• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

The Impact of Osteoporosis Medication Beliefs and Side-Effect Experiences on Non-Adherence to Oral Bisphosphonates*

Posted on: Saturday, 26 January 2008, 06:00 CST

By McHorney, Colleen A Schousboe, John T; Cline, Richard R; Weiss, Thomas W

Key words: Adherence - Bisphosphonates - Health beliefs - Osteoporosis - Side effects - Treatment beliefs ABSTRACT

Objective: Non-adherence to oral bisphosphonate medications Is a pervasive problem that blunts their potential to prevent fractures. Using multivariate modeling, we assessed the unique contribution of six classes of variables as drivers of non-adherence to bisphosphonates: (1) beliefs about osteoporosis and its prescription drug treatment, (2) ratings of the affordability of the prescription osteoporosis medications, (3) evaluations of the convenience of the bisphosphonate dosing frequency, (4) reports of troublesome side effects, (5) ratings of aspects of the bisphosphonate dosing regimen, and (6) risk factors for fracture. These categories of predictor variables were selected for investigation because they have been suggested by clinical-trial, survey, and observational studies in osteoporosis as reasons for non-adherence among patients taking prescription osteoporosis therapy.

Methods: Women aged 45 or older who filled a prescription for an oral bisphosphonate in January or February of 2006 were identified through a dispensing database of 3300 US retail pharmacies. Subjects received a mailed pre-notification letter from the retail pharmacy chain informing them that someone would be calling them to invite them to participate in a telephone survey about osteoporosis medications. Trained interviewers used a standardized telephone script to recruit patients. Our definition of adherence was provisionally based on database records across a 7-month period and then cross-validated using patient self-report during the telephone recruitment. We measured beliefs regarding bisphosphonate effectiveness and safety, osteoporosis health concerns, concerns regarding drug costs, dosing frequency convenience, and experienced side effects using multi-item scales. Data were collected by telephone interview. Bivariate analyses were conducted using chi^sup 2^ and t-tests, and multivariate analyses were conducted using logistic regression.

Results: Of the 3274 women contacted for study participation, 1092 (33%) completed the interview and 1015 had analyzable data. Multivariate analyses showed that those most symptomatic in terms of side effects and those with the most skeptical beliefs in drug effectiveness and drug safety had odds ratios for non-adherence of 6.78 (95% CI 4.67-9.86), 5.70 (95% CI 3.65-8.92), and 2.26 (95% CI 1.49-3.42), respectively. In multivariate models, osteoporosis health concerns, dosing frequency convenience, and concerns regarding medication costs were not statistically associated with non-adherence to bisphosphonate therapy.

Conclusions:The experience of troublesome side effects and patient beliefs regarding the effectiveness and safety of oral bisphosphonate medications prescribed for them are strongly associated with bisphosphonate non-adherence. Improving adherence to oral bisphosphonates may require providers to solicit and address patients' medication beliefs and to proactively address side effects. Limitations of our study include: (1) the study sample is not likely to be a national random sample of bisphosphonate users, and (2) some evidence of non-response bias was observed.

Introduction

Osteoporotic fractures constitute a major public health problem1. Among US Caucasian women and men age 50 or older, 50 and 25%, respectively, can expect to have a fracture of the hip, wrist, or a vertebra sometime during their remaining lifetime1. These fractures can result in serious long-term health consequences. For example, only one-third of those who suffer a hip fracture recover their pre- fracture ability to ambulate2,3, and a substantial proportion of those who suffer vertebral fractures will have height loss and/or chronic back pain3-5.

Medications are now available that can reduce the risk of incident clinical fractures at all skeletal sites from minor to moderate trauma by 30-50% and the risk of multiple vertebral fractures by up to 90%6. Guidelines promulgated by medical subspecialties7,8, as well as professional societies concerned specifically with osteoporosis9,10, are in agreement that medications to reduce fracture risk are indicated in those with osteoporosis by bone density criteria, with a prevalent radiographic vertebral fracture, or fractures from minor or moderate trauma.

Adherence refers to the extent to which patients conform to treatment recommendations prescribed by healthcare providers. Unfortunately, adherence to prescription osteoporosis medications in clinical practice is, on average, suboptimal as only 50% of patients are still on their medication 1 year after the start of tiierapy11- 13. Suboptimal adherence is associated with greater rates of bone loss, increased fracture rates, and increased hospitalization rates and healthcare costs14-22.

Research to date on adherence to pharmacologic fracture prevention therapy has relied overwhelmingly upon administrative databases or open-label randomized studies, both of which are severely limited in the breadth and depth of explanatory variables. With their large sample sizes, these studies have been able to describe and precisely quantify the extent and time course of non- adherence, but they shed little light on why non-adherence occurs beyond the effect of basic demographic variables and crude indices for disease burden or comorbidity.

Clinical-trial, survey, and observational studies in osteoporosis have suggested various reasons for non-adherence among patients taking prescription osteoporosis therapy, including the experience, or fear, of troublesome side effects23-37; out-of-pocket drug costs25,27,29,31,35, practical difficulties associated with the bisphosphonate regimen23,28,30,36, and inconvenient dosing frequency25,28,30,35. However, much of this extant research has been largely descriptive in nature. Little is known about the unique contribution of each of these putative adherence drivers in the context of multivariate prediction.

Moreover, both qualitative38-40 and quantitative41-51 research in chronic disease have suggested that beliefs about medications, specifically regarding their effectiveness and long-term safety, may be important predictors of medication-use behavior. The extent to which an osteoporosis medication is actually altering bone density is generally not known until follow-up bone density testing is performed, which can often be 1-2 years after prescription therapy is initiated (or even longer). Hence, sustaining a patient's belief that a medication to prevent fractures is effective can be challenging, given that feedback regarding clinical effectiveness occurs infrequently. Limited research conducted to date in osteoporosis lends support to the hypothesis that beliefs about drug effectiveness and concerns about medications influence adherence28,30,35,37. Other research in osteoporosis has shown that patient beliefs about susceptibility to osteoporosis and the perceived seriousness of osteoporosis are related to initiating prescription osteoporosis therapy5253 and engaging in osteoporosis self-care behaviors54.

It is well recognized that the adherence dilemma is multidimensional in nature55-57, and that numerous barriers to adherence exist at the patient, provider, and system level55,58-61. We designed a study to assess the unique contributions of a number of patient-level drivers of adherence within the context of multivariate modeling. Because health beliefs and medication beliefs have been demonstrated to be predictive of adherence to prescription medications in diseases other than osteoporosis41-51, we developed multi-item scales to measure patient beliefs about the perceived effectiveness of prescription osteoporosis therapy, the perceived safety of prescription osteoporosis therapy, and the salience of osteoporosis as a health concern. To corroborate whether other potential adherence drivers (suggested largely by descriptive research in osteoporosis) independently predict adherence, we also measured the affordability of the prescription osteoporosis medications, the convenience of the bisphosphonate dosing frequency, experience with troublesome side effects, aspects of the bisphosphonate dosing regimen, and risk factors for fracture. We used multivariate modeling to evaluate the unique contribution of each predictor variable on non-adherence to oral bisphosphonate therapy.

