Diagnostic Accuracy of Transbronchial Biopsy in Acute Farmer’s Lung Disease: Converting Pretest to Posttest Probabilities

19 Mar
2014

Diagnostic Accuracy of Transbronchial Biopsy in Acute Farmer’s Lung Disease: Converting Pretest to Posttest ProbabilitiesConverting Pretest to Posttest Probabilities
The figures presented in Table 4 were derived from the nomogram of Fagan (Fig 2). Low (25%), intermediate (50%), and high (75%) pretest probabilities were arbitrarily selected to illustrate their modification by any given test results. The presence of diffuse lymphocytic infiltration (which was associated with a likelihood ratio of 9.09) or the combination of diffuse lymphocytic infiltration and granulomas (which was associated with a likelihood ratio of 11.84) significantly alters the pretest probability of FL. However, their uncommon occurrence in cases of FL (36% and 11% of the cases, respectively) limits their clinical usefulness in the diagnosis of FL. Lower likelihood ratios significantly alter the pretest probability of FL only when it is intermediate. canadian health & care mall

Discussion
Since first reported, TBB has raised much attention among medical specialists, thoracic surgeons, and pathologists for the diagnosis of interstitial lung diseases. Early case series reported on the diagnostic accuracy of TBB in terms of “usefulness.”” For most of these reports, pathologic diagnoses obtained after TBB were listed without any reference to a gold standard. In most reports, cases of “interstitial pneumonitis” or “pulmonary fibrosis, with or without inflammation” represented a proportion as high as 43%> and were classified as acceptable final diagnosis. Smith et al acknowledged that many clinicians were reluctant to accept a diagnosis of nonspecific fibroinflammatoiy disease established by TBB specimens and that further study was necessary.
Few studies have actually correlated the results of TBB with those of open lung biopsies. Despite stronger methods, these studies were not exempt from bias, with the consequence of conflicting conclusions. Levin et al reported on 33 patients, including 22 with diffuse lung disease. Fifteen of these 22 patients also underwent an open lung biopsy. TBB correlated with open lung biopsy in 11 patients (73%). Three patients showed pathologic changes on TBB specimens consistent with “rheumatoid lung;” none of these were further investigated with an open lung biopsy. Even if the pathologic diagnosis of rheumatoid lung was consistent with the unspecified clinical picture, these patients could be considered as “difficult cases,” and excluding them from the analysis may inflate the figures of test accuracy. Nevertheless, the investigators concluded that TBB “can be applied to diffuse (…) lung disease with good results.” On the contrary, Wall et al did not perform open lung biopsy when TBB led to a diagnosis of sarcoidosis or when a diagnosis could be established with reasonable confidence after TBB along with clinical data. This decision left only the difficult cases for verification by open lung biopsy. The authors concluded that “TBB diagnosis of interstitial pneumonia, chronic inflammation, nonspecific reaction, and fibrosis are unreliable and often entirely misleading.”
Table 4—Posttest Probability of FL Given Different Pretest Probabilities and Results of the Assessment of TBB

Diagnostic Categories or Pathologic Criteria*LikelihoodRatiosiLow(25%)Pretest ProbabilityIntermediate(50%)High(75%)% FL Cases With This Test Result
Probable FL2.6447738953
Possible FL1.7036638427
Nonspecific0.5916376420
Alternative diagnosis0.00000
Granuloma* (1/2)1.8238658511
Diffuse LI* (>1/3)9.0975909736
Diffuse LI (>2/3)12.7381939725
Combination of granuloma and diffuse LI11.8480929711
top