Patient knowledge, beliefs, and behavior concerning the prevention and treatment of glucocorticoid-induced osteoporosis

Authors


Abstract

Objective

To obtain descriptive information concerning the extent to which patients taking oral glucocorticoids recall receiving osteoporosis prevention counseling, and to identify factors associated with the practice of 3 behaviors (i.e., calcium intake, vitamin D intake, and bone mineral density [BMD] testing) recommended for patients receiving glucocorticoid therapy.

Methods

The study assessed cross-sectional data derived from telephone interviews and mailed questionnaires completed by 227 patients who were currently taking oral prednisone. Questions assessed current calcium and vitamin D intake and history of BMD testing.

Results

Approximately one-third of participants (36.3%) reported that they had received osteoporosis prevention counseling. Among those who reported receiving counseling, most (73.2%) remembered being told about the importance of obtaining an adequate amount of calcium. Other topics were remembered less frequently. Slightly more than half of study participants (51.1%) were obtaining the recommended amount of calcium. Fewer were obtaining the recommended amount of vitamin D (35.2%) or had received a BMD test within the past year (33.5%). The most consistent predictors of behavior were counseling status and patients' perceptions of the difficulty associated with performing the behavior.

Conclusion

Most patients receiving oral glucocorticoids receive insufficient counseling concerning the prevention of osteoporosis; patients either are not being counseled or they are being counseled in a manner that is not sufficient to promote subsequent recall and behavior change. Research is needed to develop effective strategies to educate patients about the prevention of glucocorticoid-induced osteoporosis.

INTRODUCTION

Therapy with oral glucocorticoids often plays an important role in the management of inflammatory forms of arthritis by alleviating symptoms and slowing disease progression (1). Despite these benefits, however, use of oral glucocorticoids is also a well-established risk factor for osteoporosis (2–7). Past research suggests that up to 25% of patients receiving long-term steroid therapy will develop an osteoporosis-related fracture (5–8). Bone loss is most rapid during the first 6 months of therapy, with typical losses thereafter ranging from 3–10% per year (9, 10). Given the substantial risks associated with oral glucocorticoid therapy, it is imperative that patients treated with these medications receive effective counseling aimed at osteoporosis prevention and that they be monitored for deleterious effects of therapy on bone density.

Several studies have examined physician practice patterns with respect to the management and prevention of glucocorticoid-induced osteoporosis (GIOP). These studies suggest that osteoporosis prophylaxis for patients treated with oral glucocorticoids is far from routine. Precise estimates vary widely. For example, in 2 early studies conducted in the United Kingdom, rates of 6% and 14% were reported (11, 12). More recent reports have estimated the proportion of patients receiving some form of osteoporosis prophylaxis at between 30% and 62%, still leaving considerable room for improvement (8, 13–18).

Past studies have also examined the association between osteoporosis prophylaxis and a variety of patient and physician characteristics. Findings from these studies suggest, for example, that postmenopausal women and patients treated by a rheumatologist are more likely than others to receive recommendations concerning osteoporosis prophylaxis (13–17). However, most of these past studies have been based on information obtained from medical record reviews and have been limited by the lack of comprehensive information concerning patients' knowledge, beliefs, and behaviors. Thus, little is known about the role that patient-level factors play in the prevention and treatment of GIOP. The current study was designed to address this knowledge gap.

The study had 2 objectives. The first objective was to obtain in-depth descriptive information concerning the extent to which patients receiving oral glucocorticoids recall having been counseled about osteoporosis prevention; their recollection of the topics covered in counseling; and their knowledge, beliefs, and behaviors with respect to osteoporosis and osteoporosis prevention. The second objective of the study was to identify factors associated with the practice of 3 behaviors recommended by the American College of Rheumatology (ACR) Ad Hoc Committee on Glucocorticoid-Induced Osteoporosis: maintaining an adequate intake of calcium, maintaining an adequate intake of vitamin D, and obtaining annual bone mineral density (BMD) tests to monitor the effects of therapy (19). The study was guided by a conceptual model suggesting that patient behavior is influenced most proximally by patients' knowledge and beliefs about osteoporosis and osteoporosis prevention, patients' knowledge and beliefs are influenced by counseling, and patient sociodemographic characteristics may influence the likelihood of receiving counseling.