Patients and methods

Study cohort

To conduct this study, we contracted with a national retail pharmacy chain with over 3300 retail stores in 28 states. Potential participants were women identified by their retail pharmacy records based on their receipt of one of three types of marketed oral bisphosphonates: weekly alendronate (Fosamax 35 mg or 70 mg, Merck & Co., Inc., Whitehouse Station, NJ), weekly risedronate (Actonel 35 mg, Procter & Gamble Pharmaceuticals, Cincinnati, OH), or monthly ibandronate (Boniva 150 mg, Roche Therapeutics, Inc., Nutley, NJ). These medications were identified in the database based on their National Drug Codes (NDC). Participants also had to have prescriptions that required them to visit a retail pharmacy approximately once a month to obtain their osteoporosis medication.

The eligible population for this study was obtained from two contiguous time samples. The two samples were needed in order to have a large enough bolus of women to achieve our desired sample size after accounting for non-locatables, ineligibles, and refusals. The January 2006 sample consisted of women aged 45 or older who filled an index prescription in January 2006 for a weekly or monthly bisphosphonate and had no recorded prescription fills for a bisphosphonate during a 3-month look-back period (October 1, 2005 to December 31, 2005). The February 2006 sample consisted of women aged 45 or older who filled an index prescription in February 2006 for a weekly or monthly bisphosphonate and had no recorded prescription fills from November 1, 2005 to January 31, 2006 for a bisphosphonate. The February 2006 sample was intended a priori as a back-up if the desired sample size was not obtained with the January 2006 sample. The 3-month look-back period allowed us some level of confidence that patients were starting a new prescription. Patients who had a prior history of prescription osteoporosis medications, no matter how distant before the look-back period, were allowed into the sample. Hence, we did not try to establish, either using the pharmacy records or through the interview process, each patient's history of osteoporosis medication usage and whether each patient was completely naive to osteoporosis medications before their defined index prescription. Women were ineligible for study if: (1) they were younger than age 45, (2) their index prescription was for a daily oral bisphosphonate or any other medication used to treat osteoporosis, (3) their osteoporosis medication prescription was for a 60- or 90-day supply, or (4) they were not members of the January or February samples as defined above. Definition of adherence and non-adherence

Each patient was defined as either adherent or non-adherent to their oral bisphosphonate based on a two-step procedure. The first step involved a provisional classification of each patient based on electronic refill records over a 7-month period. A central database that captured all retail pharmacy records for all of the chain's stores was used for the provisional classification. The second step involved verification of the provisional classification with each patient in the course of study recruitment on the telephone.

Each eligible participant's electronic pharmacy records were reviewed over a 7-month period (January 2006 to July 2006 for the January sample and February 2006 to August 2006 for the February sample) to provisionally classify them as either bisphosphonate adherers or non-adherers (i.e., database evidence of refilling their oral bisphosphonate). January 2007 was chosen as the first month in the sampling time window because patients frequently change health plans, and hence drug benefits, before the start of each new calendar year. A 7-month observation period was selected because this provided us with the most current retail data available (i.e., July/August 2006) at the time the actual survey process began (September 2006).

Women were provisionally classified as adherers if they had >/= 140 total days' supply of their bisphosphonate across the 7-month period. The total days' supply was based on the number of prescriptions during the 7-month period and the days' supply of each prescription. For example, patients who filled five prescriptions, each with a 28-day supply (i.e., four weekly pills), would have 140 total days' supply (5 x 28 days) during the observation period and be provisionally classified as adherers. Participants were provisionally classified as non-adherers if they had < 140 days' supply of their bisphosphonate across the 7-month period. For example, patients who filled three prescriptions, each with a 28- day supply, would have 84 total days' supply (3 x 28 days) available during the observation period and be provisionally classified as non-adherers.

The final determination of bisphosphonate adherence status was made while recruiting participants for study participation. Patients received a mailed pre-notification letter from the retail pharmacy chain informing them that someone would be calling them to invite them to participate in a telephone survey about osteoporosis medications. Trained interviewers used a standardized telephone script to recruit patients. Telephone recruitment involved: (1) screening patients for eligibility, and (2) inviting patients to complete the survey if eligible. Potential participants were asked during the telephone screen: (1) if they were still taking their index prescription, (2) if not, did they take at least one dose of the index prescription (to remove from study eligibility persons who filled the index prescription but did not take a single dose), and (3) did they change prescriptions during the observation period between weekly and monthly formulations. Participants who reported continuing to take their oral bisphosphonate were classified as adherers. Participants who reported that they were no longer taking their oral bisphosphonate were classified as non-adherers. Participants who switched between weekly oral bisphosphonates (alendronate to risedronate or vice versa) were classified as adherers as long as they reported taking the weekly bisphosphonate at the time of the interview. The ten patients who switched from a weekly bisphosphonate to oral ibandronate and the 13 patients who switched from oral ibandronate to a weekly bisphosphonate were classified as adherers as long as they reported taking the oral bisphosphonate at the time of the interview. Additional detail on the survey administration process is described below.

Survey design

Our research aims were to evaluate the unique contribution of six classes of variables as drivers of non-adherence to bisphosphonates: (1) beliefs about osteoporosis and its prescription drug treatment, (2) ratings of the affordability of the prescription osteoporosis medications, (3) evaluations of the convenience of the bisphosphonate dosing frequency, (4) reports of troublesome side effects, (5) ratings of aspects of the bisphosphonate dosing regimen, and (6) risk factors for fracture. To inform the survey development process, we conducted three phases of research. First, we conducted ten focus groups in two major metropolitan areas with recent adherers and non-adherers of bisphosphonates. The purpose of this qualitative research was to identify patient-centered reasons for treatment adherence and non-adherence and to identify salient aspects of the bisphosphonate regimen that might influence adherence. Second, we conducted a comprehensive review of the literature on adherence to prescription osteoporosis medications. In that literature, we identified several patient-centered determinants of adherence, including experience or fear of side effects23-37, practical difficulties with the bisphosphonate dosing procedure23,28,30,36, inconvenient dosing frequency25,28,30,35, medication costs25,27,29,31,35, skepticism about drug efficacy/ treatment benefit28,35, concerns about drug safety26,35, lack of motivation35, and general medication concerns (e.g., concerns about taking bisphosphonates for a long period of time)30,37. Third, we reviewed the literature on drivers of adherence to prescription medications for chronic disease other than osteoporosis to inform our measurement efforts.

