Title: GORE® Embolic Filter in Carotid Stenting for High Risk Surgical Subjects (EMBOLDEN)
Reference: Not published (yet) Registration number: NCT00766493
Source: ClinicalTrials.gov
Protocol: Purpose: Compare the 30-day safety and efficacy of the GORE® Embolic Filter used in conjunction with FDA-approved carotid stents to a performance goal obtained from carotid stent studies utilising distal embolic protection.
Enrollment: 250 Start Date: January 2009
Study Completion Date: July 2010
C. Study approaches to evaluating diagnostic, screening,
monitoring and prognostic test devices
As described in Chapter 4.3 main text, non-therapeutic devices include diagnostic, screening and prognostic test devices. Test devices commonly have no direct thera- peutic effects on the health outcomes of the intended population; instead, they are tests that generate information which in turn directs health care professionals (or sometimes patients themselves – depending on the type of device) to guide clinical management (see Figure 1). The health of the targeted individuals will thus improve indirectly because they receive more appropriate or earlier treatment, and because they can avoid potentially harmful additional tests.
Researchers of such devices should consider carefully whether they will focus on the results generated by the device often referred to as (predictive) accuracy studies – or also measure the down-stream consequences of test use, for example changes in the management of health care professionals or even in the health effects induced by combining the use of the device and follow-up treatment (device test-treatment evaluations).
Medical test Test result
Clinical action
Treat No treat
Patient outcome Patient
No direct effect Side effect possible
Patient outcome Figure 1. Possible working pathways of diagnostic, screening, monitoring, prognostic test
devices.
Example of non-therapeutic devices: relation between tests and patient outcomes
C1. Diagnostic and screening device accuracy designs
Diagnostic and screening test accuracy studies aim to measure the relationship between the results of the test and the presence or absence of a certain disease or con- dition (e.g. disease or any other health state) of interest at the same time. This leads to a cross-sectional (though time) design where subjects will undergo both the index device (test) under evaluation and the reference standard. The reference standard is the best available method – usually more invasive, burdensome or costly than the index device – for determining whether a subject in fact (‘truly’) has or does not have the condition of interest. Because the interest is in the cross-sectional relationship, the device and the reference standard should ideally be performed within a short time interval. Sometimes it is not possible or ethical to establish the presence or absence of a target condition at the moment when the index device is used. Follow-up can then be used as a means of confirming or refuting the true presence or absence of the target condition. This is frequently applied in studies on the accuracy of screening devices for detecting early stages of a certain disease, e.g. breast cancer. Several accuracy meas- ures can be used to express the agreement between the index device/test under evalu- ation and the true presence or absence of disease, for example sensitivity & specificity, positive and negative predictive values, likelihood ratios, and the diagnostic odds ratio. Test accuracy studies are vulnerable to some specific types of biases, as described at length in the literature (see e.g. references in Chapter 4).22 For example, in diagnos- tic accuracy studies, the individuals of interest are often defined by their presenting symptoms, signs and setting. Using patients already diagnosed with the condition or disease of interest and a separate group of control subjects without the disease can lead to overoptimistic measures of accuracy, because patients with the disease are more typical or advanced cases of the target condition. Also, the selected con- trol group can be atypical, for instance when a healthy control group has been used. Such case-control designs (see above) using healthy controls may indeed be used to examine whether there a device test holds any promise. However, the results of such designs and the accuracy (agreement) found between the index device and the ref- erence test cannot simply be generalised to the device accuracy that would be found when the device is used in the targeted clinical population in routine practice. For diagnostic and screening test accuracy studies, it is crucial to think about the targeted individuals and setting beforehand (see Chapter 4.5). Another key quality item of such studies is that all patients will undergo the same (best available) reference standard. Finally, the results of the index diagnostic or screening test device should be deter- mined or interpreted without knowing the outcome of the reference standard; other- wise, there is the risk of producing overoptimistic estimates of accuracy for the device under evaluation.
22 Bossuyt P.M., Irwig L., Craig J., Glasziou P., Comparative accuracy: assessing new tests against existing diagnostic pathways. BMJ. 2006 May 6;332(7549):1089-92.
C1.1 Single (non-comparative) device test accuracy studies
The aim in single device test accuracy studies is to determine the accuracy of a specific device test (see Figure 2). This leads to a cross-sectional diagnostic or screening device study where the targeted individuals (ideally a consecutive series of such indi- viduals) undergo both the index device test and the reference standard. A key qual- ity item of such studies is that all patients will undergo the same optimal reference standard. The results of the index test should be determined or interpreted without knowing the outcome of the reference standard; otherwise, there is the risk of produc- ing overoptimistic estimates of accuracy.
Single device test accuracy studies are not suitable for determining whether a specific test leads to improved accuracy beyond what is achieved with available test results, or which order of testing is to be preferred (see C1.2); neither are they suita- ble for determining whether the use of the diagnostic or screening device will lead to actual (additional) health benefits in individuals (see C.3).