Decision tables and logic in decision analysis

Screening strategies for tubal factor subfertility

BACKGROUND

Different screening strategies exist to estimate the risk of tubal factor subfertility, preceding laparoscopy. Three screening strategies, comprising Chlamydia trachomatis IgG antibody testing (CAT), high-sensitivity C-reactive protein (hs-CRP) testing and hysterosalpingography (HSG), were explored using laparoscopy as reference standard and the occurrence of a spontaneous pregnancy as a surrogate marker for the absence of tubal pathology.

METHODS

In this observational study, 642 subfertile women, who underwent tubal testing, participated. Data on serological testing, HSG, laparoscopy and interval conception were collected. Multiple imputations were used to compensate for missing data.

RESULTS

Strategy A (HSG) has limited value in estimating the risk of tubal pathology. Strategy B (CAT-->HSG) shows that CAT significantly discerns patients with a high versus low risk of tubal pathology, whereas HSG following CAT has no additional value. Strategy C (CAT-->hs-CRP-->HSG) demonstrates that hs-CRP may be valuable in CAT-positive patients only and HSG has no additional value.

CONCLUSIONS

CAT is proposed as first screening test for tubal factor subfertility. In CAT-negative women, HSG may be performed because of its high specificity and fertility-enhancing effect. In CAT-positive women, hs-CRP seems promising, whereas HSG has no additional value. The position and timing of laparoscopy deserves critical reappraisal.

Assessing the gain in diagnostic performance when combining two diagnostic tests

Combining dichotomous (or dichotomized) results of two diagnostic tests will result in a trade-off in sensitivity and specificity of the combined test relative to the component tests. Because of this inherent trade-off, likelihood ratios provide a clinically relevant means of comparing the combined test with one of its components. The likelihood ratios depend on both sensitivity and specificity and hence take into account the trade-off between them. A graphical approach is used to assess whether the combined test is superior to a component test, or vice versa. Asymptotic standard errors are derived for comparing likelihood ratios when a paired study design is used. The trade-off in the expected number of additional true positive and false positive results (or true negative and false negative results) is used as the basis for deciding whether to use tests in combination when neither the combined nor a component test shows superior test performance based on their likelihood ratios. These methods are illustrated with an example that considers the combined use of Pap and HPV testing.

Representation of clinical practice guidelines in conventional and augmented decision tables

OBJECTIVE

To develop a knowledge representation model for clinical practice guidelines that is linguistically adequate, comprehensible, reusable, and maintainable.

DESIGN

Decision tables provide the basic framework for the proposed knowledge representation model. Guideline logic is represented as rules in conventional decision tables. These tables are augmented by layers where collateral information is recorded in slots beneath the logic.

RESULTS

Decision tables organize rules into cohesive rule sets wherein complex logic is clarified. Decision table rule sets may be verified to assure completeness and consistency. Optimization and display of rule sets as sequential decision trees may enhance the comprehensibility of the logic. The modularity of the rule formats may facilitate maintenance. The augmentation layers provide links to descriptive language, information sources, decision variable characteristics, costs and expected values of policies, and evidence sources and quality.

CONCLUSION

Augmented decision tables can serve as a unifying knowledge representation for developers and implementers of clinical practice guidelines.

Improving clinical guidelines with logic and decision-table techniques: application to hepatitis immunization recommendations

Clinical practice guidelines can be clarified, verified, and simplified by the use of logical analysis and the application of decision-table techniques. This methodology is applied to a CDC guideline for the prevention of perinatal transmission of hepatitis B by immunization. Relevant clinical variables are identified and possible values for each variable are defined. An exhaustive enumeration of value combinations is generated. Logically impossible combinations are identified and eliminated. The guideline's recommendations are next translated into a set of rules and compared with the remaining value combinations. Variable combinations that are not covered by guideline recommendations represent incomplete guideline specifications. Inconsistency can be identified by finding identical condition sets in two or more rules. The procedure demonstrates that the hepatitis guideline is incomplete. Logical analysis can improve the quality of clinical practice guidelines by assuring comprehensiveness and consistency.

Decision analysis--a helpful tool for clinicians to establish diagnostic-therapeutic guidelines?