Multi-item scale development

Since we selected the telephone as the survey mode of administration, it was vital to employ a survey with minimal respondent burden while simultaneously measuring an array of putative drivers of non-adherence. We thoroughly reviewed the peer- reviewed literature for extant measure that would be: (1) appropriate in terms of content (i.e., exhibit content validity vis- a-vis our objectives), and (2) psychometrically sound. We identified few measures that sufficiently met both of these criteria. Thus, we constructed multi-item scales, using well-accepted principles of psychometric theory62-67, to tap a number of the putative drivers of non-adherence. We constructed multi-item scales measuring drug efficacy beliefs specific to osteoporosis (five items), drug safety beliefs specific to osteoporosis (three items), and the salience of osteoporosis as a health concern (five items). We defined drug safety as patient subjective beliefs about possible undue harm resulting from pharmacologic therapy. We used three items to construct a multi-item scale assessing the affordability of prescription medications for osteoporosis. We defined side effects as actually experienced adverse symptoms resulting from oral bisphosphonate therapy. We used two items to construct a multi-item scale assessing the evaluation of side effects of bisphosphonate therapy (unpleasant side effects and side effects a problem for me). We used three items to construct a multi-item scale assessing dosing frequency convenience. Whenever possible, items for these scales were derived or adapted from existing surveys24,28,39,40,52,68-72.

Osteoporosis risk factors

Risk factors for osteoporosis and/or fracture are objective experiences that may motivate women to adhere to their medications. We measured seven risk factors using a yes/no format for each item: osteoporosis diagnosis status (osteoporosis, osteopenia, or normal bone density), previous fracture, any female relative with a history of osteoporosis, self-reported dowager's hump, self-reported height loss from peak height, current weight less than 127 pounds (57.6kg), and self-reported fall in the past year.

Ratings of the bisphosphonate dosing regimen

To assess perceived practical difficulties with the bisphosphonate regimen, we asked respondents to rate how problematic (1 = not a problem at all, 2 = a little of a problem, 3 = somewhat of a problem, and 4 = very much of a problem) five aspects of the bisphosphonate regimen were to them: (1) remembering to take it, (2) staying upright for a specified amount of time, (3) not being able to eat or drink for a specified amount of time, (4) interference with taking other medications, and (5) interference with normal daily activities. Each of the five items was treated analytically as single-item measures. Reasons for adherence and non-adherence

The final set of items in the survey asked respondents to rate the importance (1 = extremely, 2 = very, 3 = somewhat, 4 = a little, 5 = not at all) of six osteoporosis medication attributes as reasons for their adherence or non-adherence. The six attributes were: (1) side effects (phrased as 'absence of side effects' for adherers and 'presence of side effects' for non-adherers), (2) out-of-pocket costs for the drug, (3) the dosing frequency, (4) the drug's effectiveness, (5) the procedure for taking the drug, and (6) drug interactions with other prescription medications (phrased as 'absence of for adherers and 'presence of for non-adherers). Each of these items was treated analytically as single-item measures.

Demographic characteristics

We asked respondents to self-report the following demographic characteristics: age, race, years of formal education achieved, and marital status. In addition, we asked the excellent-to-poor, self- rated health item ('In general, how would you say your health is: excellent, very good, good, fair, poor?'), and we asked respondents to report the total number of prescription medications they currently take ('How many different prescription medications do you currently take?').

Hypotheses

Based on previous research23-37, we hypothesized that the experience of side effects would be the principal driver of bisphosphonate non-adherence. Accordingly, we posited a large effect size (i.e., > 0.80) for side effects. The literature guided us to hypothesize that beliefs about drug effectiveness and drug safety would influence non-adherence26,28,30,35,43,52,73-75, although not as strongly as side effects. Thus, we posited moderate effect sizes (i.e., 0.50-0.80) for drug effectiveness and drug safety beliefs. We also believed that medication affordability concerns25,29,3135,76,77 and osteoporosis health concerns27,28,40,52,75 would modestly influence adherence. We hypothesized a small effect size (i.e., 0.15- 0.30) for ratings of dosing frequency convenience on treatment non- adherence because: (1) research has shown rather modest improvements in adherence with weekly versus daily bisphosphonate dosing11, (2) recent research, also using data from 2006, suggests comparable persistence for weekly versus monthly formulations of bisphosphonates78-80, and (3) patient preference studies in osteoporosis have shown dosing frequency to be a minor determinant of treatment preference72,81,82. We had no specific hypotheses for our demographic variables, but we posited a small effect size (i.e., 0.15-0.30) for osteoporosis risk factors.

Multi-item scaling procedure

Each multi-item scale (drug effectiveness beliefs, drug safety beliefs, osteoporosis health concerns, affordability, dosing frequency convenience, and side effects) was constructed using Likert's methods of summated ratings83, which equally weights items and sums them into an overall score. All scales were linearly transformed to a 0-100 metric, with 100 indicating the most favorable state/attitude, 0 the least favorable state/attitude, and scores in between representing the percentage of the total possible score achieved. As an example, three items were used to assess perceived drug safety, and each item had the same 5-point, strongly disagree (1) to strongly agree (5) rating scale. The raw score could range from 3 (strongly disagree response to each item) to 15 (strongly agree response to each item). The first step in Likert scaling is to sum the items to create a raw summated score. A response of 'agree' (4) to all three items would receive a summated raw score of 12 (4+4+4). The final step is to convert the raw scale (in this example, the 3-15 raw scale) to a 0-100 metric, which is a simple linear transformation.

Sample size

Our research aim was to compare bisphosphonate adherers to non- adherers and to predict bisphosphonates non-adherence. However, we powered our study a priori to accommodate subgroup analysis. Specifically, we powered our study to compare weekly bisphosphonate adherers to monthly bisphosphonate adherers and to compare weekly bisphosphonate non-adherers to monthly bisphosphonate non-adherers. The parameters for our sample size calculations were for an 80% power (using a two-tailed test) to detect a group difference of 5 points on a 0-100 multi-item scale with a standard deviation of 19. Using these parameters, we determined that we would need 513 adherers (342 weekly and 171 monthly) and 513 non-adherers (342 weekly and 171 monthly), for a total minimum sample size of 1026. We collected 1092 questionnaires and 1015 were usable,

Survey administration

Prior to fielding the survey, we pretested it among a sample of eight women eligible for the study to assess potential logic and flow problems and any issues pertaining to reading level. It was not necessary to pretest the survey with a larger sample because negligible problems were uncovered during the pretest. The final printed survey contained 62 items, was seven pages in length, and took approximately 15 minutes to complete by telephone administration.

Pre-notification invitation letters were sent to potential participants. Participants were notified both in the pre- notification letter and during the telephone screening that they would receive a modest retail gift card for participation. Telephone calls were made to households Monday through Friday for four consecutive days or until the eligible participant was contacted, whichever came first. If the eligible participant was not contacted after 4 days, she was deemed as an 'unknown eligible' per conventional survey disposition standards84. Subject recruitment began with the January 2006 sample and proceeded through the entire sample before accessing the February 2006 sample. Subject recruitment aimed to obtain desired sample sizes for four groups: weekly adherers, weekly non-adherers, monthly adherers, and monthly non-adherers. Once the sample size was reached for any of the four groups, recruitment was closed for that group. If members of the January or February 2006 sample were not required because sample quotas had been met, they were deemed ineligible as per conventional survey disposition standards84. All surveys were conducted between September 28, 2006 and November 8, 2006. Telephone surveys were conducted by the national pharmacy chain to ensure patient privacy, and all data were de-identified prior to analysis. During the survey administration, respondents had the opportunity to speak with a clinical pharmacist regarding their osteoporosis medication. The protocol and survey were approved by Essex Institutional Review Board, Inc.

