3 (n = 2,284)Model 4 (n = 2,282)Model 5§ (n = 2,284)
  • *

    HR = hazard ratio; OOP = out-of-pocket; anti-TNFα = anti–tumor necrosis factor α.

  • The independent variable for OOP cost is weekly OOP: patients' copayments + coinsurance payments × (7/total days supplied) of anti-TNFα agents.

  • The independent variable for OOP cost is proportion of therapy cost paid by the patient: patients' copayments + coinsurance payments + deductible/total charges for anti-TNFα agents.

  • §

    The independent variable for OOP cost is whether OOP cost is above $50 per week versus below $50 week.

Weekly OOP costs for anti-TNFα agent1.008< 0.001
Proportion of anti-TNFα cost paid by patient1.027< 0.001
OOP costs for anti-TNFα exceeds $50/week1.579< 0.001
Age, years0.9960.0410.9960.0400.9960.034
Prescriptions for narcotic analgesics 6 months prior1.0720.1041.0740.0941.0710.106
Charlson comorbidity index1.0700.0021.0710.0021.0670.003


  1. Top of page
  2. Abstract

We examined the impact of patient OOP payments on compliance with biologic therapy for RA and found a significant relationship. Whether OOP costs were measured using the price of therapy or the patient's share of the overall cost of therapy, higher OOP costs significantly reduced adherence and led to earlier therapy discontinuation than would otherwise be expected. The probability that a patient would still be maintaining therapy at the end of a 1-year followup period dropped from 57% for patients with OOP costs below $50/week to 32% for patients with OOP costs above $50/week. Clinicians may be unaware that patients are not adhering to therapy, leading to inaccurate assessments of effectiveness. Additionally, lower-than-effective dosing or premature discontinuation of therapy may lead to poorer radiographic and functional outcomes. Poor disease control may lead to increased disability, disease-related surgery such as joint replacement, and increased overall health resource utilization.

Because only 8% of the patients paid over $20 per week for therapy, our analysis had a limited ability to estimate the effect of very high OOP costs (above $20/week) on compliance. However, economic theory dictates that demand is generally more elastic at higher prices, which means that OOP costs higher than those in our sample could generate larger affects on adherence. The significance of our persistence results indicates that the poor adherence is mostly due to therapy discontinuation rather than patients skipping doses. Although we did not investigate this in detail, we looked at the claims for the subset of patients who paid the most, those with OOP costs above $60/week (n = 43). Of these, 17 patients discontinued after 1 or 2 prescriptions, while 26 patients had extended periods of time between their prescriptions that were not covered by their supply. This indicates that high OOP costs may be associated with both reduced persistence and dose skipping. However, it is not possible to assess dose skipping in this data set.

The use of claims data, as in this study, is subject to some limitations. In particular, we do not have patient-level measurements of the effectiveness of the medications or of the side effects that could cause patients to discontinue treatment. Also, some variables that have been shown to affect compliance with RA medications cannot be measured using claims data. For example, a European prospective study (28) found that compliance with RA medications was higher in patients who reported more satisfactory contacts with health care professionals, more education, and higher disability levels. The study also reported higher compliance among older patients and women. Our result of improved persistence with age is consistent with this, although we did find that women were less adherent than men. Although claims data cannot measure compliance as precisely as do studies that use electronic monitoring of prescription-use behavior, a study that monitored patients with RA, gout, and polymyalgia rheumatica (29) found that most of the difference in compliance can be explained by the patient's coping style, sex, dosing frequency, and overall health. Our study found that increased comorbidities, proxied by Charlson score, increased the likelihood of therapy discontinuation but did not significantly affect adherence.

Despite the noted limitations of claims data, these data are particularly useful in measuring a patient's OOP payments because costs for which patients are responsible are specifically tracked as part of the claims process. Three other studies of compliance or dosing patterns of biologics for RA have been conducted recently using these data (21, 30, 31). For example, Harley et al (30) studied compliance with etanercept, methotrexate, and infliximab using patient enrollment and followup criteria similar to those in our study. They defined patients as compliant if they took at least 80% of the expected administrations of therapy, and they found that ∼68% of the study patients were compliant with etanercept. Their result translates roughly into an MPR of 0.54 (0.68 × 0.80), which is very close to the average MPR of 0.52 in our study.

