Original article
Efficient diagnostic test sequence: Applications of the probability-modifying plot

https://doi.org/10.1016/S0895-4356(01)00391-2Get rights and content

Abstract

Excessive diagnostic testing rarely leads to complete certainty of the disease status of a patient. Because of constraints of health care budgets, physicians should be aware of the diminishing marginal gain of each subsequent test in a testing strategy related to costs. We developed the probability-modifying plot to determine economically efficient sequence of testing without losing clinical effectiveness. Two applications of the model are presented: diagnosis of Helicobacter pylori and diagnosis of benign prostatic hyperplasia. The probability-modifying plots are drawn to visualize and calculate the economic efficiency of all possible test sequences, meaning that although the different test sequences are equally accurate, some have even lower costs than others.

Introduction

Excessive diagnostic testing rarely leads to complete certainty of the disease status of a patient 1, 2. When using diagnostic tests sequentially, the marginal gain of each test can be determined and knowledge from previously performed tests can be used to decide which, if any, test should be performed next. Because of constraints of health care budgets, physicians should be aware of the diminishing marginal gain of each subsequent test in a testing strategy related to costs. However, if many factors such as test characteristics and costs of numerous diagnostic tests are to be included in diagnostic strategy considerations, a physician's ability to intuitively integrate the results seems to reach a limit [3].

Economic evaluations aim to provide the necessary information about costs and outcome of health care interventions to policymakers [4]. An important issue when performing an economic evaluation is the choice of the competing alternative to which the intervention under study is compared, because this can influence study findings [5]. In the case of evaluating diagnostic tests, there are various alternatives for comparison in a study, including test versus no test, one test versus another test, and test versus treatment [6]. However, as mentioned above, diagnostic tests are rarely used in isolation and therefore the evaluation of a specific test should be performed not only in contrast to other tests but also, when relevant, in the context of the other tests. Thus, decisions must be made both about which diagnostic tests to perform and about the sequence of testing. The principle of including tests with respect to their gain in certainty of disease status is the basis for most methods to evaluate the sequence of diagnostic tests. These methods agree with individual clinical decisionmaking because the aims of optimizing patient health (effectiveness) is considered to be the primary objective for the clinician dealing with the individual patient. The aim of reducing costs is fundamentally different and even sometimes conflicting; however, from a global health care perspective a methodology to evaluate the economic efficiency of alternative diagnostic testing strategies is needed. Efficiency of diagnostic testing strategies is defined by the level of the costs involved in the different strategies, given that all stategies lead the same diagnostic accuracy.

Section snippets

Methods

Based on the principles of decision analysis and decision tables, we developed the probability modifying plot, a method that includes both economic efficiency and aspects of clinical effectiveness for determining the sequence of diagnostic tests. Thus, the method puts an emphasis on economically efficient sequence of testing without losing diagnostic accuracy regarding a treatment decision. The possibility that less accuracy can be outweighed by cheaper testing is not considered in this method

Description of disease and diagnosis

Helicobacter pylori (HP) is a common chronic infection in humans. The infection leads to substantial morbidity. It may cause peptic ulcer, functional dyspepsia, or gastric cancer. However, widespread noninvasive testing, such as a breath test and serology, followed by treatment of nonsymptomatic infected patients is beyond issues of feasibility and financial consideration. Besides this, information on possible unwanted side effects of this approach is lacking [10]. Therefore, whether a person

Description of disease

BPH is a histological condition that may cause a benign enlargement of the prostate and is common in elderly men. Four properties are related to the disease process of BPH: (1) histological BPH, (2) macroscopic benign enlargement of the prostate gland (BPE), (3) bladder outlet obstruction (BOO), and (4) lower urinary tract symptoms (LUTS), such as diminution in the caliber and force of the urinary stream, hesitancy initiating voiding, inability to terminate micturition abruptly, a sensation of

Discussion

Assuming that time delay is not important, sequential use of diagnostic tests is potentially more economically efficient than performing multiple tests simultaneously [19]. In our probability-modifying model, economic efficiency of the sequence of diagnostic tests was either defined as the least number of tests or the lowest total cost for testing. In fact, the model shows that the different sequences have identical clinical effectiveness, defined as the number of patients being misclassified

Acknowledgements

This work was supported by a grant from the Reinier Post Foundation, The Netherlands.

References (35)

  • M Berger

    Design of prospective cost-effectiveness clinical trialsthe critical role of the comparator group

    Drug Inform J

    (1995)
  • J.L Severens et al.

    Economic evaluation of diagnostic testsa review of published studies

    Int J Technol Assess Health Care

    (1999)
  • Severens JL, Vries Robbé PF de, Verbeek ALM. Optimisation of diagnostic test sequence: the probability modifying plot....
  • H.C Sox et al.

    Selecting and interpretation of diagnostic tests

  • S.G Pauker et al.

    The threshold approach to clinical decision making

    N Engl J Med

    (1980)
  • L Rabeneck et al.

    Helicobacter pyloriwhen to test, when to treat

    Ann Intern Med

    (1997)
  • Diagnosis of Helicobacter pylori infection. Review of diagnostic techniques and recommendations for their use in different clinical settings

    Scand J Gastroenterol

    (1997)
  • Cited by (10)

    • Test sequence of CSF and MRI biomarkers for prediction of AD in subjects with MCI

      2012, Neurobiology of Aging
      Citation Excerpt :

      Test sequence analyses were performed to investigate the added predictive value of a second biomarker for AD-type dementia or cognitive decline in specific subgroups. We identified the best test sequence of CSF Aβ1–42/tau ratio and HCV assessment and visualized it using probability-modifying plots (Severens et al., 2001). The relation between biomarkers, the sample size, and number of subjects required for screening, was tested for a fictive placebo-controlled AD drug trial targeting amyloid pathology in subjects with MCI.

    • New patient-oriented diagnostic test characteristics analogous to the likelihood ratios conveyed information on trustworthiness

      2005, Journal of Clinical Epidemiology
      Citation Excerpt :

      Analogously, similar patterns of changes may occur in PPR and NPR. The new measures could be used to compare tests that are performed sequentially or simultaneously, and to explore the efficiency of the use of diagnostic tests in alternative sequences, based on previous publications [30]. Graphical presentation of PPR and NPR could prove to be an efficient teaching tool, based on an approach that has been developed by Johnson [31].

    View all citing articles on Scopus
    View full text