Data analysis

The survey response rate was calculated according to criteria set forth by the American Association for Public Opinion Research84. To assess possible nonresponse bias, we compared responders and nonresponders on five variables from the pharmacy database that were available on all persons: age, census region (derived from state of residence), sample (January or February 2006), provisional adherence classification (adherer vs. non-adherer), and index prescription dosing frequency (weekly vs. monthly). We used chi^sup 2^ tests for nominal and categorical variables and Student's t-test for age.

We computed Cronbach's alpha coefficient85 to estimate the internal-consistency reliability of each multi-item scale. To assess the unidimensionality of each multi-item scale (i.e., the extent to which items measure just one thing in common), we compared the ratio of the first to second eigenvalue. A 2:1 or better ratio is supportive of scale unidimensionality, as is having the first eigenvalue exceed 1.086-88. To determine the range defined by each scale, we calculated measures of central tendency and the percentage of the sample achieving the lowest (floor effect) and highest (ceiling effect) possible scores.

The focus of our analyses was to characterize adherers and non- adherers according to their: (1) beliefs about osteoporosis and its prescription drug treatment (drug effectiveness beliefs, drug safety beliefs, and osteoporosis health concerns), (2) ratings of the affordability of the prescription osteoporosis medications, (3) evaluations of the convenience of the bisphosphonate dosing frequency, (4) reports of troublesome side effects, (5) ratings of aspects of the bisphosphonate dosing regimen, and (6) risk factors for fracture. Bivariate comparisons were made using chi^sup 2^ for nominal and categorical variables and t-tests for the interval- level variables. We developed a stepwise logistic regression model to identify predictors of non-adherence. We used a forward stepwise logistic regression with entry and retention criteria set at the 0.01 probability level. To test the monotonicity of the multi-item scales as independent variables, we divided each scale into tertiles and represented the tertiles as dummy variables. The highest tertile on each scale (representing the most favorable 33% of the score distribution) was the reference group. Potential predictor variables included in the stepwise logistic-regression model were the six multi-item scales, the five items assessing bisphosphonate regimen problems, the seven osteoporosis risk factors, and demographic characteristics (age, race, education, marital status, and the excellent-to-poor health status item).

We sought to reinforce the results of the logistic regression using patients' ratings of reasons for adherence or non-adherence. Because two of the six reasons were phrased slightly differently for adherers versus non-adherers, we did not conduct tests of statistical significance between the two groups. Rather, we present these data descriptively for each group. Finally, we subjected the three multi-item scales identified as statistically significant in the logistic models to a k-means cluster analysis to explore the interactive effect of these non-adherence drivers. We computed non- adherence rates for each of the derived clusters. Results

Survey response rate and non-response bias

A total of 11 220 women were potentially eligible for the study. However, 5492 women were deemed ineligible for participation and were excluded from die denominator when calculating the survey response rate. Reasons for ineligibility included: moved and no new telephone number/address (n = 11), deceased (n = 65), non-English speaking (n = 209), wrong telephone number (n = 292), disconnected telephone (n = 938), incorrect pharmacy claims (n = 946), and not needed because quota/sample size was filled (n = 3031).

Of the 5728 eligible women, we collected 1092 responses across 6 weeks of data collection. Direct refusals occurred among 1415 women. Non-interviews (e.g., eligible but requested call back) occurred among 767 women. A total of 2454 women were classified as unknown eligibility because we never made contact with them (n = 94 with busy numbers at every attempt for four consecutive days, n = 752 with answering machine pick up at every attempt for four consecutive days, and n = 1608 with no answer at every attempt for four consecutive days). When persons of unknown eligibility are excluded from the denominator of the survey response rate, the final response rate was 33.0% (1092/3274). Among the 1092 survey responders, we identified 77 respondents with incomplete interviews. These respondents were removed from the substantive data analysis for a final analyzable sample size of 1015.

To assess possible non-response bias, we compared the 1092 responders to the 2182 non-responders. There were no statistically significant differences between the two groups in sample designation (88 and 87% January 2006 sample for responders and non-responders, respectively) and index dosing frequency (69% weekly for both responders and non-responders). There were no statistically significant differences between the two groups in census-region classification (both the four and nine regional classifications). Survey responders were younger than survey non-responders by 1.1 years (66.6 vs. 67.7, p = 0.007). There was a statistically significant difference between the two groups in the percentage of women aged 80 or older (14% for responders vs. 20% for non- responders, p < 0.0001). Survey non-responders were more likely than responders to be provisionally classified on the basis of the pharmacy records as bisphosphonate non-adherers (92 vs. 42%, p < 0.0001).

Sample characteristics

As shown in Table 1, non-adherers were slightly younger than adherers (65.6 vs. 66.9, p = 0.049). The sample was 86% Caucasian. Mean years of education was 13. One half of the sample was married. Mean scores on the self-rated health item were the same for both groups, and just over one-quarter of both groups rated their health as either poor or fair. Adherers took one additional prescription medication than non-adherers (p < 0.001). By definition, all of the adherers took at least one prescription medication compared to 85% of the non-adherers (p < 0.001).

Measurement properties of multi-item scales

As shown in Table 2, Cronbach's alpha for each scale ranged from 0.66 for osteoporosis health concerns to 0.88 for side effects. For all of the multi-item scales, the first eigenvalue was always greater tiian one (range of 1.78 to 2.74), and the ratio of the first-to-second eigenvalue was always greater than 2.25. The lowest observed mean was that for affordability, indicating that respondents expressed the most concerns about affordability compared to the other five domains. The highest observed mean was that for dosing frequency, indicating that respondents expressed the most favorable evaluation of dosing frequency convenience relative to the other five domains. Ceiling effects were the highest for dosing frequency (13.6%), and floor effects were the highest for affordability and side effects (roughly 15% each).

Differences between bisphosphonate non-adherers and adherers

Only two statistically significant differences emerged between non-adherers and adherers in osteoporosis risk factors (Table 3). Adherers were more likely to report a diagnosis of osteoporosis (70 vs. 61%), while non-adherers were more likely to report having osteopenia or normal bone mass. Adherers were more likely to report having experienced any height loss from peak height (58 vs. 45%). As shown in Table 4, none of the five aspects of the bisphosphonate dosing regimen was considered problematic by a large percentage of the sample. The most problematic aspect of the regimen was remembering to take the bisphosphonate. Despite the overall low prevalence of regimen problems, adherers and non-adherers differed in their ratings of problems with remembering to take it, not being able to eat or drink for a specified amount of time, interference with normal daily activities, and interference with taking other medications (all p < 0.05).

