This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Cranney, A. Articles by Rosen, C. Search for Related Content PubMed PubMed Citation Articles by Cranney, A. Articles by Rosen, C.

IX: Summary of Meta-Analyses of Therapies for Postmenopausal Osteoporosis

	A. Abstract

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
This section summarizes the results of the seven systematic reviews of osteoporosis therapies published in this series [calcium, vitamin D, hormone replacement therapy (HRT), alendronate, risedronate, raloxifene, and calcitonin] and systematic reviews of etidronate and fluoride we have published elsewhere. We highlight the methodological strengths and weaknesses of the individual studies, and summarize the effects of treatments on the risk of vertebral and nonvertebral fractures and on bone density, including effects in different patient subgroups. We provide an estimate of the expected impact of antiosteoporosis interventions in prevention and treatment populations using the number needed to treat (NNT) as a reference. In addition to the evidence, judgements about the relative weight that one places on weaker and stronger evidence, attitudes toward uncertainty, circumstances of patients’ and societal values or preferences will, and should, play an important role in decision-making regarding anti-osteoporosis therapy.

	B. Introduction

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
POSTMENOPAUSAL OSTEOPOROSIS IS a major public health problem (1). Thirty to fifty percent of American women will, in the course of their lifetime, suffer a clinical fracture and the associated morbidity. Estrogen deficiency is a key factor in the pathogenesis of postmenopausal osteoporosis. Osteoporotic fractures are associated with substantial suffering; in particular, both vertebral and hip fractures are associated with increased mortality and diminished quality of life (1, 2). The prevalence of vertebral and hip fractures increases with advanced age. As a larger proportion of women survive into old age, osteoporotic fractures constitute a progressively greater burden of both illness and resource demand.

Although the World Health Organization definition of osteoporosis [bone mineral density (BMD) of at least 2.5 SD values below the mean for young adult women] has limitations, we do know that fracture risk is increased with decreasing BMD. Women who sustain a vertebral fracture represent a particularly vulnerable group whose risk of another vertebral fracture within the following year is higher by a factor of 3–5 (3, 4). In addition, the presence of a vertebral fracture is associated with an increased risk of hip fractures (5). The risk of any type of fracture is at least 2-fold higher among people who have had a previous fragility fracture of any type (4).

Controversial questions in the management of osteoporosis include the role of screening, selecting patients to whom treatment should be offered, and the type of treatments clinicians should recommend. New drugs for osteoporosis have increased the complexity of clinical decision making. Clinicians should base their recommendations about whom to treat, and with what therapy, on the most recent and comprehensive evidence. As discussed in Section I in this series, well-conducted systematic reviews can provide objective, comprehensive summaries of the evidence for the practicing clinician.

In this series, we presented the results of our meta-analyses of randomized trials of seven different osteoporosis therapies. Our goal was to provide the most precise pooled estimates of the magnitude of treatment effects for each therapy. We focused on vertebral and nonvertebral fractures, and on bone density. In this final section, we summarize the findings of these reviews. We highlight key results and provide our perspective on important issues in interpretation. In presenting the summary of our results here, we have included meta-analyses of etidronate (6) and fluoride (7).

	C. Methodological quality of the studies

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
We included only randomized controlled trials (RCTs) in our systematic review. For every trial, we evaluated four aspects of study design and conduct that have an impact on validity—that is the likelihood of biased results—of randomized trials (8). We chose the four methodological issues based on empirical evidence and a theoretical rationale suggesting a systematic impact on the magnitude of treatment effects. For three of these issues, concealment of randomization, blinding, and whether investigators included all patients in the groups to which they were randomized (intention-to-treat analysis), empirical evidence suggests a larger treatment effect in weaker studies (9, 10, 11). We also included completeness of follow-up, in which the direction of bias differs across studies. Table 1 summarizes the methodological quality of the studies included in each systematic review.

Concealment refers to whether those individuals responsible for determining eligibility knew which treatment participants would receive if they entered the trial. Trials that inadequately conceal patient allocation tend to overestimate treatment effects (9, 10, 11, 14).

In our systematic reviews, we noted one instance in which concealment appeared to have a systematic impact on the magnitude of the treatment effect. We found a large amount of variability among trials that assessed the effect of calcitonin on bone density of the lumbar spine and femoral neck. The effect of calcitonin on the lumbar spine was significantly greater in those trials that did not conceal treatment allocation vs. those that did (14.6% vs. 2.7%, P < 0.01) (see Section VI). Similarly, when we analyzed the impact of calcitonin at the femoral neck, there were significantly larger treatment effects noted with trials that did not conceal allocation (10.0% vs. -0.8%, P = 0.05) (see Section VI).