PATIENTS AND METHODS

Patient recruitment.

Study participants were recruited via a pharmacy claims database maintained by a major pharmaceutical benefits management (PBM) company. The database was queried to identify individuals who were age 18 years or older; had filled at least 2 prescriptions for prednisone within the previous 6 months; and had filled at least 1 prescription for sulfasalazine, etanercept, gold, leflunomide, methotrexate, penicillamine, or hydroxychloroquine within the previous 6 months. These later medications were used as pharmaceutical markers to limit the query to individuals likely to have either rheumatoid arthritis or another type of inflammatory arthritis. Two queries were performed, the first in January 2002 and the second in October 2002. A total of ∼2,650 individuals were identified and were sent a letter from the PBM informing them of the study. The letter indicated that the study was limited to individuals with rheumatoid arthritis who were currently taking prednisone. A form was included with the letter that individuals could return if they believed they were eligible for the study and were interested in participating. A self-addressed, stamped envelope was also enclosed. A total of 426 individuals (16%) returned the interest form. These individuals were then contacted by telephone to conduct a brief screening interview. In 18 cases, we were unable to contact the individual to conduct the screening interview despite repeated attempts. Eleven other individuals refused to participate in the screening interview. Therefore, a total of 397 screening interviews were completed. The screening interview determined that 85 individuals were not eligible to participate in the study. Reasons for ineligibility were not currently taking prednisone (n = 62), instructed to limit calcium intake (n = 7), instructed to limit vitamin D intake (n = 5), history of breast/prostate cancer (n = 5), non-English speaking (n = 3), in ACR Functional Class IV (n = 2) (20), deceased (n = 2), and pregnant or breast-feeding (n = 1). Of the remaining 312 individuals, 227 (72.8%) completed all baseline data collection procedures and formed the study sample.

Data collection procedures.

Data were collected via mailed questionnaire and telephone interview. The questionnaire included mostly fixed-choice items, whereas the interview included many open-ended questions and multipart questions with complex skip patterns.

Immediately following the screening interview, eligible participants were mailed the questionnaire to complete and return by mail. Followup procedures were modeled after the recommendations of Dillman (21). Approximately 1 week after the questionnaire was mailed, the participant was sent a reminder/thank you letter, regardless of whether the questionnaire had been returned. Participants who had not returned the questionnaire were mailed a followup packet containing a replacement questionnaire at 3 weeks and again at 8 weeks after the mailed questionnaire was sent. After the questionnaire was returned, participants were contacted by telephone to conduct the telephone interview. Interviews lasted an average of 20 minutes.

Measures.

Behavior.

Calcium and vitamin D intake were assessed using a brief food frequency questionnaire (FFQ) derived from the Health Habits and History Questionnaire (HHHQ) (22, 23).The HHHQ is often used in epidemiologic studies to assess nutrient intake, and substantial data support its reliability and validity (24–27). The brief FFQ was limited to 23 foods/beverages that are high in either calcium or vitamin D. Participants recorded how frequently they ate/drank each food/beverage during the past year. For each food/beverage, participants also recorded their usual serving size (i.e., small, medium, large) according to guidelines provided. Responses to the frequency items were recorded on a 9-point scale. For foods, response options ranged from “never or less than once a month” to “2+ times per day.” For beverages, response options ranged from “never or less than once a month” to “6+ times per day.” In a small pilot study using 7-day food diaries as the criterion measure of nutrient intake, the brief FFQ was found to perform as well as the full HHHQ in estimating dietary intake of calcium and vitamin D (23). In addition, during the telephone interview, we obtained information about the amount of calcium and vitamin D participants received from vitamin and mineral supplements and from fortified foods/beverages (e.g., orange juice fortified with calcium). These amounts then were added to the estimates of dietary calcium/vitamin D intake to yield the final estimates of total daily calcium/vitamin D intake. Participants were categorized as meeting the ACR guideline for calcium and vitamin D intake (19) if their estimated daily consumption of these nutrients was at least 1,500 mg and 800 IU, respectively.

