3 - Example


3 Overview

3.1 Introduction

This chapter provides an in-depth walk-through of a worked example detailing the clinical-effectiveness of pharmacological interventions for preventing fractures.

3.1.2 In this chapter

Overview

  • Step 1: Effect Measure and Number of Treatments
  • Step 2: Comparing Alendronate with Placebo
  • Step 2a: Weighted Effect Measure (1,2)
  • Step 3: Comparing Etidronate with Placebo
  • Step 3a: Weighted Effect Measure (2,3)
  • Step 4: Calculating Results

3.2 Overview

Osteoporosis is associated with important medical, social, and financial implications, and its incidence is expected to increase significantly as the Canadian population ages. Many of the consequences of osteoporosis are potentially lessened through the use of a number of non-pharmacological and pharmacological interventions. The oral bisphosphonate drugs etidronate, alendronate, and risedronate have been introduced as pharmacological options for the primary and secondary prevention of osteoporotic fractures.

We have conducted a systematic review assessing the clinical-effectiveness of etidronate, alendronate, and risedronate in the primary and secondary prevention of osteoporotic fractures in postmenopausal women receiving these agents compared with untreated women over a follow-up period of at least one year. A systematic literature search of the evidence from randomized placebo-controlled trials of each of the three drugs was conducted using a standardized Cochrane Collaboration approach to literature search, article selection, data extraction, and quality assessment. Clinical data analysis was conducted according to the methodology of the Cochrane Collaboration for systematic reviews and meta-analyses.

Considering the data available for the longest treatment duration in the trials and using the follow-up denominators for the number of patients in the trial, a detailed worked example will be considered. For this detailed worked example, the weighted RR effect estimates of fracture after treatment with the bisphosphonates alendronate and etidronate compared with placebo will be used to derive an indirect estimate. The indirect treatment comparison method will be used to evaluate the head-to-head comparison of alendronate to etidronate, using the placebo as the bridging group in the one-step comparison (i.e., k = 3).

3.3 Step 1: Effect Measure & Number of Treatments

In the main window, the effect measure of interest is identified, and information for each consecutive pair of treatments of interest is requested in terms of the point estimate and 95% CI of the effect measure for each direct comparison involved in the indirect comparison.

The effect measure of interest is the Relative Risk (RR), which is chosen by selecting the corresponding option button.

There are three treatments involved in this indirect comparison — alendronate, etidronate, and placebo — and the number 3 is entered in the Number of Treatments field.

For each consecutive pair of treatments, the direct estimates of the measure of association and the 95% LCL and UCL must be provided. The order of entry of the treatment pairs must follow the exact sequence indicated with the bridging comparison groups linking the treatment pairs. The interest here is to compare alendronate with placebo and then placebo with etidronate and, in so doing, use placebo as the bridging comparison group.

3.4 Step 2: Comparing Alendronate to Placebo

A systematic review was conducted for trials that compared alendronate with placebo for primary or secondary prevention. Non-vertebral fractures were reported in eight trials. One trial did not report fractures separately by treatment groups, and one trial reported that no fractures occurred in either treatment group.

The pooled estimate of the RR of non-vertebral fractures from the five trials that could be analyzed demonstrated a significant reduction (16%) in non-vertebral fractures (RR: 0.84 [95% CI 0.74 to 0.94]).

The direct estimate from this meta-analysis for the RR (0.84) and the 95% lower confidence limit (0.74) and 95% upper confidence limit (0.94) are entered on the first direct comparison line (1 = alendronate, 2 = placebo).

To request the weights used for calculating the weighted effect measure for the direct comparison, the arrow to the right of the (1,2) line is clicked. If the test of association is not of interest, this step can be skipped.

3.5 Step 2a: Weighted Effect Measure (1,2)

The pooled estimate of the RR of non-vertebral fractures (RR 0.84, 95% CI 0.74 to 0.94) was based on five trials. To calculate the weights that were used for this weighted RR risk, the Derived option and the Fixed effect option are selected because heterogeneity was not an issue.

Rates for the treatment and control groups for each study are requested in the form of numerator (number of events) / denominator (number of subjects). From the systematic review:

Treatment
(n/N)

Control
(n/N)

122/1022

148/1005

261/2214

294/2218

3/46

1/45

45/500

38/332

19/792

37/841

 

Rates for the treatment and control groups for each study are requested in the form of numerator (number of events) / denominator (number of subjects). From the systematic review:

3.6 Step 3:  Comparing Etidronate with Placebo

A systematic review was conducted for trials that compared etidronate with placebo for primary or secondary prevention. Non-vertebral fractures were reported in seven trials.

The pooled estimate of the RR of non-vertebral fractures from the seven trials indicated a lack of effect of etidronate on non-vertebral fractures. The 95% CI around the RR estimate for all non-vertebral fractures was wide with a relative risk reduction of approximately 32% and a relative risk reduction increase of 42% (RR 0.95, 95% CI 0.66 to 1.36). Results were consistent across the seven trials.

The direct estimate from this meta-analysis for the RR (0.95) and the 95% lower confidence limit (0.66) and 95% upper confidence limit (1.36) are entered on the second direct comparison. The results entered compare etidronate with placebo. By selecting Reverse, the results will be reversed so placebo is compared with etidronate and the (2,3) will correspond to (2 = placebo, 3 = etidronate).

To request the weights used for calculating the weighted effect measure for the direct comparison, the arrow to the right of the (2,3) line is clicked. If the test of association is not of interest, this step can be skipped.

3.7 Step 3a: Weighted Effect Measure (2,3)

The pooled estimate of the RR of non-vertebral fractures (RR 0.95, 95% CI 0.66 to 1.36) was based on seven trials. To calculate the weights that were used for this weighted RR, the Derived option and the Fixed effect option are selected because heterogeneity was not an issue.

Rates for the treatment and control groups for each study are requested in the form of numerator (number of events) / denominator (number of subjects). From the systematic review:

Treatment
(n/N)
Control
(n/N)
3/39 5/35
2/25 3/24
3/45 6/46
5/20 6/20
20/92 16/89
14/93 12/89
1/14 1/14

These results are entered in the corresponding lines provided. Click Close to save the entries and close the window.

3.8 Step 4: Calculating Results

Once all the data are entered, the resulting indirect comparison estimates for the effect measure and the 95% CI as well as the P value for the test of association corresponding to this effect measure are provided in the main window for the comparison of treatments (1,3) using treatment 2 as the bridging comparison.

When all data is entered, click Calculate.

The indirect treatment effect estimate for the RR of alendronate compared with etidronate was 0.88 with the 95% CI (0.60 to 1.29). The result indicates that alendronate and etidronate are not significantly different. This is confirmed with the P value for the test of association of 0.79.

To save these results to a .txt file, click Save.