The validity of the alendronate trials proved the strongest with respect to blinding, concealed allocation and follow-up (Table 1 and Section II). Near-complete follow-up of randomized patients ensures that the integrity of the randomization is maintained. Substantial loss to follow-up, even if similar in the treatment arms, threatens trial validity. A differential loss to follow-up in the control vs. treatment arms poses further risk of an imbalance in risk factors in patients for whom outcome data are available, and can result in biased estimates of the treatment effect (15). We selected three cut points for loss to follow-up: 1%, 5%, and over 20%.

Loss to follow-up was the major methodological weakness in these trials (Table 1). Although the majority of trials suffered between 5 and 20% loss to follow-up, a number of the larger trials lost over 20% of patients (Refs. 6, 16 , and 17 , and Section III). Calcitonin, risedronate, and etidronate trials had particularly large losses to follow-up. Loss to follow-up can create bias in either direction, spuriously increasing or decreasing the magnitude of the treatment effect. Large losses to follow-up may bias against the treatment if individuals at greater risk of fracture are lost preferentially from the control arm (Section III). This appears to be the case for one of the larger vertebral fracture risedronate trials (17). In this study, patients lost to follow-up in the placebo arm had a higher prevalence of vertebral fractures at the time they began the study than other patient groups (including other placebo-arm patients, and those lost to follow-up in the treatment arm).

The fourth aspect of study validity we examined in each of the trials was whether the investigators applied the intention-to-treat principle in their analyses (18). The intention-to-treat principle dictates the inclusion of all participants in the group to which they were allocated independent of whether or not they received treatment or completed the trial (18). There was considerable variability in the number of trials that respected the intention-to-treat principle. For example, only 1 of 15 calcium trials and 4 of 30 calcitonin trials analyzed all patients in the groups to which they were allocated (Table 1). More recent RCTs consistently followed the intention-to-treat principle. We found only one instance in which failure to apply the intention-to-treat principle appeared to be associated with an inflation of the treatment effect. This occurred in the etidronate meta-analysis in the evaluation of lumbar spine BMD (6).

Optimal methodology in osteoporosis trials would include blinding of all relevant groups, including the data analysts; application of strategies to ensure adequate concealment of randomization; applying the intention-to-treat principles to the analysis; and, in particular, ensuring minimal loss to follow-up. Trials should explicitly describe the procedures and results undertaken in each area.

	D. Comparison of treatment effects between meta-analyses

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
In the next paragraphs, we present tables that juxtapose the apparent impact of osteoporotic treatment on vertebral and nonvertebral fractures and bone density. It would be natural to directly compare these different apparent effects. We discourage readers from making strong inferences about the relative impact of therapies on the basis of data from our summary tables for the following reasons. First, readers should note the confidence intervals (CIs) around the magnitude of treatment effect sometimes overlap. If this occurs, apparent differences in the point estimates may not reliably indicate true underlying differences in the magnitude of the effect.

Second, even if CIs are nonoverlapping, between-trial comparisons of treatment effects are unreliable (19). Patient populations may differ in their responsiveness to treatment effect because of differences in bone density, prevalent fractures, postmenopausal status, co-interventions, and co-morbidity. Thus, an apparently more effective treatment may actually have been tested in a more responsive population. The doses of the different therapies may not be comparable. Differences in administration of other interventions may impact on the apparent effectiveness of antiosteoporosis drugs. For instance, many of the recent RCTs of antiresorptive agents have included calcium and vitamin D in the both the treatment and control arm. In contrast, many of the earlier HRT trials did not include calcium and vitamin D.

Furthermore, the methodological differences, rather than the relative treatment potency, may explain apparently different treatment effects. Such differences may include differential loss to follow-up and methods used to analyze incident fractures.

These considerations suggest that secure conclusions about the relative effectiveness of different osteoporosis therapies must await results of head-to-head comparisons in randomized trials. The relative merit of the therapies that are clearly effective are likely, therefore, to remain speculative, and their use determined more often by their side effect profile, cost, convenience, and possibly the impact of commercial marketing efforts.

	E. Vertebral fractures—results

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
Table 2 presents a summary of the magnitude of the pooled treatment effect of different therapies on vertebral fractures. Calcium (see Section VII), fluoride (7), and HRT (see Section V) showed trends toward reduction in vertebral fractures. However, in each case, the CI overlaps 1.0, indicating that the data have not excluded a null or even detrimental effect with these agents. Thus, inferences about the impact of these agents on vertebral fracture are weak.

The variability of results from trial to trial reflected in the P value for heterogeneity is the fourth variable to examine. The alendronate, risedronate, and etidronate results are all extremely consistent from study to study, and the vitamin D results are relatively consistent.

