EBM David Kirk 1/18/01
General Question:
Should a hospitalized, pregnant lady with lupus be placed on prophylactic anticoagulation?
Paper:
Prophylactic Antithrombic Therapy for Patients With Systemic Lupus Erythematosus With and Without Antiphospholipid Antibodies. Arch Intern Med. 2000; 160:2042-2048
Design of Study:
Decision Analysis
Brief Discussion:
Decision analysis uses quantitative methods to examine the various consequences to different strategies. The foundation of the study is a decision tree. (FIGURE 1 on 2044)
Probabilities are obtained for each chance circle in the decision tree. To obtain the information, the authors must select and use the best studies available. The techniques used to select and appraise the literature should be explained. The data are obtained in similar fashion as for a meta-analysis. (TABLE 1 on 2045)
Utilities represent quantitatively how the various outcomes will affect a patient. Credible ratings would best come directly from a large group of patients with the various conditions. However, it is also often obtained through the literature from various published quality-of-life ratings. (TABLE 2 on 2045)
The probabilities and utilities are used to calculate the outcomes. Based on these calculations, the authors concluded that prophylactic aspirin was the preferred strategy as “the number of arterial and venous thromboembolic events prevented by this option exceeded that of major bleeding episodes included in all risk scenarios.”( FIGURE 2 on 2046). Table 4 on 2047 shows the quality-adjusted survival gains in months provided by aspirin in the three risk groups.
As a decision analysis tries to account for both the quality and quantity of life, the results are often reported as “quality-adjusted life years.” The gain in life expectancy can be spread out over a large course of time instead of just prolonging the ending of life. Through the literature it is generally accepted that a gain in life expectancy of 2 or more months should be considered a significant gain; anything less than that is probably not clinically relevant.
Published literature can not be perfectly precise. Uncertainty introduced during the data collection can become magnified in the outcomes. To reduce this problem, decision analysts use sensitivity analysis to explore how the various degrees of uncertainty could affect the outcomes of the study. Both probability and utility values should be tested with sensitivity analysis. Figure 3 and 4 are included to show that the wide ranges in values such as aspirin efficacy are less important than one might expect.
Evaluation:
§ Are the results valid?
o Were all important strategies and outcomes included?
§ All the expect outcomes were included in the study; however, several potential treatments such as LMWH and heparin were not explored.
o Was an explicit and sensible process used to identify, select, and combine the evidence into probabilities?
§ The assumptions required are described in the “Material and Methods” section of the paper on 2043. Other than being stated that data obtained were “derived from published systematic reviews and meta-analyses,” the authors failed to describe their exact process of evaluating the literature that was chosen.
o Were the utilities obtained in an explicit and sensible way from credible sources?
§ The utility values were obtained through the literature; however, the process by which was done is not well documented.
o Was the potential impact of any uncertainty in the evidence determined?
§ Sensitivity analysis was used to show that the imprecision in the probabilities and rates would not greatly affect the end results of the study. These same techniques were not used on the utility values.
§ What are the results?
o Does one strategy result in a clinically important gain for patients?
§ Aspirin appears to add greater than 1 quality-adjusted life year for certain subgroups of patients. According to these calculations, aspirin provides signifigant gains in all patient groups except in low risk patients greater than the age of 50.
o How strong is the evidence used in the analysis?
§ Garbage in = garbage out. Without knowing the details of the methods used to obtain the data, the reader is forced to trust the authors abilities to filter through and extract from the literature.
o Could the uncertainty in the evidence change the result?
§ Figures 3 and 4 show that uncertainty in the probabilities and rates change the conclusions very little. It is unknown if the uncertainty in utility values would be of any consequence.
§ Will the results help me in caring for my patients?
o Does the paper address my patient’s clinical features?
§ Errr, not much. This paper fails to address several important characteristics of our patient in question. Heparin would be the logical choice for our patient; however, this is not an included strategy. This paper also fails to address pregnancy in any way.
§ Do the utilities reflect how my patients would value the outcomes of the decision? Is the quality of a day of somebody with lupus really half that of a normal person? Who knows.
Comments:
While this paper is riddled with multiple flaws, I find it interesting enough that I’ll discuss the risks and benefits of aspirin therapy with my lupus patients in the future.
As it happens so frequently, the literature fails to provide us answers to an important clinical question. Most anticoagulation studies regarding pregnant women with lupus and antiphospholipid antibodies are attempting to prevent miscarriages not DVTs. Luckily, as physicians we are trained to be problem solvers and thinkers. We can function with a head full of common sense and without a fist full of double blinded randomized controlled trials.