In this paper we focus on the question: Does decision analysis provide a framework to assess the value of diagnostic tests in clinical practice and how can it be used by clinicians in establishing diagnostic-therapeutic guidelines. To study this question we performed two analyses concerning the use of pelvic lymphadenectomy and pedal lymphography for staging prostate cancer. Both analyses yielded similar results as far as the preferred strategy was concerned, yet the approach and set up of the two analyses were different. The first analysis was performed in accordance with the textbooks on decision analysis. However, using this traditional approach we encountered some difficulties: in structuring the decision tree, in eliciting values for the quality of life parameters, and in interpreting the results. These difficulties urged us to modify the approach, presented in the second analysis. In this second analysis, the decision problem was split into several consecutive decision problems which corresponded to the questions posed by the clinicians. Longevity and quality of life were considered separately and the consequences of treatment and testing, which affect the quality of life of the patients, were indicated by just two parameters. Finally, the result of the analysis was expressed in clinically meaningful terms. The second analysis is compared with different approaches presented in the literature for analyzing decision problems involving diagnostic tests. Despite some unresolved methodological problems it is concluded that decision analysis provides a good framework for clinicians to structure and analyze complex decision problems.

The area under the ROC curve and its competitors

The area under the receiver operating characteristic (ROC) curve is a popular measure of the power of a (two-disease) diagnostic test, but it is shown here to be an inconsistent criterion: tests of indistinguishable clinical impacts may have different areas. A class of diagnosticity measures (DMs) of proven optimality is proposed instead. Once a regret(-like) measure of diagnostic uncertainty is agreed upon, the associated DM is uniquely defined and, indeed, calculable from the ROC curve configuration. Two scaled variants of the ROC are introduced and used to advantage in the analysis. They may also be helpful to students of medical decision making.

Threshold analysis using diagnostic tests with multiple results

Clinical problems represented by decision trees can be analyzed in terms of the probability threshold model, which provides management recommendations based on the prior probability of disease, the test threshold, and the test-treatment threshold. As originally proposed, the threshold model assumes that diagnostic tests provide information about a single event that is relevant to the decision. For some problems, however, a diagnostic test may provide information about more than one such event (e.g., a computed tomography [CT] scan gives information about both mediastinal and hilar metastases in lung cancer). The authors extend the probability threshold model to cases in which a single test provides information about two events that are relevant to the decision. They derive four thresholds that determine the best strategy for any combination of test results. The approach is illustrated for the decision to use a CT scan to stage lung cancer. The analysis reveals that: 1) the range of prior probabilities for which testing is optimal increases; 2) for some prior probabilities only test results about one event are important; 3) for some prior probabilities test results about both events are important; and 4) failure to account fully for information provided by a test can lead to erroneous test and treatment recommendations.

Reporting clinical trials from the viewpoint of a patient's choice of treatment

Those who report a clinical trial should acknowledge the right of the 'consumer' to make decisions based on his own valuation of the beneficial and adverse effects which rival treatments may have. Suppose a new patient is inclined to trade one unit of benefit for c units of complication. Then he should (should not) be given the treatment if his estimated utility gain, x1-cx2, is positive (negative) and statistically significant according to the data of the trial; here x1 (x2) denotes the observed average benefit (complication level). If the estimated gain is not statistically significant, the data do not allow any firm recommendation. This c-dependent recommendation in general cannot be determined from inspection of a joint confidence region for the two means concerned. Therefore investigators should present the outcome of the significance test as a function of c (inverted inference). Typically there are several types of adverse effect or benefit, in which case the quantity c must be generalized into a vector of personal relative utility weights.

Threshold analysis of decision tables

When a decision table is used to find a maximum expected utility testing strategy, it is based on a given prior probability distribution of diseases. In the two-disease situation, a threshold analysis over all prior probabilities can be done using threshold transformations of the points of indifference between treatments. This results in a set of prior probability intervals each with its own unique decision rule. The Boolean expression for the table indicates the acceptable testing strategies. A decision table analysis may then be extended to include invasive or costly investigations. The technique represents a saving in time and effort compared with standard decision tree approaches, especially where investigative recommendations are to be made for a broad range of prior probabilities, e.g., where initial symptoms and signs are considered before the investigations.