History of BMD testing was assessed via the telephone interview. Participants were asked if they had ever undergone a BMD test and, if so, if they had obtained one within the past year. Participants then were categorized as meeting the ACR guideline for BMD testing (19) if they had received a BMD test within the past year.

Counseling.

At the beginning of the telephone interview, participants were asked if they had heard of osteoporosis prior to being asked to participate in the current study; if they had ever been told that they are at increased risk for osteoporosis; and, if so, who provided this information. These introductory questions were followed by items assessing participants' recall of counseling they had received concerning osteoporosis prevention. Participants first were asked if they had ever received counseling or advice about osteoporosis prevention. Participants who answered affirmatively then were asked who gave the counseling/advice and what they were told about osteoporosis prevention.

Knowledge and belief variables.

The questionnaire included measures of 11 knowledge and belief variables. Five of these measures were specific to the behaviors of interest (i.e., calcium intake, vitamin D intake, and BMD testing). Of these, 3 assessed only beliefs about calcium intake, due to the lack of standardized measures for vitamin D intake and BMD testing. Perceived calcium barriers (e.g., “Calcium-rich foods are high in fat”) and perceived calcium benefits (e.g., “Eating calcium-rich foods would make my bones stronger”) were assessed by 8 and 6 items, respectively. Participant responses were recorded on a 5-point Likert scale with endpoints labeled 1 (strongly disagree) and 5 (strongly agree). Calcium self-efficacy was assessed by 3 items from the Osteoporosis Self-Efficacy Scale (28). Responses were recorded on a 10-cm visual analog scale with endpoints labeled “not at all confident” and “very confident.” Respondent ratings were measured to the nearest millimeter. Thus, responses ranged from 0 to 100. The remaining 2 behavior-specific variables were assessed by a single item for each behavior. All items used a 5-point Likert response scale. To assess perceived behavioral control, participants were asked how easy/hard it would be to perform the behavior in question. Response options ranged from 1 (very easy) to 5 (very hard). To assess subjective norms, participants were asked to indicate their level of agreement with the statement, “Most people who are important to me think I should (perform behavior).” Response options ranged from 1 (strongly disagree) to 5 (strongly agree).

The remaining 6 variables were not behavior specific. Osteoporosis concern (3 items, e.g., “In general, how concerned are you about osteoporosis?”) and perceived susceptibility to osteoporosis (3 items, e.g., “How likely do you think it is that you will ever get osteoporosis?”) were assessed using a 5-point Likert scale with endpoints labeled 1 (not at all) and 5 (extremely). Health motivation (5 items, e.g., “Keeping healthy is very important to me”), perceived severity of osteoporosis (6 items, e.g., “The thought of getting osteoporosis scares me”), and perceived controllability of osteoporosis (5 items, e.g., “Osteoporosis can be prevented”) were assessed using 5-point Likert scales with endpoints labeled 1 (strongly disagree) and 5 (strongly agree). Finally, osteoporosis knowledge was assessed by 20 true/false questions. Correct answers were summed to yield an index with scores ranging from 0 to 20. For all multi-item scales, with the exception of knowledge, a summary score was created by summing responses across items and dividing by the number of items answered. If more than one-third of items were missing, a summary score was not computed. Most of the multi-item measures were adapted from the Osteoporosis Health Belief Scale (29). All of the multi-item measures have been used in previous research and have been shown to possess desirable psychometric properties (30, 31).

