EBM Conference 9/14/01

Marion McCrary MD

Study: Perrier, Amaud et al. “Performance of Helical Computed Tomography in unselected outpatients with suspected Pulmonary Embolism.” Annals of Internal Medicine. July 17, 2001. Vol 135, No 2, p 88-97.

Background/Objective of Study: PE is a common and potentially fatal disorder. Prevalence of PE in suspected cases is approx. 25-35%. Most pts with a low clinical prob of PE may be managed entirely by noninvasive means such as plasma D-dimer, lower limb compression ultrasonography, V/Q lung scans, and helical CT. If combos of the first 3 means are used, pulmonary angiography (the gold std) is necessary in only 4-11% of cases, as recently demonstrated in 2 large-scale outcome trials. The possibility of using CT is another issue currently under study. Initial studies were very optimistic however recent meta-analyses have shown a wide variety of sensitivity (53-100%) and specificity (73-100%) and concluded that better, more complete trials were necessary. Certain methodological criteria were suggested. The objective of this study was to eval the sensitivity and specificity of CT of PE diagnosis using a validated strategy that fulfilled these criteria suggested in a recent meta-analysis.

Design: Observational study

Setting: ED of a teaching and community hospital in Geneva, Switzerland of 25 months

Patients: Initially a cohort of 1108 consecutive, nonselected pts presenting the ED with a

suspected PE. INCLUSION CRITERIA: Clinical suspicion of PE, older than 16, and plasma D-dimer >500. EXCLUSION CRITERIA: Plasma D-dimer <500 (386-35%),

then 267 pts (24% of 1108) with abnormal D-dimers were excluded b/c of they had CI of CT,{Cr> 1.69 (47), contrast allergy (14), asthma (12), or pregnancy (9),}, declined to

participate or were unable to consent (139), had already been tx with oral anticoag at study entry(13), had CI to anticoag (5), were likely to be impossible to follow (6), or were suspected to survive less than 3 months (22). Then, CT was unavailable or could not be used for study purposes in 108 pts (10% of 1108): 13 could not perform CT for technical reasons, 10 had had a CT performed elsewhere, 7 pts had the attending demand the CT results, 2 died, 21 had alternative dx established, and 55 left the hospital or were transferred before CT was performed. Then the diagnostic workup was incomplete (45) or interrupted (3) b/c of anticoagulation for another indication in 48 patients (4% of 1108). Overall, 423 with abnormal D-dimer levels (total 38% of 1108) were excluded from the study. 299 of the 1108 pts (27%) were included in the study and of those 12 were further excluded from the final determination of sens and spec secondary to inconclusive results on CT leaving approx 26% of the initial cohort determining the sens and spec. The authors also report that there was no clinically significant differences b/t the included and the excluded in terms of age, sex, RF, clinical presentation, and clinical prob of PE. Table 1 has characteristics of the 299 included only. Later in the paper they comment that there was a difference in median age b/t the pos and neg D-dimer pts (age 69 comp to 45).

Set up: They compared CT to a “recently validated” diagnostic algorithm for PE: Pulmonary

angiography showing PE, High Prob V/Q (according to the revised PIOPED criteria), or

DVT on ultrasound and a clinical suspicion for PE. PE was considered absent in the presence of a normal angio, or a low prob V/Q, or a combo of low clinical prob plus nondiagnostic V/Q plus neg u/s. In pts with the last combo, PE is very unlikely. This algorithm also included 3-month clinical f/u where venous thromboembolic events and episodes of bleeding were documented. F/u by Family MDs and phone interviews. F/U was completed for all pts included in the study.

Techniques for test described in methods. Tech who performed the ELISA D-dimer tests was unaware of clinical data. U/S happened by trained staff w/in 24 hrs of presentation. CT evaluated up to the segmental arterial level. The CTs were read by a subspecialty trained chest radiologist who “prospectively recorded the findings.” The results of the CT were withheld from the attending MDs. Exceptions were made for rare cases (nodule) that necessitated a reg chest CT and these people were excluded from the study. Then each CT was presented at least 3 months later to radiologists (also a CV and a general radiologist) who were blinded to all clinical data and the other test results.

Funding: Not an issue.