For both raloxifene and calcitonin, we have chosen to present the results of the largest study, rather than the pooled results of all available studies. For raloxifene, the ratio of patients in the larger to the smaller study was approximately 150:1, and the disparate results suggest that the larger study provides the only trustworthy estimate. For calcitonin, the pooled estimate (relative risk 0.46, 95% CI 0.25–0.87) is driven by the results of 3 small trials with a sample size of 45–164 and very large relative risk reductions (20, 21, 22). This raises concern about possible publication bias. In addition, the random-effects models we chose for our analyses because of its theoretical appeal and its generally wider, and thus more conservative, CIs gives relatively larger weight to small studies in comparison to fixed-effect models. In this case, in which smaller studies have yielded larger effects, the point estimate of relative risk are driven downward, potentially inflating the treatment effect. In response to these considerations, Table 2 presents the 21% relative risk reduction suggested by PROOF (16) rather than the 54% suggested by the pooled estimate.

For prevention (women in the normal or near normal range of BMD) vs. treatment populations (women with osteoporosis), the pooled relative risks for vertebral fractures were similar for most drugs (for example, with alendronate, relative risks were, respectively, 0.45 vs. 0.53, P value on difference between estimates 0.87) (see Section II). Differences between the relative risk reduction in prevention and treatment populations did not reach statistical significance for any of the therapies, though typically there were few events in the prevention populations, considerably limiting the power of these analyses.

With many of the therapies, it was difficult to adequately evaluate the effect of dose on vertebral fractures because of sparseness of data. With alendronate, the treatment effect was similar across doses.

	F. Nonvertebral fractures—results

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
Table 3 summarizes the magnitude of the treatment effect on nonvertebral fractures for each therapy. Alendronate and risedronate were the only two therapies that had a significant pooled treatment effects on nonvertebral fracture reduction, reflected in CIs around the relative risk that do not overlap 1.0 (that is, the CIs exclude an effect size of 0).

Although we had hoped to estimate the magnitude of the relative risk reduction in prevention vs. treatment trials, the very small number of events in the former studies prevent us from making any but the weakest inference. Thus, we view our finding of a similar treatment effect on nonvertebral fractures in the prevention vs. treatment trials, respectively, for alendronate (point estimate 0.79 vs. 0.49, P value for the difference = 0.40) and risedronate (0.49 vs. 0. 73, P = 0.58) with skepticism. We are particularly cautious because we did not have access to individual patient data, and analyses conducted by the investigators of the primary studies support a different conclusion. These analyses suggest a smaller relative risk reduction in nonvertebral fractures in patients with higher, vs. those with lower, bone density (23). Thus, the impact of bisphosphonates on nonvertebral fracture reduction in low-risk populations remains questionable. We noted larger treatment effects on nonvertebral fractures with larger doses of alendronate with a pooled relative risk of 0.51 in the 10- to 40-mg dose and 0.87 in the 5-mg trials (P value of the difference <0.01).

For a single treatment, alendronate, we had data on fracture site in a sufficient number of patients that we were able to estimate the relative effect on the incidence of osteoporotic vs. nonosteoporotic fractures. Osteoporotic fractures were defined by using a prior study (24) that indicated an association between low calcaneal bone density and the particular type of fracture (relative risk of fracture 1.5 or greater). These sites included forearm, hip, rib, leg, patella, pelvis, and hands (osteoporotic fractures) vs. all fractures in which the risk was less than 1.5 (nonosteoporotic fractures). With 10–40 mg of alendronate, the pooled relative risk was 0.46 for osteoporotic and 0.57 for the nonosteoporotic fractures (see Section II). The relative risk reductions were larger with both types of fractures with the 10- to 40-mg of alendronate dose in comparison to the 5-mg dose.

The treatment effects were very similar with alendronate across all fracture types, and thus very similar for hip fractures vs. other nonvertebral fractures. The consistent effect of alendronate on osteoporotic and nonosteoporotic fractures supports applying the pooled relative risk estimate of 0.51 and the associated CIs to all types of nonvertebral fractures. In particular, the consistency of results across fracture type suggests that clinicians should apply the pooled relative risk reductions for nonvertebral fractures to hip fractures. Although we found similar results with other treatments, only with alendronate did our analyses have sufficient power to fully explore this issue. Thus, on the basis of our analyses, the inference that pooled nonvertebral fracture reduction relative risks apply to all such fractures is stronger for alendronate than other treatments.

	G. Absolute differences in event rates

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
The results, as we have presented them, do not convey the absolute differences in event rates that women might expect with antiosteoporotic treatment. Table 4 presents the absolute risk of fracture in prevention and treatment populations, and estimates of the number of patients clinicians must treat for a period of 2 yr to prevent a fracture with each effective therapy. Table 4 includes only therapies in which the CI excludes no effect. Thus, calcium, calcitonin (using the PROOF data only), HRT, and fluoride do not appear in Table 4. As described above, we feel sufficiently unsure about the relative risk reduction for nonvertebral fractures in low-risk populations that we have not provided estimates of NNT in this patient group. The NNT would be smaller over longer periods of time, as well as in individuals from higher risk groups.

