Diagnostic Accuracy of Transbronchial Biopsy in Acute Farmer’s Lung Disease: Pathologic Criteria

17 Mar

Contingency tables for the pathologic criteria are given in Table 2. For simplicity, the results obtained from only one pathologist (R.S.F.) are reported since both pathologists graded interstitial infiltrations on slightly different scales. As a pathologic criterion, diffuse lymphocytic infiltration proved useful to generate positive changes in pretest probabilities: when present (graded > 1/3), this criterion had a positive likelihood ratio of 9.09 (95% Cl, 2.24 to 36.95). Moderate-to-severe infiltration (graded 2/3) was associated with a positive likelihood ratio of 12.73 (95% Cl, 1.73 to 93.32). Even though diffuse lymphocytic infiltration could significantly increase the likelihood of FL, it was found in only 20 of 55 cases of FL (36.4%). There is a trend for the presence of loosely formed granulomas (graded 1/2) to favor the diagnosis of FL (likelihood ratio, 1.82; 95% Cl, 0.48 to 6.89), and for well-formed granulomas to favor an alternative diagnosis that was most often sarcoidosis (likelihood ratio, 0.50; 95% Cl, 0.20 to 1.24). Such a finding was expected because of the current opinion that loosely formed granulomas (graded 1/2) may favor the diagnosis of FL, whereas well-formed granulomas (graded 2/2) would favor an alternative diagnosis. The combination of diffuse lymphocytic infiltration and granulomas (both of any grade) was an uncommon occurrence in FL cases (6/55; 10.9%) and was associated with a likelihood ratio of an infinite value; to avoid a division by zero, the addition of 0.5 to all cells to calculate the associated 95% Cl resulted in a likelihood ratio of 11.84 (95% Cl, 1.22 to 119.51). canadian health&care mall

Overall Assessment
The decision matrices obtained after the overall assessment from both pathologists are provided in Table 3. The likelihood ratios for each diagnostic category and associated 95% CIs are provided as well. Fifty-one of 105 TBB specimens (48.6%) were read by the first observer as nonspecific, while 23 of 105 (21.9%) of the TBB specimens were read by the second observer as nonspecific. Neither the diagnosis of FL nor an alternative diagnosis was then favored (likelihood ratio, 0.87; 95% Cl, 0.59 to 1.29 and 0.59; 95% Cl, 0.31 to 1.13, respectively). For each diagnostic category, there was no statistically significant difference between the likelihood ratios obtained by the pathologists.
Table 2—Contingency Table for the Pathologic Criteria

Pathologic CriteriaDiffuse Lymphocytic Infiltration, ScoreGranuloma,Score
01230 1 2
FL (n=55)35611343 6 6
Control samples4811036 3 11

Table 3—Results of the Overall Assessment of the TBBs: Contingency Table and Likelihood Ratios of Each Diagnostic Category

DiagnosticCategoriesObserver 1Observer 2
FLLikelihood RatioFLLikelihood Ratio
Probable FL651.09 (0.35-3.35)29102.64 (1.44-4.84)
Possible FL1982.16(1.04-4.49)1581.70 (0.79-3.67)
Nonspecific25260.87 (0.59-1.30)11170.59 (0.31-1.13)
Alternative diagnosis5110.41 (0.15-1.11)015O.OO