Descriptive statistics for the knowledge and belief variables are shown in Table 1. Cronbach's α is also presented for each multi-item scale.

Table 1. Descriptive statistics for knowledge and belief measures*
 Mean ± SDα
  • *

    NA = not applicable; BMD = bone mineral density.

Health motivation4.01 ± 0.630.84
Controllability3.53 ± 0.590.66
Concern2.81 ± 0.860.73
Knowledge10.51 ± 3.920.80
Perceived severity3.10 ± 0.670.78
Perceived susceptibility3.66 ± 0.990.80
Calcium barriers2.28 ± 0.580.77
Calcium benefits3.80 ± 0.510.62
Calcium self-efficacy77.47 ± 19.400.95
Perceived behavioral control  
 Calcium1.78 ± 1.01NA
 Vitamin D1.75 ± 0.94NA
 BMD testing1.76 ± 1.05NA
Subjective norms  
 Calcium3.86 ± 0.92NA
 Vitamin D3.69 ± 0.91NA
 BMD testing3.54 ± 0.99NA

Background characteristics.

The mailed questionnaire assessed information concerning participant age, sex, race, education, marital status, employment status, and disease duration. Participants' rheumatologic diagnosis and current prednisone dose were collected during the eligibility interview.

Statistical analyses.

Descriptive statistics were used to characterize the sample. Bivariate relationships were examined using t-tests and logistic regression for numerical and categorical variables, respectively. To assess the independent effects of demographic characteristics, counseling, and knowledge and belief variables on behavior, a 3-step backward, hierarchical logistic regression analysis was performed for each behavior (i.e., calcium intake, vitamin D intake, and BMD testing). In step 1, 7 sociodemographic variables (i.e., age, sex, race, education, marital status, prednisone dose, and osteoporosis diagnosis) were included in the model. In step 2, an indicator variable indexing counseling status was added to the model. In step 3, the relevant knowledge and belief variables were added to the model. Within each step, variables were removed if they failed to reach statistical significance. All relationships were evaluated with α set at 0.05 using 2-tailed significance tests.

RESULTS

Sample characteristics.

The mean ± SD age of study participants was 55.1 ± 13.0 years. Most participants were women (76.7%), white (75.3%), married (64.9%), and employed either full time or part time (53.3%). Overall, participants were well educated with a mean of 14.3 ± 2.2 years of schooling. Almost all participants (91.2%) had a diagnosis of rheumatoid arthritis. Other rheumatologic diagnoses included systemic lupus erythematosus (n = 9), psoriatic arthritis (n = 4), and polymyalgia rheumatica (n = 3). Mean disease duration was 13.9 ± 10.3 years. Most participants (69.6%) took at least 5 mg of prednisone per day. A subset of participants (15.4%) reported that they had been diagnosed as having osteoporosis. Study participants did not differ from individuals who completed the screening interview but who failed to complete baseline data collection procedures with respect to age, sex, or presence of osteoporosis.

Counseling.

Almost all participants (97.4%) had heard of osteoporosis and slightly more than half (50.2%) reported having been told that they are at increased risk for developing osteoporosis. Participants most often cited their doctor as the source of this information (93.9%).

Approximately one-third of the participants (36.3%) reported that they had been counseled about osteoporosis prevention. In almost all cases (92.6%) participants reported that their physician had provided this counseling. As shown in Figure 1, participants reported that most counseling sessions included a discussion of the importance of obtaining an adequate amount of calcium. Other topics mentioned less frequently were use of bisphosphonates, vitamin D, weight-bearing exercise, BMD testing, and exercise with type unspecified.

Figure 1.

Osteoporosis prevention counseling topics in 227 patients.

As shown in Table 2, whether or not one reported having been counseled about osteoporosis prevention did not vary as a function of age, sex, marital status, prednisone dose, or disease duration. However, after controlling for other sociodemographic characteristics, patients were more likely to report having been counseled if they were white, reported having been diagnosed with osteoporosis, and had attended some college.