Our results showing that higher OOP payments were associated with reduced compliance are corroborated by 3 studies that used US managed-care claims data to study other chronic diseases. In particular, higher patient OOP costs were shown to be associated with lower adherence to statins (23), and Medicare patients whose pharmacy benefits were limited to $1,000 had lower adherence with antidiabetics, antihypertensives, and lipid-lowering medications than did patients who had supplemental drug insurance (32). In another study, doubling copayments was associated with reductions in the taking of all 8 therapeutic classes of drugs studied: NSAIDs, antihistamines, antihyperlipidemics, antiulcerants, antiasthmatics, antihypertensives, antidepressants, and antidiabetics (1).

Anti-TNFα therapy has been shown to control inflammation, inhibit progression of radiographic damage, and prevent loss of function when given at the approved dose and dosing interval (9–11). Poor adherence and/or premature discontinuation of effective therapy may lead to poorer outcomes. These outcomes may be associated with increased overall health care costs. Our data, which demonstrate a negative association of adherence and persistence with anti-TNFα therapy and OOP costs, suggest that increased cost sharing by patients could have the effect of increasing health care costs in the long term.


  1. Top of page
  2. Abstract

The study was designed by Amgen and funded by Immunex Corporation, a wholly owned subsidiary of Amgen. The Amgen-affiliated authors were involved in the design of this study and in the hypotheses set forth; thus, they were involved in discussion of potential analyses prior to conducting the study. Study sponsors were not involved in the analysis or interpretation of the data, which remained the responsibility of other, non–Amgen-affiliated authors. The Amgen-affiliated authors were involved in the development of a manuscript outline and critical review of the manuscript. Amgen assisted with the writing of the manuscript. All others contributed intellectually to the content of the manuscript, had full access to the data, and vouch for the completeness and accuracy of the data and data analyses. The Amgen-affiliated and non–Amgen-affiliated authors made collaborative decisions about submission for publication.


  1. Top of page
  2. Abstract

Dr. Curkendall had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Curkendall, Patel, Gleeson, Zagari, Dubois.

Acquisition of data. Zagari.

Analysis and interpretation of data. Curkendall, Patel, Gleeson, Zagari, Dubois.

Manuscript preparation. Curkendall, Campbell, Zagari, Dubois.

Statistical analysis. Curkendall, Gleeson.


