Observer variability in the assessment of clinical probability in patients with suspected pulmonary embolism

Abstract

5 Nolan TE, Smith RP, Devoe LD. Maternal plasma D-Dimer levels. in normal and complicated pregnancies. Obstet. Gynaecol. 1993; 81: 235. 6 Observer variability in the assessment of clinical probability in patients with suspected pulmonary embolism Assessment of the clinical pretest probability in patients with clinically suspected pulmonary embolism (PE), based on clinical judgement alone (further referred to as Ôclinical probability estimateÕ) or a structured prediction model, in order to guide further diagnostic testing has aroused new interest. Wells and colleagues derived a clinical model based on information from the medical history, physical examination, blood gas analyses, chest X-ray, electrocardiography and the likelihood of an alternative diagnosis [1]. This model was subsequently simplified [1,2]. These models have now successfully been applied in several management studies in patients with suspected PE [3][4][5]. However, the diagnostic accuracy of the clinical probability assessment in previous studies varied significantly [1][2][3][4]6]. The reasons for this observed heterogeneity in test accuracy is not fully explained. An important determinant of test accuracy is its reproducibility, which so far has not been studied. We therefore evaluated the interobserver variability of the original clinical probability assessment using a previously described structured clinical model. In addition, we compared the results of the clinical model with those of the clinical probability estimate in the same patient as assessed by the same physician.The investigation was performed within the framework of a management study in a large teaching hospital, in which the outcome of the clinical probability was used in combination with a D-dimer test to guide further diagnostic and therapeutic management [4]. In consecutive in-and outpatients with clinically suspected PE, clinical probability was assessed by two independent physicians using estimate and a structured clinical model as described by Wells and colleagues (further referred to as Ôextended clinical modelÕ) with the classification of low, moderate or high probability of PE [1]. After completion of the study, the obtained information was used to calculate the result (low, moderate or high and unlikely or likely probability) of the simplified clinical model, as described by Wells and colleagues [2]. The primary analysis was performed in the patient group with a duplicate assessment of the clinical probability of PE using the extended and simplified clinical model. The agreement between distinct observers was expressed by a weighted kappa (j). A weighted j value of 1 corresponds to perfect agreement, 0 to agreement as expected by chance. We also calculated a weighted j for the agreement between the results of the extended clinical model and the clinical probability estimate.In a total of 45 consecutive patients independent, duplicate assessments of the clinical pretest probability of PE using the clinical model were available for analyses. Of the 45 patients, 13 h...

Alternative diagnosis other than pulmonary embolism as a subjective variable in the Wells clinical decision rule: not so bad after all

“…The latter criterion carries a major weight in the score, is obviously dependent on the interpretation of the physician and cannot be standardized. The moderate reproducibility because of interobserver variability [4] and lack of standardization are the major reported points of criticism of the Wells rule [3]. The Geneva score requires a blood gas analysis while breathing room air, and was only evaluated in patients in an emergency ward.…”

Update in the diagnosis of deep-vein thrombosis and pulmonary embolism

“…The Wells simplified score (Table 2) and the Geneva score are the most widely used and validated predictions rules used in diagnosis of pulmonary embolism [51 • ]. One criticism of the Wells score has been the need to assess whether an alternative diagnosis is more probable than pulmonary embolism -a factor contributing to only moderate interobserver agreement when using the score [52,53]. Nevertheless, the Wells score was recently validated again as being able to stratify accurately PTP of pulmonary embolism in a outpatient population [53].…”

Diagnosis of venous thromboembolism: an update

“…26 The Tinaquant ® test, an immunoturbimetric assay, has also been validated in three outcome studies that included 2071 patients. 16,27,28 The result was negative, ruling out PE in 857 (41%) patients who had either a low clinical probability of PE or were classified as 'PE unlikely' according to the dichotomized Wells' rule. 16 The 3-month thromboembolic risk was 0.6% (95% CI, 0.2 to 1.4).…”