Table 2. Univariate and multivariate associations between sociodemographic characteristics and counseling status*
CharacteristicUnadjusted odds ratio95% CIPAdjusted odds ratio95% CIP
  • *

    Models predict the probability of being counseled about osteoporosis prevention. 95% CI = 95% confidence interval; NS = not significant.

Age ≥ 57 years0.990.58–1.72NS0.770.42–1.44NS
Married0.650.37–1.14NS0.620.33–1.16NS
Some college1.960.97–3.96NS2.451.12–5.360.03
Female1.130.59–2.17NS1.140.56–2.32NS
White2.951.42–6.130.013.011.37–6.620.01
Prednisone <5 mg/day0.800.44–1.47NS0.830.43–1.60NS
Osteoporosis2.881.34–6.200.012.841.22–6.600.02
Disease duration1.000.98–1.03NS1.000.97–1.03NS

Multivariate analysis of variance procedures revealed a significant overall association between counseling and patient knowledge and beliefs (F[15,204] = 3.38, P < 0.0001). As shown in Table 3, followup bivariate analyses found that, relative to patients who did not recall having been counseled about osteoporosis prevention, those who did report such counseling exhibited greater health motivation, concern about osteoporosis, osteoporosis knowledge, and perceived susceptibility to osteoporosis. Patients who reported having been counseled also reported fewer barriers to obtaining an adequate amount of calcium and stronger subjective norms regarding all 3 behaviors than did patients who did not report having been counseled. Finally, patients who reported having been counseled believed that it would be easier to obtain a BMD test than did patients who did not recall having been counseled.

Table 3. Relationship of counseling status to knowledge and beliefs*
Knowledge and belief measuresNot counseled (n = 138)Counseled (n = 82)P
  • *

    With the exception of knowledge and self-efficacy, all variables had a possible range of 1–5, with higher values reflecting higher levels of the attribute assessed. Knowledge had a possible range of 0 to 20, with higher values reflecting greater knowledge. Self-efficacy had a possible range of 0 to 100, with higher values reflecting greater self-efficacy. NS = not significant.

Health motivation3.924.19< 0.01
Osteoporosis knowledge and beliefs   
 Controllability3.493.61NS
 Concern2.683.00< 0.01
 Knowledge10.0311.60< 0.01
 Perceived severity3.113.08NS
 Perceived susceptibility3.543.89< 0.05
Behavior-specific beliefs   
 Calcium   
  Perceived barriers2.362.12< 0.01
  Perceived benefits3.793.85NS
  Self-efficacy78.6676.05NS
  Perceived behavioral control4.214.24NS
  Subjective norms3.704.15< 0.001
 Vitamin D   
  Perceived behavioral control4.204.32NS
  Subjective norms3.583.94< 0.01
 Bone mineral density testing   
  Perceived behavioral control4.094.52< 0.01
  Subjective norms3.383.80< 0.01

Behavior.

Slightly more than half of study participants (51.1%) were meeting the ACR guideline for calcium intake (19). Fewer were meeting the guideline for vitamin D intake (35.2%). With respect to BMD testing, 68.7% of participants had received a BMD test at some time in the past and, as recommended by the ACR guidelines, 33.5% had received a test within the past year. As shown in Figure 2, participants who reported having been counseled about osteoporosis prevention were significantly more likely than others to meet each of the guidelines (P < 0.05).

Figure 2.

Relationship between counseling status and whether participants were meeting the American College of Rheumatology Guidelines for calcium intake, vitamin D intake, and bone mineral density testing.