Results: PE in 118/299 (39%), if include nml D-dimer pts-118/685 (17%)

NO PE in 181/299 pts

TABLE 2- TP 81, FN 35, FP15, TN 156, Inconclusive scans 12 (motion or insuff contrast)

Sensitivity: Pos CT/ Pos PE = TP/(TP+FN) = 81/116 = 0.698 = 70% (95% CI= 62%-78%)

Specificity: Neg CT/ Neg PE = TN/(TN+FP) = 156/171 = 0.912 = 91% (95% CI= 86%-95%)

Pos LR= Sen/(1-spec) = TPR/FPR= 0.698/(1-0.912)= 7.9= 8

Neg LR= (1-sen)/Spec = FNR/TNR= (1-0.698)/.912 = 0.33 = 0.3

FN results (35) were diagnosed by hi prob V/Q in 19, U/S in 12, 3 with angio, & 1 on f/u

FP results (15) had PE s R/O by nml V/Q in 3, angio in 7, low clinical prob + nondx V/Q + nml +

U/S + nml f/u in 5 pts. FP were in segmental and subsegmental arteries (Table 4).

More stringent dx criteria would not change sens or spec in this sample (Table 3).

Interobserver agreement was excellent (Table 5)

Combined strategies= if do CT if u/s neg =overall accuracy 87% & FNR of 21% (30% CT alone)

if do CT if u/s and v/q neg=overall accuracy 94% &FNR 5%/FPR 7%

F/U=7/118 pt with PE died (6%) (3-PE, 4-CA); 2 Major Bleed (1-GI, 1-RP)

9/181 pt w/o PE died (5%) (5-CA,2-ResF,0-PE), 1 Major Bleed (Fem Art for angio)

Authors conclusions: Sens in this study was among the lowest found to date and the Spec was also in the

lower range. This study fulfilled most of the methodologic criteria for a valid assessment of the

characteristics of a diagnostic test. The spectrum of pts were broad. The interpreters were blinded to the other studies and the interpreters of the other studies had no knowledge of the CT findings. No major bias was suspected. These findings: the sensitivity was too low to r/o PE without addtl tests. Neg LR was close to that of a low-prob V/Q- also shown to be unsatisfactory for excluding PE. If CT shows a main PA PE, treat but if it is in more distal vessels the possibility of a FP is higher and should be treated if the clinical prob is intermediate or high. However if low clinical prob with suggestion of distal PE by CT, PE is less certain. Those with inconclusive scans should undergo further tests. The authors also state that CT is “potentially” useful for dx of PE when used in combo with other tests. (would at least need normal u/s, nondx V/Q for CT to attempt to replace angio in an algorithm= second combined strategy in results). Limitations: may not apply to hospitalized pts- and less likely to for CT to perform better secondary to all their comorbid conditions)

Assessment of Study:

Are the Results Valid?

1. Was there an independent, blind comparison with a reference standard? YES. The reference standard is acceptable according to a paper also first authored by the same person. The only interesting thing is that the reference standard involved options of several pathways.

2. Did the patient sample include an appropriate spectrum of patients to whom the diagnostic test will be applied in clinical practice? YES. Table 1

3. Did the results of the test being evaluated influence the decision to perform the reference standard? NO. The results of the test were not known to the MDs taking care of the pts. The only possible bias was if the radiologist reading the CT initially was also giving advice to the MD about which avenue to pursue.

4. Were the methods for performing the test described in sufficient detail to permit replication? YES. The type of scanners used were discussed, the amount of contrast, etc.

5. Overall, are the results of the study valid? YES.

What are the results?

What are the likelihood ratios for the test results? Pos LR is 8, Neg LR is 0.3. Neither is

incredibly big.

Will the results help me in my patient care?

1. Will the reproducibility of the test result and its interpretation be satisfactory in my setting? YES. I think so. These results were for outpatients being initially eval’d in the ED. I think they would also apply to clinic pts. However, I don’t think they could be extrapolated to inpatients. Setting is also similar to UNC.

2. Are the results applicable to my patients? YES. We have similar distribution of disease severity and competing diseases in this tertiary care hospital.

3. Will the results change my management? YES. The spiral CT is not as good alone as I thought. The LRs are OK but not great. However, these authors relied on D-dimer as a discriminating test (which we don’t) and it interests me as an area to think of as well.

4. Will patients be better off because of the test? Well, PE needs to be diagnosed and fairly risk free test and effective treatment. If get initially and has main PA PE then will treat, but if distal to that, may consider other testing prior to commit to long term therapy anticoag. If negative test initially and pos D-dimer, then may consider other testing if clinical suspicion still high. But is it good enough to stop with a neg CT and pos D-dimer if clinical suspicion low. More studies are needed.