We provide the example of the effect of vitamin D on vertebral fractures in low-risk patients to show how we made the calculations. Low-risk patients have a risk of 0.12% (12 in 10,000) of a vertebral fracture over a period of 2 yr. The pooled estimate of randomized trial results shows a relative risk reduction of 37% with vitamin D (Table 2). Thus, the absolute risk reduction with vitamin D is 0.0012 x 0.37 = 0.000444 (0.04% or 4 in 10,000). The NNT is the inverse of the absolute risk reduction or 1/0.000444 = 2252.

The boundary of the CI consistent with the largest plausible relative risk reduction associated with vitamin D is 55%. Were this the true relative risk reduction, the absolute risk reduction in vertebral fractures with vitamin D in the low-risk population would be 0.0012 x 0.55 = 0.00066. The NNT associated with this absolute risk reduction would be 1/0.00066 or 1515.

These calculations assume a constant relative risk reduction in vertebral fracture across low and high-risk populations. As we have mentioned, the data support this assumption. In nonvertebral fractures, where the assumption of constant relative risk is uncertain, we have not calculated the NNT.

	H. Bone density—results

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
Table 5 presents a summary of the pooled treatment effects on lumbar spine bone density vs. placebo. The largest treatment effects on the lumbar spine were seen with alendronate (10–40 mg) and HRT, with intermediate effects seen with risedronate and etidronate. All these drugs show quite narrow confidence intervals around the treatment effects, but study-to-study variability in results was almost uniformly significant.

Longer treatment durations with alendronate, risedronate, raloxifene and HRT resulted in larger treatment effects on bone density of the lumbar spine. This was not the case for vitamin D, calcitonin, or calcium.

The effect of standard vitamin D on bone density was much smaller than for hydroxylated vitamin D. There was a significant difference in bone density for total body (0.40 vs. 2.06, P value of the difference 0.03) and combined forearm (-0.48 vs. 5.53, P = 0.01) between standard and hydroxylated vitamin D preparations.

In our a priori hypotheses, we stated that the magnitude of the treatment effects might differ depending on use of concurrent therapies with calcium or vitamin D. With higher doses of calcium (>500 mg), the effect on BMD of the lumbar spine and total body was greater in those individuals treated with vitamin D (standard dose). Alendronate trials with a higher recorded calcium intake (dietary and supplementation combined) demonstrated a greater treatment effect on lumbar spine BMD than those with a lower calcium intake.

	I. Management implications

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 
Although evidence should play a crucial role in clinical decision-making, evidence alone never completely informs any decision (25). Other facets of the decision-making process include judgements about the relative weight that one places on weaker and stronger evidence, attitudes toward uncertainty, circumstances, and, above all, values or preferences. Decisions may differ depending on one’s perspective. Consider the following individuals: a patient over 65 whose drugs are largely covered by a government drug plan; an employee whose drugs are partially covered by a work benefit package; an indigent person paying out of pocket for her drugs; and a government decision-maker determining what coverage to provide to citizens within her jurisdiction. Facing the same evidence, these individuals may well make different decisions about optimal management of osteoporosis.

Because the relationship between evidence and optimal decision-making is so complex, and so highly dependent on values or preferences, we have avoided commenting on the treatment implications of the results of our individual reviews. In this concluding article, however, we reflect on some of the issues we believe decision-makers should ponder.

First, how might one best summarize the evidence from our reviews? First, it is likely that a number of drugs reduce vertebral fractures. These include vitamin D (hydroxylated), calcitonin, raloxifene, and the bisphosphonates etidronate, risedronate, and alendronate. The inferences regarding reduction of vertebral fractures are, on the basis of the methodological quality of the studies, the magnitude of the treatment effect, narrowness of the confidence intervals, and the consistency of the results from study to study, strongest for alendronate and risedronate.

HRT may also reduce vertebral fractures. However, the CI around the pooled estimate from randomized trials still includes the possibility of an increase in fracture rate with HRT (Table 2). Evidence suggesting an appreciable relationship between changes in lumbar spine bone density and the magnitude of vertebral fracture reduction suggest that HRT will ultimately prove to have a large positive impact on vertebral fracture incidence (25, 26, 27, 28, 29).

Further support for the inference that subsequent trials will show a large reduction in vertebral fracture with HRT comes from the large body of case-control and cohort studies that have suggested a 25–50% reduction in hip fractures and 50% reduction in vertebral fractures in HRT users vs. nonusers (30, 31). However, even well-conducted cohort studies with appropriate adjustment for prognostic factors may produce misleading results because of differences in unmeasured or unmeasurable determinants of outcome in experimental and control groups. For instance, whereas observational studies suggest a HRT results in a large reduction in cardiovascular risk with HRT, data from randomized trials suggest no impact on cardiovascular risk whatsoever (32). Large RCTs of HRT such as the Womens’ Health Initiative will help to clarify the impact of HRT on fracture reduction (33). The majority of the evidence for the efficacy of HRT is based on results of case-control and cohort studies. These studies have suggested a 25–50% reduction in hip fractures and 50% reduction in vertebral fractures in HRT users versus nonusers (30, 31). However, even well-conducted cohort studies that adjust for confounders are subject to selection bias (healthy users bias) and may overestimate the magnitude of the treatment effect.