Table 1. Sociodemographic characteristics of non-adherers and adherers

Table 2. Properties of multi-item measures

Table 5 presents unadjusted differences between non-adherers and adherers on the multi-item scales. As hypodiesized, the largest difference between the two groups was observed for side effects. On a 0-100 scale, non-adherers had an average score of 39.9 compared to 70.0 for adherers (mean difference of 30.1 points and effect size of 0.97, p < 0.0001). As hypothesized, non-adherers expressed more skepticism than did adherers about drug safety (46.8 vs. 63.7, p < 0.0001) and drug effectiveness (64.5 vs. 76.3, p < 0.0001). Modest differences were observed between non-adherers and adherers on dosing frequency and osteoporosis health concerns, with non- adherers expressing less favorable evaluations of dosing frequency convenience (80.4 vs. 86.5, p < 0.0001) and more osteoporosis health concerns (73.7 vs. 77.6, p < 0.0001).

Multivariate predictors of non-adherence

Table 6 presents the results from the stepwise, multivariate logistic regression model predicting bisphosphonate non-adherence. Only variables statistically significant at the p < 0.01 level were tabulated. The c statistic was 0.816, suggesting that the data fit the model well. The strongest predictor of bisphosphonate non- adherence was side effects: women most symptomatic in terms of side effects were 6.8 times more likely to be non-adherers than those least symptomatic. Women with the most skeptical beliefs about drug effectiveness were nearly six times more likely to be non-adherers than women with the most favorable beliefs about drug effectiveness. Women with middling beliefs about drug effectiveness (middle tertile) were twice as likely to be non-adherers compared to women with the most favorable beliefs about drug effectiveness. Women with the most skeptical beliefs about drug safety were over twice as likely to be non-adherers than women with the most favorable beliefs about drug safety. Women reporting no loss of height from peak height were 1.88 times more likely to be non-adherers than diose self-reporting any height loss. We conducted sensitivity tests using different categorizations of the multi-item scales (e.g., quartiles and quintiles vs. tertiles). Our observed results were maintained across all sensitivity tests. In none of the models did dosing frequency convenience, osteoporosis health concerns, or affordability emerge as a statistically significant predictor of bisphosphonate non-adherence. We also conducted sensitivity tests excluding the 23 patients who switched between weekly and monthly formulations. Our observed results were completely maintained across this sensitivity test.

Table 3. Differences between non-adherers and adherers in osteoporosis risk factors

Table 4. Differences between non-adherers and adherers on ratings of the bisphosphonate regimen

Table 5. Unadjusted mean differences between non-adherers and adherers on multi-item scales

Table 6. Multivariate adjusted odds ratios: variables associated with non-adherence

Table 7 compares non-adherers and adherers on their reasons for adherence (or lack thereof). The foremost reason for non-adhering to bisphosphonates was the presence of side effects: 67% of non- adherers rated the presence of side effects as an 'extremely' or 'very important' reason for non-adherence. For 55% of non-adherers, inferior drug effectiveness beliefs were an 'extremely' or 'very important' reason for non-adherence. Adherers rated each of the six medication attributes as important in influencing their adherence. For example, 92% of adherers endorsed that belief in the drug's effectiveness was an 'extremely' or 'very important' reason for their adherence and 83% endorsed that the absence of side effects was an 'extremely' or 'very important reason for adherence'.

Three clusters emerged from the k-means cluster analysis of side effects, drug effectiveness beliefs, and drug safety beliefs. Women in cluster I were characterized by the absence of troublesome side effects and the presence of confident beliefs in both the effectiveness and safety of bisphosphonates. Cluster I represented 60% of the sample, and they had non-adherence rates of 24%. Women in cluster II were characterized by troublesome side effects, confident beliefs about drug effectiveness, but skeptical beliefs about drug safety. Cluster II represented 22% of the sample, and they had non- adherence rates of 69%. Women in cluster III were characterized by troublesome side effects and skeptical beliefs about both drug effectiveness and safety. Cluster III represented 18% of the sample, and they had non-adherence rates of 82%.

Table 7. Reasons for non-adherence/adherence Discussion

Poor compliance with and non-adherence to oral bisphosphonates is pervasive11-13 and significantly blunts the potential of these therapies to prevent fractures and their associated adverse health consequences14-22. Therefore, bone clinicians and researchers, as well as those in primary care, widely recognize that there is a compelling need to identify the determinants of non-adherence to bisphosphonates in order to design appropriate strategies to improve fulfillment, adherence, and persistence.

In our survey of women aged 45 and older who had filled one or more prescriptions for an oral bisphosphonate medication in 2006, we found skeptical beliefs in the effectiveness and safety of pharmacologic fracture prevention therapy to be independent and strong predictors of adherence to bisphosphonates (even after adjusting statistically for the experience of troublesome side effects). Three studies in osteoporosis also found medication concerns, variously measured, to be associated with non- adherence28,30,35. Unson's40 qualitative work in osteoporosis also discovered that women's lack of trust in medications influenced adherence. Patient preference research in osteoporosis72,74,81,82 and other chronic diseases73,89-96 has repeatedly identified perceived drug effectiveness as a key determinant of medication preferences. Medication concerns41-51,97,98 and drug effectiveness beliefs43,99 have been found to be highly predictive of non- adherence in myriad chronic diseases other than osteoporosis.

Consistent with past research in osteoporosis23-37, we found that the experience of troublesome side effects powerfully predicted non- adherence to oral bisphosphonates. Our observed findings are also in harmony with recent patient preference research in osteoporosis reporting that the absence of side effects ranks very highly as a preferred treatment attribute among women with or at risk for osteoporosis72,81.

We found that women who self-reported no loss of height from peak height were more likely to be non-adherent to bisphosphonate therapy. Since height loss is a visible and obvious manifestation of poor bone health, the absence of personally noticeable change in height may give some women false confidence about their bone health and their need for pharmacologic fracture prevention therapy.

We were unable to confirm an association in this sample between concerns about medication cost and medication non-adherence, which is in contradiction to recent research on cost-related medication underuse among chronically ill adults100-102. However, a noteworthy proportion of both of our study's adherers and non-adherers had concerns about the costs of their bisphosphonates. Out of our six multi-item scales, our affordability scale had the lowest mean score (mean of 34 out of 100, Tables 2 and 5) and the highest percentage of floor effects (15%, Table 2). Thus, affordbility concerns were common in our sample but were not differential between adherers and non-adherers. A recently-published study also reported no association between drug costs and adherence to bisphosphonate therapy36.

