Bruce Ettinger, Alice Pressman, Jeffrey Schein, Julia Chan, Paula Silver, and Nancy Connolly

ABSTRACT: The objective of this study was to determine the extent to which women who receive alendronate therapy for osteoporosis comply with instructions regarding its use, find alendronate tolerable, and continue with the treatment.

Through telephone interview surveys and retrospective database analysis, the authors surveyed 812 women with a mean (SD) age of 68.7 (11.7) years, who filled at least one prescription for 10 mg alendronate between October 1, 1995, and October 31, 1996. All the patients are members of the Kaiser Permanente Medical Care Program in Northern California.

Based on interviews conducted a mean (SD) 8.0 (3.1) months after alendronate therapy was begun, 91.8% of the women recalled receiving verbal or printed instructions, but only 60% reported that they received all required instructions. During the course of therapy, 55.8% did not comply with one or more instructions for taking the drug. Specifically, 13.5% failed to comply with at least one of the recommended safety rules, and 51.7% disregarded rules relating to proximate ingestion of food, liquids (other than water), or medications. New upper-gastrointestinal symptoms ascribed to use of alendronate were reported by 32.7% of all users. The proportion of women discontinuing was 28.6% as based on self-report and was 34.9% as based on review of prescription refills. The most common reason for discontinuation was gastrointestinal problems, reported by 51.9% of women who stopped taking the drug.

Almost one in three women receiving alendronate as therapy for osteoporosis has alendronate-related gastrointestinal complaints, which often cause them to discontinue treatment. Patients who fail to comply with instructions for taking alendronate may incur side effects from the drug or may fail to achieve full therapeutic effect from its use.

KEY WORDS: Alendronate, Bone density, Bisphosphonates, Osteoporosis, Post-menopausal

J Managed Care Pharm 1998: 488-492

In clinical trials, alendronate (Fosamax, Merck & Co.) has been shown to reduce fracture risk by 50%.^1-3 However, it must be administered daily, continued for an extended period, and taken properly to have salutary skeletal effects. Health care providers and patients may be wasting resources and time if patients discontinue using alendronate or use it improperly.

Reports conflict about alendronate’s side effects and tolerability. Based on extensive clinical trial testing, alendronate ap-pears to be well tolerated, and doses of 5 mg–10 mg have not been associated with any increase in risk of gastrointestinal or other side effects.^1-3 In the past one to two years, as clinicians around the world have gained experience with alendronate, esophageal irritation and upper-gastrointestinal symptoms have been reported.⁴ As a class, aminobisphosphonates have a pro-pensity to cause esophageal irritation and, occasionally, ulceration.⁵ On March 15, 1996, Merck & Co. issued a letter warning physicians that alendronate can irritate the esophagus, but that careful adherence to the dosing instructions can reduce the likelihood of these side effects.⁶ The product circular was revised to better instruct patients regarding the need to take alendronate with at least six ounces of water, not to chew or suck on the tablet, and to remain upright at least 30 minutes after dosing. It is unclear how much these steps reduce the potential for esophageal irritation.

Women participating in clinical trials probably are healthier and better counseled than those receiving the drug in clinical practice. Women with active gastrointestinal disease, using H2 blockers, using glucocorticoids, and those who smoked heavily or who consumed alcohol heavily were excluded from most clinical trials of alendronate. Therefore, we decided to monitor the continuation of drug therapy and adverse events associated with alendronate use in a large health maintenance organization (HMO) where these treatment restrictions did not apply.