Our systematic reviews provide convincing evidence for nonvertebral fracture reduction for only two agents, risedronate and alendronate (Table 3). Randomized trials suggest that etidronate and raloxifene probably have only small, if any, effects on nonvertebral fracture. The sparse data for calcium and calcitonin provide little information. Although there is an appreciable trend suggesting that HRT reduces nonvertebral fracture, the confidence interval overlaps no effect. The experience with raloxifene, which reduces vertebral but not nonvertebral fractures, suggests continued caution in making any assumptions about the effect of HRT on nonvertebral fractures.

As we have mentioned earlier, observational studies have suggested that bone density is associated with fracture risk. Based on the results of our meta-analyses small to intermediate changes in BMD may be associated with substantial reductions in vertebral fractures as seen with vitamin D and raloxifene. When we explored the relationship between bone density and fracture through a regression analysis using data from these meta-analyses, we found that BMD was helpful in predicting the impact of therapies on vertebral but not nonvertebral fractures (34). These results provide further support for the inference that HRT reduces vertebral fractures, and reinforces caution concerning the inference that HRT reduces nonvertebral fractures.

What of the magnitude of the treatment effect? Relative risk reductions are of the order of one half for alendronate, both for vertebral and nonvertebral fractures. The relative risk reductions for risedronate are slightly more than one third for vertebral fractures, and one quarter for nonvertebral fractures. We have already cautioned against strong inferences that the effect of alendronate is greater than rised-ronate, both because of the overlapping confidence intervals and because of the general problems making inferences from indirect, rather than head-to-head, comparisons of drugs.

Decision-makers should also pay careful attention to issues of absolute risk. Patients whose absolute risk is low can expect, at best, small absolute benefits from treatment. Patients at higher risk can anticipate much greater absolute benefits, and may thus be willing to tolerate more in the way of inconvenience, costs, or medication-induced side effects. The number of patients one must treat to prevent a vertebral or nonvertebral fracture is one way of capturing the absolute impact of treatment. Clinicians will want to take in to account the NNTs for osteoporosis therapy summarized in Table 4 when making recommendations to their patients. The absolute risk of harm (and the corresponding number of patients one needs to treat to cause a toxic effect in one patient) also warrants careful consideration.

Attitudes toward strength of evidence may influence decision-making. Those who are inclined to credit the results of observational studies may continue to believe that HRT leads to a substantial reduction in cardiac events, including myocardial infarction and death (35). Those who believe the best treatment estimates come from randomized trials will, on the basis of available data, assume that HRT has a negligible effect on secondary prevention of cardiovascular risk (32).

For an individual woman with postmenopausal osteoporosis, many factors weigh in the final treatment decision. The strength of the evidence, additional benefits, risks, adverse effects, and price associated with different medications all weigh heavily in treatment decisions. When one considers the magnitude of effect on vertebral or nonvertebral fractures for the various medications, the bisphosphonates alendronate and risedronate have strong evidence of their efficacy. Other treatment options such as HRT, vitamin D, or calcitonin do not meet as stringent criteria as those for alendronate or risedronate. Those who would choose a treatment with proven impact, and who also feel that nonvertebral fracture is the most important outcome, will have little difficulty choosing alendronate or risedronate. The results of future RCTs may generate new efficacy estimates and so the inclusion of these trials in the pooled results may yield different estimates of the magnitude of treatment efficacy of HRT.

One’s attitude toward different adverse health events of therapies may influence a treatment decision. For instance, women who have serious concerns regarding breast cancer and less concern about avoidance of osteoporotic nonvertebral fractures might be less willing to take HRT and prefer raloxifene. In addition, women who are concerned about the potential increased risk of a venous thromboembolism may chose not to take either HRT or raloxifene.

Attitudes toward inconvenience associated with therapy may influence patient choice. For instance, patients prescribed the regimens tested in the clinical trials included in our analyses must take their daily alendronate while sitting up, or upright for at least 30 min before a meal, as a strategy for avoiding esophageal ulceration. Those who find such a requirement inconvenient may be inclined to choose another agent or dosing regimen. Such issues also highlight the challenge facing clinicians in encouraging long-term adherence with osteoporosis treatments (36, 37).

Values and preferences will also bear heavily on price sensitivity. Those reluctant to spend resources on preventing osteoporotic fractures may be more inclined to consider traditional HRT rather than a bisphosphonate. Attitudes toward cost may also influence the decision regarding when to begin therapy with a bisphosphonate, and how long one continues. Major uncertainties remain concerning the impact of alendronate and risedronate as well as all other therapies on nonvertebral fractures in low risk women, and on the effects of long-term therapy. Those who are more price-sensitive would be inclined to start treatment later, and discontinue therapy sooner.