Osteoporosis health concerns and dosing frequency convenience were associated with adherence at the bivariate level, but these scales did not enter into our multivariate stepwise logistic regression model. It is not surprising that dosing frequency convenience was unrelated to non-adherence because patient preference research in osteoporosis has shown dosing frequency to be a negligible determinant of treatment preference72,81,82. Further, database research in osteoporosis has shown only modest improvements in persistence (defined as the duration of time from initiation to discontinuation of therapy103) with weekly versus daily bisphosphonate therapy11 and, to date, no differences in persistence with weekly versus monthly bisphosphonate therapy78-80. Evaluations of the bisphosphonate regimen with respect to meals, other medications, staying upright, and activity interference were cited as concerns by only a minority of the sample and were not associated with non-adherence in the final multivariate model.

We corroborated the results of the multivariate regression analyses with a descriptive presentation of reasons for non- adherence and adherence. The two sets of results were exceedingly complimentary. Only two of the six reasons for non-adherence were endorsed by a majority of non-adherers: the presence of side effects (67% of non-adherers) and skeptical beliefs in drug effectiveness (55% of non-adherers). On the other hand, each of the six reasons for adherence was endorsed by a majority of adherers, but foremost among them was a belief in drug effectiveness (92% of adherers) and the absence of side effects (83% of adherers).

Our investigation follows on the heels of nine other survey- based studies of adherence to prescription osteoporosis medications. However, most of these extant studies can be characterized by limitations in study design or analysis that severely blunt their contributions to understanding treatment adherence in osteoporosis. For example, some of the past survey-based studies had small23 or unrepresentative samples (e.g., women who had a diagnostic bone densitometry in a single city in Wisconsin31 and Israel29). Others employed ad-hoc single-item measures23,29-31,36, and analyses were limited to simple descriptions of adverse events or reasons for non- adherence23,29,31,33,35,37. Only five of the extant survey-based studies employed multivariate methods to predict adherence to prescription osteoporosis medications26,30,35-37, but none of these studies comprehensively measured patient attitudes or beliefs specific to osteoporosis and/or its treatment.

In terms of strengths, our study is the only investigation to date that surveyed women across each of the nine US census regions, achieved a large sample size with ample power to test our hypotheses, measured an array of patient-centered drivers of adherence using reliable multi-item scales, and employed multivariate methods to isolate the unique contributions of our hypothesized drivers of adherence. In addition, our operational definition of adherence was provisionally based on retail pharmacy dispensing records and then cross-validated using patient self- report. Use of such a dual classification system is likely to be more accurate than sole reliance on either claims data or self- report alone.

The results of this study should be interpreted in light of several limitations. First, the design of the study prohibits us from drawing causal inferences with respect to the impact of patient beliefs on bisphosphonate adherence. It is unknown the extent to which such beliefs were ingrained a priori or whether they resulted from adherence or non-adherence to the prescribed bisphosphonate regimen. Second, the multi-item scales employed in this study were adapted from prior work and were kept brief to facilitate use in a telephone survey. As such, some small ceiling effects and floor effects were observed, and associations between these scales and non- adherence could have been slightly attenuated.

Several factors imply that our study sample is not likely to be a national random sample of bisphosphonate users. The pharmacy chain that supplied the dispensing records from which the study sample was drawn operated in only 28 states at the time this investigation was conducted. Although survey respondents represented all nine US census regions, this study is not truly national in scope. We only included in the study women who filled at least one prescription for an oral bisphosphonate. As a result, our results do not generalize to women who fail to fill or refuse an initial prescription for an oral bisphosphonate drug.

A notable limitation of our study pertains to the response rate of 33% and the possibility of selection bias due to survey non- response. We only had five variables to characterize non-response bias: age, state of residence, the index formulation (weekly or mondily), the provisional classification of bisphosphonate non- adherence, and the study sample (January or February 2006). No differences were observed between responders and non-responders on index dosing frequency, study sample, or census region. Small differences were observed for age with non-responders being slightly older than responders. The observed finding of a small bias associated with age is consistent with past research, which has documented worse survey response rates among older persons104,105. Further, under-representation of older persons has been found to be greater in telephone surveys than mail surveys105 or face-to-face interviews106.

We found non-responders more likely than responders to be provisionally classified as bisphosphonate non-adherers on the basis of pharmacy dispensing records. The most plausible explanation for this finding is the weak salience the topic of osteoporosis medications has for treatment non-adherers. The literature is rich in work linking respondent personal interest in a survey topic to survey participation105-108. The relationship is intuitive and is explained by leverage-salience tiieory108, which posits that people differ in the importance they attach to a request to participate in a survey. Survey nonresponse will be greater when the survey topic is not of special interest or relevance to potential sample member. Women who were non-adherent to their bisphosphonate could be expected to have weaker interest in the survey request than those who were adherent.

Although leverage-salience theory is useful in explaining our observed results, it remains unknown the extent to which, and in what possible direction, non-respondent non-adherers were different from respondent non-adherers in their beliefs about drug effectiveness and drug safety, osteoporosis health concerns, affordability concerns, dosing frequency evaluations, side-effect experiences, evaluations of the bisphosphonate regimen, and osteoporosis risk factors. However, the osteoporosis literature has repeatedly reported the experience or fear of side effects to be a major driver of bisphosphonate discontinuation23-37. Further, recent research in osteoporosis suggests that dosing frequency is a minor determinant of women's preferences for prescription bisphosphonate therapy72,81,82. Finally, the literature both within72,74,79,81,82 and outside43,90,92-94,96,99,109 of osteoporosis suggests that beliefs in drug effectiveness are key drivers of both treatment preferences and treatment adherence. Extrapolating from this literature, we hypothesize that our observed results for side effects, dosing frequency convenience, and drug effectiveness beliefs are unbiased, but this hypothesis remains untested. Conclusion

In conclusion, we believe our study is the first investigation using multivariate methods, a large sample, and reliable measures to demonstrate that beliefs regarding medication effectiveness and medication safety are strongly associated with non-adherence to oral bisphosphonates. These results suggest that, to improve adherence to oral bisphosphonate medication, healthcare providers need to directly solicit and address their patients' salient beliefs regarding the effectiveness and safety of medications being prescribed to them to prevent fractures. Specifically, patient beliefs that are at odds with published scientific data need to be patiently and carefully refuted. Unearthing patient belief systems about prescription medications and their treatment preferences for prescription medications lies at the heart of patient-centered care, shared decision-making, and informed choice. One avenue for uncovering such beliefs is to segment patients using brief, reliable scales such as those we used in this study. Future research should seek to develop conceptually cohesive, unidimensional, and reliable scales upon which to segment patients prior to or shortly after initiating prescription osteoporosis therapy. Finally, vis-a-vis our findings about side effects, a care plan should be collaboratively developed by providers and patients to address the possibility of side effects before therapy is initiated and to detect them promptly, should they occur.

Acknowledgments

Declaration of interest: This research was funded by Merck & Co., Inc.