The results of analyses predicting whether or not participants' behavior met the ACR guidelines are presented in Table 4. Three findings are noteworthy. First, race and education were not associated with the practice of any of the behaviors, despite the association observed between these variables and recall of counseling. Second, individuals who reported having been counseled about osteoporosis prevention were more likely than others to be practicing each of the behaviors, even after controlling for sociodemographic characteristics. Third, perceived behavioral control was a significant predictor of all 3 behaviors, even after controlling for sociodemographic characteristics and counseling status. For each behavior, individuals who rated the behavior as easier to do were more likely than others to meet the ACR guideline for that behavior.

Table 4. Regression of behavior on demographic variables, counseling status, and knowledge and belief variables*
 CalciumVitamin DBMD testing
Odds ratioPOdds ratioPOdds ratioP
  • *

    NS = variable was not statistically significant at P < 0.05 in the model for that behavior. Variables that were not statistically significant in any of the models were: step 1, sex, race, education, and prednisone dose; step 3, perceived barriers, perceived benefits, self-efficacy, perceived susceptibility, perceived severity, health motivation, and perceived controllability. BMD = bone mineral density.

Step 1      
 Age ≥57 years1.92< 0.05NS 2.53< 0.01
 Married0.42< 0.010.53< 0.05NS 
 OsteoporosisNS NS 6.44< 0.0001
Step 2      
 Counseling received1.93< 0.052.78< 0.0012.93< 0.001
Step 3      
 Perceived behavioral control1.36< 0.051.55< 0.052.73< 0.0001
 Subjective norm1.42< 0.05NS NS 
 Knowledge1.08< 0.05NS NS 
 ConcernNS 1.53< 0.05NS 

DISCUSSION

The findings from this study are consistent with previous research suggesting that current practices aimed at the prevention of GIOP are insufficient. Only approximately one-third of study participants recalled having been counseled about osteoporosis prevention, and among those who recalled having been counseled, calcium intake was the only topic routinely remembered. Counseling status was not associated with prednisone dose; therefore, participants who were receiving higher doses of prednisone were not more likely to recall having been counseled. However, participants who belonged to minority groups and those with less formal education were less likely than others to recall having been counseled.

When interpreting these findings, a limitation of the study must be considered. That is, all of the data obtained were derived from patient self report. Thus, the low rates of patient counseling reported by study participants may be explained by 2 possibilities: either most patients had never been counseled about osteoporosis prevention or the counseling was not sufficient to facilitate subsequent recall. Consistent with the latter explanation, research in other areas suggests that patients often do not recall physician counseling about a wide variety of prevention topics (32, 33). For example, Flocke and Stange found that only 44% of patients with chronic illnesses remembered being counseled about diet immediately following the visit where such counseling was provided (33). Regardless of the explanation, the low rates of recall concerning counseling about osteoporosis prevention observed in the current study are an important warning signal. Patients cannot follow advice they do not remember. To facilitate patient recall, it may be necessary for health care providers to devote more time to the topic of osteoporosis prevention during their interactions with patients. In their study, Flocke and Stange found that time devoted to a particular topic was the strongest predictor of patient recall and that every 1-minute increase in the time devoted to a topic was associated with a 2.5-fold increase in patient recall (33).

With respect to behavior, slightly more than half of study participants were currently meeting the ACR guideline for calcium intake, and approximately one-third were meeting the guideline for vitamin D intake. Although more than two-thirds of participants reported having received a BMD test at some point in the past, only about one-third were meeting the ACR guideline for BMD testing, which recommends that repeat BMD tests be performed annually. As observed with counseling, whether or not participants were meeting the ACR guidelines was not associated with prednisone dose. Thus, those taking a higher prednisone dose were not more likely than others to be meeting the guidelines. In addition, whether or not participants were meeting the ACR guidelines was not associated with either race or education, despite the association between these sociodemographic characteristics and patient recall of counseling. Finally, for all 3 behaviors, individuals who reported having been counseled about osteoporosis were more likely than others to be meeting the guidelines.