These management implications highlight the continued uncertainties regarding therapy for osteoporosis. The most important unanswered questions include the impact of HRT on vertebral and nonvertebral fractures, the relative impact of different therapies on vertebral and nonvertebral fractures, the optimal duration of therapy with antiresorptive agents, and the magnitude of the impact of therapies to reduce vertebral fracture on health-related quality of life. Answering these questions should be a priority for future research. It is evident that decisions regarding treatment of osteoporosis remain complex, challenging, and fraught with uncertainty. It is also clear that data from randomized trials have provided great insights into the strength of the evidence for effectiveness of different therapies. Our systematic reviews have clarified what we know, and what remains in question. Those responsible for recommending management strategies for osteoporosis should take full advantage of these data.

	Footnotes

 
Abbreviations: BMD, Bone mineral density; CI, confidence interval; HRT, hormone replacement therapy; NNT, number needed to treat; RCT, randomized controlled trial.

	References

Top A. Abstract B. Introduction C. Methodological quality of... D. Comparison of treatment... E. Vertebral... F. Nonvertebral... G. Absolute differences in... H. Bone density—results I. Management implications References

 

1. Cauley JA 2000 Risk of mortality following clinical fractures. Osteoporos Int 11:556–561[CrossRef][Medline]
2. Kado DM, Browner WS, Palmero L, Nevitt MC, Genant HK, Cummings SR 1999 Vertebral fractures and mortality in older women. Study of Osteoporotic Fractures Research Group. Arch Intern Med 159:1215–1220[Abstract/Free Full Text]
3. Lindsay R, Silverman SL, Cooper C, Hanley DA, Barton I, Broy S, Licata A, Benhamou L, Geusens P, Flowers K, Stracke H, Seeman E 2001 Risk of new vertebral fracture in the year following a fracture. JAMA 285:320–323[Abstract/Free Full Text]
4. Klotzbuecher CM, Ross PD, Landsman PB, Abbott T, Berger M 2000 Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 4:721–739
5. Black DM, Arden NK, Palmero L, Pearson J, Cummings SR 1999 Prevalent vertebral deformities predict hip fractures and new vertebral deformities but not wrist fractures. J Bone Miner Res 14:821–828[CrossRef][Medline]
6. Cranney A, Guyatt G, Griffith L, Krolicki N, Welch V, Adachi JD, Shea B, Tugwell P, Wells G 2001 A meta-analysis of etidronate for the treatment of postmenopausal osteoporosis. Osteoporos Int 12:140–151[CrossRef][Medline]
7. Haguenauer D, Welch V, Shea B, Tugwell P, Adachi JD, Wells G 2000 Fluoride for the treatment of postmenopausal osteoporotic fractures: a meta-analysis. Osteoporos Int 11:727–738[CrossRef][Medline]
8. Guyatt G, Cook D, Devereaux PJ, Meade M, Straus S 2001 Therapy. In: Guyatt G, Rennie D, eds. Users’ guides to the medical literature: a manual for evidence based practice. Chicago: AMA Press; 55–79
9. Schulz KF, Chalmers I, Hayes RJ, Altman DG 1995 Empirical evidence of bias. Dimensions of methodological quality assoicated with estimates of treatment effects in controlled trials. JAMA 273:408–412[Abstract]
10. Chalmers TC, Celano P, Sacks HS, Smith Jr H 1983 Bias in treatment assignment in controlled trials. N Engl J Med 309:1358–1361[Abstract]
11. Moher D, Pham B, Jones A, Cook D, Jadad AR, Moher M, Tugwell P, Klassen TP 1998 Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses. Lancet 352:609–613[CrossRef][Medline]
12. Devereaux PJ, Manns BJ, Ghali WA, Quan H, Lachetti C, Montori VM, Bhandari M, Guyatt GH 2001 Physician interpretations and textbook definitions of blinding terminology in randomized controlled trials. JAMA 285:2006–2007[Free Full Text]
13. Begg C, Cho M, Eastwood S, Horton R, Moher D, Olkin I, Pitkin R, Rennie D, Schulz KF, Simel D, Stroup DF 1996 Improving the quality of reporting or randomised controlled trials. The CONSORT statement. JAMA 276:637–639[CrossRef][Medline]
14. Moher D, Schulz KF, Altman DG 2001 The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. Ann Intern Med 134:657–662[Abstract/Free Full Text]
15. Guyatt GH, Sackett DL, Cook 1993 Users’ guide to the medical literature. II How to use an article about therapy or prevention. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 270:2598–2601[Free Full Text]
16. Chesnut III CH, Silverman S, Andriano K, Genant H, Giomona A, Harris S, Kiel D, LeBoff M, Maricic M, Miller P, Moniz C, Peacock M, Richardson P, Watts N, Baylink D 2000 A randomized trial of nasal spray salmon calcitonin in postmenopausal women with established osteoporosis: the prevent recurrence of osteoporotic fractures study. Am J Med 109:267–276[CrossRef][Medline]