* These results have been previously presented at the International Society for Clinical Densitometry, Tampa, FL, March 2007 and the 7th International Symposium on Osteoporosis, National Osteoporosis Foundation, Washington, DC, April 2007

References

1. National Osteoporosis Foundation. Fast facts. Available at: http://www.nof.org/osteoporosis/diseasefacts.htm [Last accessed 7 November 2006]

2. Magaziner J, Simonsick EM, Kashner TM, et al. Predictors of functional recovery one year following hospital discharge for hip fracture: a prospective study. J Gerontol 1990;45:M101-7

3. Fink HA, Ensrud KE, Nelson DB, et al. Disability after clinical fracture in postmenopausal women with low bone density: the fracture intervention trial (FIT). Osteoporos Int 2003;14:69-76

4. Huang C, Ross PD, Wasnich RD. Vertebral fractures and other predictors of back pain among older women. J Bone Miner Res 1996;11:1026-32

5. Nevitt MC, Ettinger B, Black DM, et al. The association of radiographically detected vertebral fractures with back pain and function: a prospective study. Ann Intern Med 1998;128:793-800

6. Black DM, Cummings SR, Karpf DB, et al. Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Fracture Intervention Trial Research Group. Lancet 1996; 348:1535-41

7. Hodgson SF, Watts NB, Bilezikian JP, et al. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the prevention and treatment of postmenopausal osteoporosis: 2001 edition, with selected updates for 2003. Endocr Pract 2003;9:544-64

8. ACOG Committee on Practice. ACOG practice bulletin. Clinical management guidelines for obstetrician-gynecologists. Osteoporosis. Number 50. Obstet Gynecol 2004;103:203-16

9. Brown JP, Josse RG. 2002 clinical practice guidelines for the diagnosis and management of osteoporosis in Canada. CMAJ 2002;167:S1- 34

10. National Osteoporosis Foundation. Physician's Guide to Prevention and Treatment of Osteoporosis. Washington, DC: National Osteoporosis Foundation, 2003

11. Cramer JA, Gold DT, Silverman SL, Lewiecki EM. A systematic review of persistence and compliance with bisphosphonates for osteoporosis. Osteoporos Int 2007;18:1023-31

12. Papaioannou A, Kennedy CC, Dolovich L, et al. Patient adherence to osteoporosis medications: problems, consequences and management strategies. Drugs Aging 2007;24:37-55

13. Seeman E, Compston J, Adachi J, et al. Non-compliance: the Achilles' heel of anti-fracture efficacy. Osteoporos Int 2007;18:711- 19

14. Yood RA, Emani S, Reed JI, et al. Compliance with pharmacologic therapy for osteoporosis. Osteoporos Int 2003;14:965- 8

15. Caro J, Ishak K, Huybrechts K, et al. The impact of compliance with osteoporosis therapy on fracture rates in actual practice. Osteoporos Int 2004;15:1003-8

16. McCombs J, Thiebaud P, McLaughlin-Miley C, Shi J. Compliance with drug therapies for the treatment and prevention of osteoporosis. Maturitas 2004;48:271-87

17. Huybrechts KF, Ishak KJ, Caro JJ. Assessment of compliance with osteoporosis treatment and its consequences in a managed care population. Bone 2006;38:922-8

18. Siris ES, Harris ST, Rosen CJ, et al. Adherence to bisphosphonate therapy and fracture rates in osteoporotic women: relationship to vertebral and nonvertebral fractures from 2 US claims databases. Mayo Clin Proc 2006;81:1013-22

19. van den Boogaard CH, Breekveldt-Postma NS, Borggreve SE, et al. Persistent bisphosphonate use and the risk of osteoporotic fractures in clinical practice: a database analysis study. Curr Med Res Opin 2006;22:1757-64

20. Gold DT, Martin BC, Frytak JR, et al. A claims database analysis of persistence with alendronate therapy and fracture risk in post-menopausal women with osteoporosis. Curr Med Res Opin 2007;23:585-94

21. Penning-van Beest FJ, Erkens JA, Olson M, Herings RM. Loss of treatment benefit due to low compliance with bisphosphonate therapy. Osteoporos Int 2007. Published article online: 14 Sep 14 2007 doi:10.1007/s00198-007-0466-1

22. Weycker D, Macarios D, Edelsberg J, Oster G. Compliance with osteoporosis drug therapy and risk of fracture. Osteoporos Int 2007;18:271-7

23. Hamilton B, McCoy K, Taggart H. Tolerability and compliance with risedronate in clinical practice. Osteoporos Int 2003;14:259- 62

24. Turbi C, Herrero-Beaumont G, Acebes JC, et al. Compliance and satisfaction with raloxifene versus alendronate for the treatment of postmenopausal osteoporosis in clinical practice: an open-label, prospective, nonrandomized, observational study. Clin Ther 2004;26:245-56

25. Segal E, Tamir A, Ish-Shalom S. Compliance of osteoporotic patients with different treatment regimens. Isr Med Assoc J 2003;5:859-62

26. Tosteson AN, Grove MR, Hammond CS, et al. Early discontinuation of treatment for osteoporosis. Am J Med 2003;115:209- 16

27. Burgener M, Arnold M, Katz JN, et al. Older adults' knowledge and beliefs about osteoporosis: results of semistructured interviews used for the development of educational materials. J Rheumatol 2005;32:673-7

28. International Osteoporosis Foundation. The adherence gap: why osteoporosis patients don't continue with treatment. Available at: http://www.osteofound.org/publications/ adherence_gap_report.html [Last accessed 14 July 2006]

29. Zafran N, Liss Z, Peled R, et al. Incidence and causes for failure of treatment of women with proven osteoporosis. Osteoporos Int 2005;16:1375-83

30. Carr AJ, Thompson PW, Cooper C. Factors associated with adherence and persistence to bisphosphonate therapy in osteoporosis: a cross-sectional survey. Osteoporos Int 2006;17:1638-44

31. Pickney CS, Arnason JA. Correlation between patient recall of bone densitometry results and subsequent treatment adherence. Osteoporos Int 2005;16:1156-60

32. Ideguchi H, Ohno S, Hattori H, Ishigatsubo Y. Persistence with bisphosphonate therapy including treatment courses with multiple sequential bisphosphonates in the real world. Osteoporos Int 2007;18:1421-7

33. Jakob F, Marin F, Martin-Mola E, et al. Characterization of patients with an inadequate clinical outcome from osteoporosis therapy: the Observational Study of Severe Osteoporosis (OSSO). QJM 2006;99:531-43

34. Penning-van Beest FJ, Goettsch WG, Erkens JA, Herings RM. Determinants of persistence with bisphosphonates: a study in women with postmenopausal osteoporosis. Clin Ther 2006;28:236-42

35. Rossini M, Bianchi G, Di Munno O, et al. Determinants of adherence to osteoporosis treatment in clinical practice. Osteoporos Int 2006;17:914-21

36. Kamatari M, Koto S, Ozawa N, et al. Factors affecting long- term compliance of osteoporotic patients with bisphosphonate treatment and QOL assessment in actual practice: alendronate and risedronate. J Bone Miner Metab 2007;25:302-9