Alendronate, like all other bisphosphonates, is poorly ab-sorbed from the gastrointestinal tract;⁷ availability of the oral dose is <1%.⁸ Absorption is further reduced if alendronate is taken with any substance that could bind to it. Almost any food or drink, except water, can reduce bioavailability.^8,9 Taking alendronate with food reduces absorption by as much as 90%.^8,9 Taking alendronate with black coffee or with orange juice re-duces absorption by 60% compared to taking alendronate with water.⁸ In a clinical trial of etidronate, another bisphosphonate, coadministration of calcium supplements eliminated the skeletal benefit of the drug.¹⁰ Thus, strict enforcement of restrictions on food and medication use are necessary to prevent a marked decrease in alendronate efficacy. The current alendronate package circular mentions that the tablet should be taken on arising (not at bedtime), that it should not be taken with liquids other than water, and that patients should wait at least 30 minutes after dosing before eating, drinking, or taking any other medications. We wanted to determine how often these recommendations were being followed in clinical practice and whether certain types of patients were more likely to demonstrate poor compliance with these instructions.

Therefore, we designed a telephone survey of women members of the Kaiser Permanente health plan who had taken 10 mg alendronate between October 1, 1995, and October 31, 1996. We hypothesized that women older than 75 years would have poorer compliance with dosing guidelines and a higher rate of discontinuation. We also hypothesized that women receiving written warnings and instructions about how to take alendro-nate would be more likely to comply with dosing guidelines.

The Kaiser Permanente Medical Care Program serves 2.6 million plan members at 48 facilities in Northern California. Health plan pharmacies record all prescriptions in a computer database (Pharmacy Information Monitoring Service, or PIMS). Included in the database are patient identification; prescriber information; dispensing date; drug information, including name, strength, and number of pills dispensed; directions for use; and cost. Nearly all health plan members have prescription copayment plans. Thus, the median cost for a year’s supply of alendronate for study participants was $20 (range: $0–$631); 90% of subjects in this study paid less than $100 for a year’s supply.

From PIMS, we identified 981 women who filled a prescription for alendronate tablets, 10 mg. Excluded were those whose prescription directions indicated other than "daily" or "as directed" and those receiving 40 mg dosage (prescribed for Paget’s disease). We then mailed a letter to eligible women inviting them to participate in a 10-minute telephone interview; 6.3% refused to be interviewed. Others were ineligible for the following reasons: 0.8% had died; 4.0% could not be reached by telephone or could not be interviewed; 2.1% stated that they had not filled a prescription for alendronate; and 2.9% had filled a prescription but did not take any of the pills. Eight hundred twelve women remained who were eligible for interview and database analysis. In 96% of cases, the subject was interviewed; in 4% of cases, family members provided the information.

Women who reported that they had stopped using alendronate were asked two open-ended questions: "What was the primary reason that you discontinued?" and "What other reason (if any) did you have to discontinue?" The survey also asked yes/no questions regarding various new upper-gastrointestinal symptoms that the subjects specifically attributed to alendronate use. Specific yes/no questions were asked about factors that affect alendronate absorption, including timing of meals, drinks, calcium supplements, vitamins, antacids, and other medications in relation to ingestion of alendronate. Subjects also were asked specifically about factors that could affect drug safety, including how much water (if any) the subjects took with the alendronate pills, and whether they remained upright for 30 minutes after taking the pills.

We determined whether subjects had continued using alendronate by examining PIMS for alendronate refills. We also searched PIMS for all other prescriptions. Based on the therapeutic classification of drugs in PIMS, we determined the total number of drug types used and the total number of different therapeutic classes represented by these drugs for all survey subjects. We particularly assessed use of drugs that could be associated with a higher risk of upper-gastrointestinal disorders, including nonsteroidal antiinflammatory drugs (NSAIDs), glucocorticoids, and drugs for acid-related disorders (ARD), such as H2 blockers and proton-pump inhibitors. To determine if patients taking these drugs were less tolerant of alendronate, we analyzed the subjects’ use of drugs in these therapeutic classes for the period six months before their first alendron-ate prescription through the end of alendronate usage.