17. Harris ST, Watts NB, Genant HK, McKeever CD, Hangartner T, Keller M, Chesnut III CH, Brown J, Eriksen EF, Hoseyni MS, Axelrod DW, Miller PD 1999 Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis A randomized controlled trial. JAMA 282:1344–1352[Abstract/Free Full Text]
18. Guyatt G, Devereaux PJ 2001 Therapy and validity. In: Guyatt G, Rennie D, eds. Users’ guide to the medical literature: a manual for evidence-based clinical practice. Chicago: AMA Press; 267–273
19. Bucher HC, Guyatt GH, Griffith L, Walter SD 1997 The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol 50:683–691[CrossRef][Medline]
20. Gennari C, Agnusdei D, Montagnami M, Gonneli S, Civitelli R 1992 An effective regimen of intranasal salmon calcitonin given in early postmenopausal bone loss. Calcif Tissue Int 50:381–383[CrossRef][Medline]
21. Overgaard K, Hansen MA, Jensen SB, Christiansen C 1992 Effect of salcalcitonin given intranasally on bone mass and fracture rated in established osteoporosis: a dose-response study. BMJ 305:556–561
22. Hizmetli S, Elden H, Kaptanoglu E, Nacitarhan V, Kocagil S 1996 The effect of different doses of calcitonin on bone mineral density and fracture risk in postmenopausal osteoporosis. Int J Clin Pract 52:453–455
23. Cummings SR, Black DM, Thompson DE, Applegate WB, Barrett-Connor E, Musliner TA, Palermo L, Prineas R, Rubin SM, Scott JC, Vogt T, Wallace R, Yates JA, LaCroix AZ 1998 Alendronate reduces the risk of vertebral fractures in women without pre-existing vertebral fractures: results of the Fracture Intervention Trial. JAMA 280:2077–2082[Abstract/Free Full Text]
24. Seeley DG, Browner WS, Nevitt MC, Genant HK, Scott J, Cummings SR 1991 Which fractures are associated with low appendicular mass in elderly women? The Study of Osteoporotic Fractures Research Group. Ann Intern Med 115837–42
25. Guyatt G, Straus S, McAlister F, Haynes B, Sinclair J, Devereaux PJ, Lacchetti C 2001 Moving from evidence to action. In: Guyatt G, Rennie D, eds. Users’ guide to the medical literature: a manual for evidence-based clinical practice. Chicago: AMA Press; 567–657
26. Wasnich RD, Miller PD 2000 Antifracture efficacy of antiresorptive agents are related to changes in bone density. J Clin Endocrinol Metab 85:231–236[Abstract/Free Full Text]
27. Black DM, Sarkar S, Mitlak B, Knickerbocker R, Cummings SR 1999 What proportion of the effects of raloxifene (RLX) on vertebral fracture risk can be directly attributed to its bone mineral density (BMD) effects? J Bone Miner Res 4(Suppl 1):S158 (Abstract 1101)
28. Black DM, Pearson JB, Karpf DB, Harris F, Genant HK, LaCroix AZ, How much of the reduction in risk of vertebral fractures by alendronate is explained by increased spine BMD? ASBMR 21^st Annual Meeting, St. Louis, MO, 1999 (Abstract 1107)
29. Li Z, Meredith MP, Hoseyni MS 2001 A method to assess the proportion of treatment effect explained by a surrogate endpoint. Stat Med 20:3175–3188[CrossRef][Medline]
30. Grady D, Rubin SM, Petitti DB, Fox CS, Black D, Ernster VL 1992 Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med 117:1016–1037
31. Cauley JA, Seeley DG, Ensrud K, Ettinger B, Black D, Cummings SR 1995 Estrogen replacement therapy and fractures in older women: a prospective study. Ann Intern Med 122:9–16[Abstract/Free Full Text]
32. Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, Vittinghoff E 1998 Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. Heart and Estrogen/progestin Replacement study (HERS) Research Group. JAMA 280:605–613[Abstract/Free Full Text]
33. The Women’s Health Initiative Group 1998 Design of the Women’s Health Initiative clinical trial and observational study. Control Clin Trials 19:61–109[CrossRef][Medline]
34. Guyatt G, Cranney A, Griffith L, Walter S, Krolicki N, Favus M, Rosen C, Summary of meta-analyses of therapies for postmenopausal osteoporosis and the relationship between bone density and fractures. Endocrinol Metab Clin North Am, in press
35. Stampfer MJ, Colditz GA 1991 Estrogen replacement therapy and coronary heart disease: a quantitative assessment of the epidemiologic evidence. Prev Med 20:47–63[CrossRef][Medline]
36. Pressman A, Forsyth B, Ettinger B, Tosteson ANA 2001 Initiation of osteoporosis treatment after bone mineral density testing. Osteoporos Int 12: 337–42
37. Faulkner DL, Young C, Hutchins D, McCollam JS 1998 Patient noncompliance with hormone replacement therapy: a nationwide estimate using a large prescription claims database. Menopause 5:226–229[Medline]