  1. Top of page
  2. Abstract
  • 1
    Goldman DP, Joyce GF, Escarce JJ, Pace JE, Solomon MD, Laouri M, et al. Pharmacy benefits and the use of drugs by the chronically ill. JAMA 2004; 291: 234450.
  • 2
    Fairman KA, Motheral BR, Henderson RR. Retrospective, long-term followup study of the effect of a three-tier prescription drug copayment system on pharmaceutical and other medical utilization and costs. Clin Ther 2003; 25: 314761.
  • 3
    Motheral B, Fairman KA. Effect of a three-tier prescription copay on pharmaceutical and other medical utilization. Med Care 2001; 39: 1293304.
  • 4
    Motheral BR, Henderson R. The effect of a copay increase on pharmaceutical utilization, expenditures, and treatment continuation. Am J Manag Care 1999; 5: 138394.
  • 5
    Huskamp HA, Deverka PA, Epstein AM, Epstein RS, McGuigan KA, Frank RG. The effect of incentive-based formularies on prescription-drug utilization and spending. N Engl J Med 2003; 349: 222432.
  • 6
    Brixner DI, Joish VN, Odera GM, Avey SG, Hanson DM, Cannon HE. Effects of benefit design change across 5 disease states. Am J Manag Care 2007; 13: 3706.
  • 7
    Landsman PB, Yu W, Liu X, Teutsch SM, Berger ML. Impact of 3-tier pharmacy benefit design and increased consumer cost-sharing on drug utilization. Am J Manag Care 2005; 11: 6218.
  • 8
    Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing: associations with medication and medical utilization and spending and health. JAMA 2007; 298: 619.
  • 9
    American College of Rheumatology Subcommittee on Rheumatoid Arthritis Guidelines. Guidelines for the management of rheumatoid arthritis: 2002 update. Arthritis Rheum 2002; 46: 32846.
  • 10
    Weinblatt ME, Keystone EC, Furst DE, Moreland LW, Weisman MH, Birbara CA, et al. Adalimumab, a fully human anti–tumor necrosis factor α monoclonal antibody, for the treatment of rheumatoid arthritis in patients taking concomitant methotrexate: the ARMADA trial. Arthritis Rheum 2003; 48: 3545.
  • 11
    Haraoui B. The anti-tumor necrosis factor agents are a major advance in the treatment of rheumatoid arthritis. J Rheumatol Suppl 2005; 72: 467.
  • 12
    Bathon JM, Martin RW, Fleischmann RM, Tesser JR, Schiff MH, Keystone EC, et al. A comparison of etanercept and methotrexate in patients with early rheumatoid arthritis. N Engl J Med 2000; 343: 158693.
  • 13
    Klareskog L, van der Heijde D, de Jager JP, Gough A, Kalden J, Malaise M, et al. Therapeutic effect of the combination of etanercept and methotrexate compared with each treatment alone in patients with rheumatoid arthritis: double-blind randomised controlled trial. Lancet 2004; 363: 67581.
  • 14
    Smolen JS, Han C, van der Heijde D, Emery P, Bathon JM, Keystone E, et al. Infliximab treatment maintains employability in patients with early rheumatoid arthritis. Arthritis Rheum 2006; 54: 71622.
  • 15
    Yelin E, Trupin L, Katz P, Lubeck D, Rush S, Wanke L. Association between etanercept use and employment outcomes among patients with rheumatoid arthritis. Arthritis Rheum 2003; 48: 304654.
  • 16
    Kobelt G, Eberhardt K, Geborek P. TNF inhibitors in the treatment of rheumatoid arthritis in clinical practice: costs and outcomes in a follow up study of patients with RA treated with etanercept or infliximab in Southern Sweden. Ann Rheum Dis 2004; 63: 410.
  • 17
    Farahani P, Levine M, Gaebel K, Wang EC, Khalidi N. Community-based evaluation of etanercept in patients with rheumatoid arthritis. J Rheumatol 2006; 33: 66570.
  • 18
    Bresnihan B. Effects of anakinra on clinical and radiological outcomes in rheumatoid arthritis. Ann Rheum Dis 2002; 61 Suppl 2: ii747.
  • 19
    Paulus HE, Weaver AL, Yu EB, Woolley JM, Xia HA, Louie J. Employed rheumatoid arthritis patients experienced decreased work loss on etanercept treatment. Poster presented at the Annual European Congress of Rheumatology; 2005 June 8–11; Vienna, Austria. Vienna: EULAR; 2005.
  • 20
    Dor A, Encinosa W. Does cost sharing affect compliance? The case of prescription drugs. NBER working paper no. 10738. 2004. URL:
  • 21
    Goldman DP, Joyce GF, Lawless G, Crown WH, Willey V. Benefit design and specialty drug use. Health Aff (Millwood) 2006; 25: 131931.
  • 22
    Soumerai SB, Pierre-Jacques M, Zhang F, Ross-Degnan D, Adams AS, Gurwitz J, et al. Cost-related medication nonadherence among elderly and disabled Medicare beneficiaries: a national survey 1 year before the Medicare drug benefit. Arch Intern Med 2006; 166: 182935.
  • 23
    Gibson TB, Mark TL, McGuigan KA, Axelsen K, Wang S. The effects of prescription drug copayments on statin adherence. Am J Manag Care 2006; 12: 50917.
  • 24
    Ozminkowski RJ, Marder WD, Hawkins K, Wang S, Stallings SC, Finkelstein SN, et al. The use of disease-modifying new drugs for multiple sclerosis treatment in private-sector health plans. Clin Ther 2004; 26: 134154.
  • 25
    Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother 2006; 40: 12808.
  • 26
    Physicians' desk reference. Montvale (NJ): Thomson Healthcare; 2004.
  • 27
    Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992; 45: 6139.
  • 28
    Viller F, Guillemin F, Briancon S, Moum T, Suurmeijer T, van den Heuvel W. Compliance to drug treatment of patients with rheumatoid arthritis: a 3 year longitudinal study. J Rheumatol 1999; 26: 211422.
  • 29
    De Klerk E, van der Heijde D, Landewe R, van der Tempel H, Urquhart J, van der Linden S. Patient compliance in rheumatoid arthritis, polymyalgia rheumatica, and gout [published erratum appears in J Rheumatol 2003;30:423]. J Rheumatol 2003; 30: 4454.
  • 30
    Harley CR, Frytak JR, Tandon N. Treatment compliance and dosage administration among rheumatoid arthritis patients receiving infliximab, etanercept, or methotrexate. Am J Manag Care 2003; 9 Suppl 6: S13643.
  • 31
    Gilbert TD Jr, Smith D, Ollendorf DA. Patterns of use, dosing, and economic impact of biologic agent use in patients with rheumatoid arthritis: a retrospective cohort study. BMC Musculoskelet Disord 2004; 5: 36.
  • 32
    Hsu J, Price M, Huang J, Brand R, Fung V, Hui R, et al. Unintended consequences of caps on Medicare drug benefits. N Engl J Med 2006; 354: 234959.