When determining duration of alendronate use from prescription refill data, we allowed 30 days to elapse between the time the prior supply of tablets would have been exhausted and the time of the subsequent refill. We divided prescriber type into specialists likely to be skilled in care of osteoporosis patients (endocrinology and rheumatology) versus other physicians. We dichotomized subjects into early users (therapy initiated before March 15, 1996) versus late users (therapy initiated on or after that date); March 15, 1996, was about when the manufacturer of alendronate issued a "Dear Doctor" letter providing further clarification on dosing instructions to clinicians. We classified drinking four ounces or less of water as noncompliance with the patient instruction to drink at least six ounces of water when taking alendronate. To test the relationship between dichotomous predictors and outcomes, we used chi-squared analyses. We used logistic regression to test relationships between patient variables and dichotomous outcomes. All analyses were done by using SAS software (Statis-tical Analysis Systems of Cary, North Carolina).

Table 1 shows characteristics of the 812 women who were interviewed. Most (55.7%) were more than 70 years old, and all used some concomitant medications. Subjects used medications from a median of seven different therapeutic drug classes (range: 1–37 drug classes).

The number of women beginning alendronate therapy increased considerably during the year of study. Before March 15, 1996, 31% (n=252) of patients had begun taking alendronate, compared with 69% (n=560) after that date.

Many different health care providers (n=593) prescribed alendronate to these women. General internists constituted the largest single prescriber class. Specialists likely to be knowledgeable about osteoporosis prescribed alendronate to 41.3% of subjects.

Of the women interviewed, 40% did not recall receiving all of the instructions (See Figure 1). Between 16.6% and 23.1% of subjects did not recall receiving one of the instructions. Women who filled their first prescription after March 15, 1996, were more likely to report that they had received instructions than those whose first prescription was filled earlier (See Table 2).

During the course of therapy, 55.8% of those interviewed did not comply with at least one patient instruction (See Figure 2). Specifically, 13.5% failed to comply with at least one of the recommended safety rules, and 51.7% did not comply with at least one of the instructions relating to absorption of alendro-nate. Women over 75 years of age were not distinguished in their noncompliance—57.5% of older women and 55.24% of younger women failed to comply with at least one instruction (p=.53). No association was seen between prescriber type and patients’ compliance with instructions (data not shown). Those who began alendronate therapy before March 15, 1996, showed a nonsignificant tendency to be less compliant with instructions (59.9% for earlier versus 53.9% for later, p=.11). Compared with women who stated they had received all patient instructions, those who stated they had not received all the instructions were more likely not to comply with them (67.7% versus 48.0%, respectively, p<.001).

In the entire group, 32.7% reported new upper-gastrointestinal symptoms ascribed to use of alendronate (See Table 3). Those who filled prescriptions for drugs to treat ARD were more likely (p=.01) to report new gastrointestinal problems than women not using these drugs. The risk of experiencing new gastrointestinal symptoms was not affected by age, compliance with all safety instructions, or use of NSAIDs or glucocorticoids (data not shown). Among those who reported compliance with absorption instructions (i.e., avoiding food, drink, or medications that could reduce absorption of the drug), the proportion experiencing new gastrointestinal symptoms was significantly higher than that of those who did not avoid such binders (p=<.01).

We assessed discontinuation in two different ways—by subject self-report and by using PIMS to observe prescription refills. On the basis of telephone interviews, 28.5% of women reported discontinuation of alendronate within a mean eight months of initiating therapy (range: 2.0–15.3 months). Based on PIMS database analysis, 34.9% of women discontinued by six months. The probability of discontinuation was moderately linear for the first six months of use (See Figure 3). The two most common symptoms these women reported as the reason for discontinuing alendronate were gastrointestinal (51.9%) and musculoskeletal (9.4%).

The risk of discontinuation was not related to age, prescriber type, compliance with recommended drug safety instructions, use of NSAIDs, or use of glucocorticoids. However, those using drugs for ARDs were more likely to discontinue using alendro-nate (RR=1.79, 95% CI 1.27–2.52, p<.001).

We found that gastrointestinal adverse events and the discontinuation of alendronate in a clinical practice setting are much higher than reported in randomized, double-blinded, clinical trials.^3,11,12 Further, more than half of women taking alendronate fail to comply with recommendations regarding food, drink, and medications, possibly severely curtailing or eliminating the skeletal benefit of the drug.