This article has been cited by other articles:

				B. Sander, V. Elliot-Gibson, D. E. Beaton, E. R. Bogoch, and A. Maetzel A Coordinator Program in Post-Fracture Osteoporosis Management Improves Outcomes and Saves Costs J. Bone Joint Surg. Am., June 1, 2008; 90(6): 1197 - 1205. [Abstract] [Full Text] [PDF]

				S. M. Cadarette, J. N. Katz, M. A. Brookhart, T. Sturmer, M. R. Stedman, and D. H. Solomon Relative Effectiveness of Osteoporosis Drugs for Preventing Nonvertebral Fracture Ann Intern Med, May 6, 2008; 148(9): 637 - 646. [Abstract] [Full Text] [PDF]

				A. M. Cheung and A. S. Detsky Osteoporosis and Fractures: Missing the Bridge? JAMA, March 26, 2008; 299(12): 1468 - 1470. [Full Text] [PDF]

				S. R. Majumdar MD MPH, J. A. Johnson PhD, F. A. McAlister MD MSc, D. Bellerose BScN, A. S. Russell MD, D. A. Hanley MD, D. W. Morrish MD PhD, W. P. Maksymowych MD, and B. H. Rowe MD MSc Multifaceted intervention to improve diagnosis and treatment of osteoporosis in patients with recent wrist fracture: a randomized controlled trial Can. Med. Assoc. J., February 26, 2008; 178(5): 569 - 575. [Abstract] [Full Text] [PDF]

				M. B. Herndon, L. M. Schwartz, S. Woloshin, and H. G. Welch Implications Of Expanding Disease Definitions: The Case Of Osteoporosis Health Aff., November 1, 2007; 26(6): 1702 - 1711. [Abstract] [Full Text] [PDF]

				S. R. Majumdar, L. A. Beaupre, C. H. Harley, D. A. Hanley, D. A. Lier, A. G. Juby, W. P. Maksymowych, J. G. Cinats, N. R. Bell, and D. W. Morrish Use of a Case Manager to Improve Osteoporosis Treatment After Hip Fracture: Results of a Randomized Controlled Trial Arch Intern Med, October 22, 2007; 167(19): 2110 - 2115. [Abstract] [Full Text] [PDF]

				P. F. Lata and M. E. Elliott Patient Assessment in the Diagnosis, Prevention, and Treatment of Osteoporosis Nutr Clin Pract, June 1, 2007; 22(3): 261 - 275. [Abstract] [Full Text] [PDF]

				M. J. Bolland, A. B. Grey, A. M. Horne, S. E. Briggs, M. G. Thomas, R. B. Ellis-Pegler, A. F. Woodhouse, G. D. Gamble, and I. R. Reid Annual Zoledronate Increases Bone Density in Highly Active Antiretroviral Therapy-Treated Human Immunodeficiency Virus-Infected Men: A Randomized Controlled Trial J. Clin. Endocrinol. Metab., April 1, 2007; 92(4): 1283 - 1288. [Abstract] [Full Text] [PDF]

				N. D. Nguyen, J. A. Eisman, J. R. Center, and T. V. Nguyen Risk Factors for Fracture in Nonosteoporotic Men and Women J. Clin. Endocrinol. Metab., March 1, 2007; 92(3): 955 - 962. [Abstract] [Full Text] [PDF]

				W F Lems Clinical relevance of vertebral fractures Ann Rheum Dis, January 1, 2007; 66(1): 2 - 4. [Full Text] [PDF]

				J. B. Cannata-Andia, M. Rodriguez-Garcia, N. Carrillo-Lopez, M. Naves-Diaz, and B. Diaz-Lopez Vascular Calcifications: Pathogenesis, Management, and Impact on Clinical Outcomes J. Am. Soc. Nephrol., December 1, 2006; 17(12_suppl_3): S267 - S273. [Abstract] [Full Text] [PDF]

				E. M. Lewiecki and A. J. Laster Clinical Applications of Vertebral Fracture Assessment by Dual-Energy X-Ray Absorptiometry J. Clin. Endocrinol. Metab., November 1, 2006; 91(11): 4215 - 4222. [Abstract] [Full Text] [PDF]

				D. T Gold, I. M Alexander, and M. P Ettinger How Can Osteoporosis Patients Benefit More from Their Therapy? Adherence Issues with Bisphosphonate Therapy Ann. Pharmacother., June 1, 2006; 40(6): 1143 - 1150. [Abstract] [Full Text] [PDF]

				{blacktriangledown}Strontium ranelate for osteoporosis? DTB, April 1, 2006; 44(4): 29 - 32. [Abstract] [Full Text] [PDF]

				C. J. Rosen Postmenopausal Osteoporosis N. Engl. J. Med., August 11, 2005; 353(6): 595 - 603. [Full Text] [PDF]