37. Ringe JD, Christodoulakos GE. Patient compliance with alendronate, risedronate and raloxifene for the treatment of osteoporosis in postmenopausal women. Curr Med Res Opin 2007. Published article online: 19 Sep 2007 doi:10.1185/03007x226357

38. Conrad P. The meaning of medications: another look at compliance. Soc Sci Med 1985;20:29-37

39. Horne R, Weinman J. Patients' beliefs about prescribed medicines and their role in adherence to treatment in chronic physical illness. J Psychosom Res 1999;47:555-67

40. Unson C, Siccion E, Gaztambide J, et al. Nonadherence and osteoporosis treatment preferences of older women: a qualitative study. J Womens Health 2003;12:1037-45

41. Horne R, Sumner S, Jubraj B, et al. Haemodialysis patients' beliefs about treatment: implications for adherence to medication and fluid-diet restrictions. Int J Pharm Pract 2001;9:169-75 42. Horne R, Weinman J. Self-regulation and self-management in asthma: exploring the role of illness perceptions and treatment beliefs in explaining non-adherence to preventer medication. Psychol Health 2002;17:17-32

43. Spire B, Duran S, Souville M, et al. Adherence to highly active antiretroviral therapies (HAART) in HIV-infected patients: from a predictive to a dynamic approach. Soc Sci Med 2002;54:1481- 96

44. Butler JA, Peveler RC, Roderick P, et al. Modifiable risk factors for non-adherence to immunosuppressants in renal transplant recipients: a cross-sectional study. Nephrol Dial Transplant 2004;19:3144-9

45. Horne R, Buick D, Fisher M, et al. Doubts about necessity and concerns about adverse effects: identifying the types of beliefs that are associated with non-adherence to HAART. Int J STD AIDS 2004;15:38-44

46. Ross S, Walker A, MacLeod MJ. Patient compliance in hypertension: role of illness perceptions and treatment beliefs. J Hum Hypertens 2004;18:607-13

47. Brown C, Battista DR, Bruehlman R, et al. Beliefs about anti- depressant medications in primary care patients: relationship to self-reported adherence. Med Care 2005;43:1203-7

48. Neame R, Hammond A. Beliefs about medications: a questionnaire survey of people with rheumatoid arthritis. Rheumatology 2005;44:762-7

49. De Smet BD, Erickson SR, Kirking DM. Self-reported adherence in patients with asthma. Ann Pharmacother 2006;40:414-20

50. Phatak HM, Thomas J. Relationships between beliefs about medications and nonadherence to prescribed chronic medications. Ann Pharmacother 2006;40:1737-42

51. Horne R, Cooper V, Gellaitry G, et al. Patients' perceptions of highly active antiretroviral therapy in relation to treatment uptake and adherence: the utility of the necessity-concerns framework. J Acquir Immune Defic Syndr 2007;45:334-41

52. Cline RR, Farley JF, Hansen RA, Schommer JC. Osteoporosis beliefs and antiresorptive medication use. Maturitas 2005;50:196- 208

53. Unson CG, Fortinsky R, Prestwood K, Reisine S. Osteoporosis medications used by older African-American women: effects of socioeconomic status and psychosocial factors. J Community Health 2005;30:281-97

54. Cline RR, Worley MM. Osteoporosis health beliefs and self- care behaviors: an exploratory investigation. J Am Pharm Assoc 2006;46:356-63

55. Miller NH, Hill M, Kottke T, Ockene IS. The multilevel compliance challenge: recommendations for a call to action. A statement for healthcare professionals. Circulation 1997;95:1085-90

56. World Health Organization. Adherence to Long-Term Therapies. Geneva Switzerland: World Health Organization, 2003

57. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med 2005;353:487-97

58. Dunbar-Jacob J, Mortimer-Stephens MK. Treatment adherence in chronic disease. J Clin Epidemiol 2001;54:S57-60

59. Ockene IS, Hayman LL, Pasternak RC, et al. Task force #4 adherence issues and behavior changes: achieving a long-term solution. J Am Coll Cardiol 2002;40:630-40

60. Morrison A, Wertheimer AI. Compilation of quantitative overviews of studies of adherence. Drug Inf J 2004;38:197-210

61. Krueger KP, Berger BA, Felkey B. Medication adherence and persistence: a comprehensive review. Adv Ther 2005;22:313-56

62. Nunnally JC. Psychometric Theory. New York: McGraw-Hill, 1978

63. McHorney C. Methodological and psychometric issues in health status assessment across populations and applications. In: Albrecht GL, Fitzpatrick R, eds. Advances in Medical Sociology, Vol. 5. Stamford CT: JAI Press, Inc., 1994:281-304

64. McHorney CA, Ware JE, Lu JFR, Sherbourne CD. The MOS 36-item short-form health survey (SF-36): III. Tests of data quality, scaling assumptions and reliability across diverse patient groups. Med Care 1994;32:40-66

65. McHorney C. Generic health measurement: past accomplishments and a measurement paradigm for the 21st century. Ann Intern Med 1997;127:743-50

66. McHorney C. Health status assessment methods for adults: past accomplishments and future challenges. Annu Rev Public Health 1999;20:309-35

67. McHorney C. Ten recommendations for advancing patient- centered outcomes measurement for older persons. Ann Intern Med 2003;139:403-9

68. International Osteoporosis Foundation. How fragile is her future? Available at: http://www.iofbonehealth.org/ policy-advocacy/ iof-policy-publications/how-fragile-is-her-future-2000-2002.html [Last accessed 18 July 2006]

69. Resnick B, Wehren L, Orwig D. Reliability and validity of the self-efficacy and outcome expectations for osteoporosis medication adherence scales. Orthop Nurs 2003;22: 139-47

70. National Osteoporosis Foundation. Health issues survey: attitudes and actions regarding osteoporosis. Available at: http:// www.nof.org/news/pressreleases/Health_Issues_ Survey_Osteoporosis.pdf [Last accessed 18 Jul 2006]

71. Jachna C, Forbes-Thompson S. Osteoporosis: health beliefs and barriers to treatment in an assisted living facility. J Gerontol Nurs 2005;31:25-30

72. Weiss T, Gold D, Silverman S, McHorney CA. An evaluation of patient preferences for osteoporosis medications and their attributes: results from the PREFER-US Study. Curr Med Res Opin 2006;22:949-60

73. Miller L, Huffman H, Weidmer B, Hays R. Patient preferences regarding antiretroviral therapy. Int J STD AIDS 2002;13:593-601

74. Keen R, Jodar E, Iolascon G, et al. European women's preference for osteoporosis treatment: influence of clinical effectiveness and dosing frequency. Curr Med Res Opin 2006;22:2375- 81

75. DiMatteo M, Haskard K, Williams S. Health beliefs, disease severity, and patient adherence: a meta-analysis. Med Care 2007;45:521-8

76. Elliott RA, Ross-Degnan D, Adams AS, et al. St

Source: Current Medical Research and Opinion

More News in this Category