In clinical trials of 10 mg alendronate, adverse gastrointes-tinal problems were as common among women receiving pla-cebo as among those taking the active drug.^3,11,12 However, eso-phageal and gastric irritation in patients receiving aminobisphosphonate drugs is well recognized.⁵ In response to reports

of such problems, the manufacturer has made recommendations to reduce likelihood that the tablets will remain in contact with the esophagus for long periods, and it has warned that patients with esophageal or gastric disease should avoid using alendronate.¹³

The degree to which these safety precautions will reduce the incidence of upper-gastrointestinal symptoms is unclear. Our study suggests that incidence of these problems is similar among women who comply with all safety instructions and those who do not, and in those who avoid drugs known to damage the gastrointestinal lining as well as those who do not. The higher rate of gastrointestinal problems associated with full compliance with instructions to avoid food or medications that bind alendronate suggests that the amount of absorbed drug may be more important in causing these symptoms than failure to comply with recommended safety precautions. Further supporting this hypothesis is our finding that new gastrointestinal symptoms were more frequent among women using drugs for ARDs (largely H2 blockers), which can increase absorption and thereby the serum level of alendronate.⁸

Discontinuing drug therapy is common in clinical practice. For example, nearly half of women starting hormone replacement therapy stop after a mean nine months;¹⁴ among patients starting antihypertensive therapy, 50%–60% stop within six months.¹⁵ A double-blinded, randomized clinical trial of a drug tends to underestimate the likelihood of drug discontinuation in clinical practice. For alendronate, clinical trials have reported discontinuation rates ranging from 8%–13% after three years.^3,12 Many factors could contribute to the relatively high rate of discontinuation we observed. Screening clinic patients for exclusionary diagnoses may not have been as thorough as screening of study participants; clinic patients may not have been highly motivated because they were not volunteers in a research study; and clinicians may not have provided the patient education, encouragement, and support provided by a motivated research team. The similarity in continuation rates between women treated by presumed osteoporosis specialists and other health care providers suggests that physician knowledge alone is not sufficient to obtain excellent continuation of therapy. On the basis of self-report, most patients who discontinued therapy did so because of drug-induced side effects.

A high proportion of patients recalled being given instructions regarding alendronate use, but it is possible that they did not understand the purpose of the therapy, its potential benefits, or the importance of following the dosing instructions. It is likely that simply providing the information is not adequate to achieve a high degree of compliance with dosing instructions. Physicians and pharmacists may vary in their knowledge and their ability to convey it when initiating drug treatment for women. The influence of patient education on continuation deserves further study.

One limitation of our study involved the sensitivity of the survey instrument, which may not have been sensitive enough to distinguish between women who had minor and intermittent noncompliance with dosing instructions and those who had more serious and recurrent dosing violations. Patients were asked whether they "ever" failed to comply with each instruction. We did not determine how often these dosing violations had occurred.

We had the advantage of studying a moderately large, homo-geneous population that had relatively easy access to physicians and pharmacies. The PIMS database provided a convenient and reliable method of tracking both alendronate and concomitant drug use; discontinuation results were similar whether we used the PIMS database approach or telephone interview data. Within our HMO, women would be unlikely to visit pharmacies outside the health plan because the prescription cost would usually be considerably greater. Women within an HMO have few barriers to receiving health care, and nearly all had received treatment from internists and subspecialists who are more likely to be well versed in osteoporosis management. Together with the low cost of prescriptions, these factors may have caused our subjects to continue treatment longer than less-privileged women.

We conclude that, in usual clinical practice, most patients taking alendronate for osteoporosis are at risk of losing therapeutic effect by failing to comply with dosing guidelines aimed at preventing poor absorption of the drug. Discontinuation of alendronate therapy within the first six months of treatment occurs in about 30% of patients, most often because of gastrointestinal complaints. Thus, efforts should be made to improve tolerability of alendronate. Our data suggest that, in clinical practice, considerable medical resources may be expended without obtaining the skeletal benefits of alendronate.