These findings suggest that patient recall of counseling may facilitate guideline adherence. However, it seems likely that most patients would have been counseled about osteoporosis prevention at the time when a BMD test was ordered. Therefore, the observation that more than two-thirds of participants reported having had a BMD test at some point in the past, whereas only approximately one-third recalled having been counseled about osteoporosis prevention, underscores the possibility that many patients may have difficulty recalling that they have received counseling in the past.

We also examined the mechanisms through which recall of counseling may facilitate guideline adherence. Our findings suggest that counseling may affect patient behavior through effects on patients' knowledge and beliefs. We found, for example, that individuals who reported having been counseled exhibited greater knowledge about osteoporosis prevention than those who did not recall having been counseled. Individuals who reported having been counseled also exhibited beliefs likely to foster prevention efforts (e.g., greater concern about osteoporosis, beliefs in personal susceptibility to osteoporosis, perceived social pressure to adhere to prevention guidelines). However, the cross-sectional nature of this study makes it impossible to determine if patients' knowledge and beliefs changed in response to counseling. An alternative explanation is that patients with knowledge and beliefs consistent with osteoporosis prevention were more likely to remember having been counseled. Therefore, to determine the direction of causality operating, longitudinal research, preferably using an experimental design, is needed.

Finally, we found that even after controlling for recall of counseling, individuals who believed that meeting the guidelines was relatively easy were more likely than others to meet them. This observation underscores findings from previous research suggesting that overcoming perceived barriers to behavioral change is a critical step in the behavior change process (30, 34). Thus, a portion of counseling efforts should focus on identifying potential barriers to the adoption of recommended behaviors and helping patients develop strategies to overcome, minimize, or circumvent those barriers.

This study has a number of limitations that must be noted. First, participants were self-selected into the study. Only 16% of those contacted about the study returned the form indicating an interest in participating. This is not a true participation rate because many of those who did not return interest forms may have judged themselves to be ineligible (e.g., they did not have rheumatoid arthritis, or they had stopped taking prednisone). Although we cannot know the true participation rate, it is unlikely that study participants are representative of the population of interest. Instead, it seems likely that individuals who were more concerned about health issues in general, and osteoporosis in particular, would have been more likely to volunteer to participate in the study. This would tend to bias our findings in the direction of overestimating the true prevalence of counseling and patients meeting the ACR guidelines. Consistent with this view, in a review of patients' medical records, Solomon and colleagues recently reported that only 23% of patients studied had ever had a BMD test, whereas 33.5% of our patients reported having had such a test within the past year (18).

Second, we did not assess adherence with medications used to treat or prevent osteoporosis, such as bisphosphonates. This was a purposeful decision because medications are recommended for only a subset of patients (i.e., those with BMD T-scores <–1), whereas the guidelines that we focused on are relevant for all patients receiving oral glucocorticoids.

Finally, several studies have found that adherence to osteoporosis prevention guidelines is higher among patients treated by a rheumatologist compared with other specialties (13–15). It is likely that other physician practice characteristics may influence patient counseling and guideline adherence. However, we lacked sufficient information about participants' physicians to examine this issue.

Despite these limitations, the findings from this study, combined with those from past research, suggest that there is a need to evaluate interventions aimed at increasing adherence to the ACR guidelines. Findings from a recent study suggest that it will not be easy. That study, which tested an innovative intervention targeted toward physicians, failed to improve guideline adherence (35). Future efforts might focus on patients directly or on patient-physician dyads. In addition, innovative programs might target other health care providers involved in the medication use process. For example, pharmacists would be ideal candidates to provide counseling on the prevention of GIOP.

In conclusion, oral glucocorticoids remain an important class of medications for the management of inflammatory forms of arthritis. However, optimal long-term therapy requires careful attention to the prevention and treatment of GIOP. This study adds to the growing literature suggesting that insufficient attention is paid to this issue in clinical practice. Hopefully, future studies will demonstrate effective strategies for improving guideline adherence so that patients might receive the benefits of glucocorticoid therapy while minimizing their risk of developing osteoporosis as a consequence.

Ancillary