The diagnosis of hypersensitivity pneumonitis (HP) is difficult and often relies on histopathology. Our objective was to identify diagnostic criteria and to develop a clinical prediction rule for this disease. Consecutive patients presenting a condition for which HP was considered in the differential diagnosis underwent a program of simple standardized diagnostic procedures. High-resolution computed tomography scan and bronchoalveolar lavage (BAL) defined the presence or absence of HP. Patients underwent surgical lung biopsy when the computed tomography scan, BAL, and other diagnostic procedures failed to yield a diagnosis. A cohort of 400 patients (116 with HP, 284 control subjects) provided data for the rule derivation. Six significant predictors of HP were identified: (1) exposure to a known offending antigen, (2) positive precipitating antibodies to the offending antigen, (3) recurrent episodes of symptoms, (4) inspiratory crackles on physical examination, (5) symptoms occurring 4 to 8 hours after exposure, (6) and weight loss. The area under the receiver operating characteristic curve was 0.93 (95% confidence interval: 0.90–0.95). The rule retained its accuracy when validated in a separate cohort of 261 patients. The diagnosis of HP can often be made or rejected with confidence, especially in areas of high or low prevalence, respectively, without BAL or biopsy.
Hypersensitivity pneumonitis (HP) is a pulmonary disease with symptoms of dyspnea and cough resulting from the inhalation of an antigen to which the patient has been previously sensitized. Acute and subacute HP represent the most active forms of the disease, which may become chronic while remaining progressive. HP may also evolve to end-stage lung (1). The diagnosis of HP has most often relied on an array of nonspecific clinical symptoms and signs developed in an appropriate setting (2), with demonstration of interstitial markings on chest radiographs, serum precipitating antibodies against offending antigens, a lymphocytic alveolitis on bronchoalveolar lavage (BAL), and/or a granulomatous reaction on lung biopsies.
When considered separately, none of these findings has proved useful in the diagnosis of HP. It is accepted that 20% of chest radiographs are normal in acute cases (3). Many reports and reviews have challenged the diagnostic value of serum precipitins for case finding (4, 5). BAL can provide supportive elements in the diagnosis of the disease. A normal lymphocyte BAL count rules out all but residual disease (6), but an alveolar lymphocytosis is not specific to HP (7, 8). Transbronchial biopsies are of limited usefulness, even when granulomas are found (9). Also, HP gives a typical but nonspecific pattern on high-resolution computed tomograms (HRCT) (10).
Several groups have recommended diagnostic criteria for HP (11–13), without their diagnostic accuracy being tested. Consequently, we conducted a prospective multicenter cohort study of patients presenting with a suspected diagnosis of acute, subacute, or chronic HP. The objective was to develop a clinical prediction rule for diagnosis of active HP (that is a clinical tool that quantifies the contribution that various components of the history, physical examination, and basic laboratory results make toward the diagnosis in an individual patient ). Such a rule aims at helping clinicians to arrive at a more accurate estimate of probability of HP and to decide whether further investigation is needed to either rule in or rule out HP. Some of the results of this study have been previously reported in the form of an abstract (15).
The HP study involved seven clinical sites from as many countries (see Appendix). Consecutive patients aged 18 years or older presenting with a pulmonary syndrome for which HP was considered in the differential diagnosis were included in the study. This cohort thus comprised patients with and without HP in a proportion that was unspecified a priori. We excluded patients with suspected Stage 1 sarcoidosis, those with a previous diagnosis of HP, and those referred for therapeutic evaluation of a known interstitial disease.
A review of the literature and consultations with content experts guided the development of a set of potential diagnostic criteria. These criteria included data usually collected during the initial investigation of patients with suspected HP (clinical history, physical examination, chest radiography marking patterns, pulmonary function tests, arterial blood gases, complete blood count, serum-precipitating antibodies). Other clinical characteristics unlikely to be found in HP (such as pleural effusion or serum antinuclear antibodies) were also considered, given their potential for making the diagnosis of HP less likely.
Attending clinicians systematically recorded the relevant data in every patient before making the final diagnosis. The symptoms and physical signs were recorded as present or absent on a standardized form during the initial clinical interview. The patients performed spirometry according to the American Thoracic Society requirements (16), lung volume measurement by plethysmography (17), and carbon monoxide diffusion capacity measurement by the single-breath method (18). The predicted values currently used within each laboratory were accepted. Arterial blood gases were measured while breathing room air. Partial pressure in oxygen was adjusted for altitude and reported as if it were obtained at sea level (19). The investigators tested for serum-precipitating antibodies in all patients using either the ELISA or the electrosyneresis methods (20) on a panel of antigens most likely to be encountered in their respective environments. The presence of antinuclear antibodies was tested by indirect immunofluorescent antibody staining of fixed cells (21). We reported precipitating antibodies and antinuclear antibodies as being either positive or negative according to center-specific predefined threshold values. Finally, the chest X-ray available at the time of the initial consultation was assessed.
The investigators had to classify each patient as HP or non-HP. In the absence of a unique gold standard defining the presence or absence of HP, the final diagnosis relied on findings of BAL, HRCT, and, if needed, other diagnostic procedures. BAL lymphocytosis (⩾ 30% for non- and exsmokers and ⩾ 20% for current smokers ) and bilateral ground-glass or poorly defined centrilobular nodular opacities on HRCT (10) were required for a diagnosis of HP to be accepted without resorting to additional diagnostic procedures. When the association of HRCT and BAL did not allow the investigators to arrive with confidence at a final diagnosis of HP or non-HP, the decision regarding additional procedures (including for instance BAL fluid cytology or culture, transbronchial or endobronchial biopsy, or mediastinoscopy) was not protocol-based but left to the investigators, according to clinical circumstances and their usual practice. Patients underwent surgical lung biopsy when the HRCT, the BAL, and other diagnostic procedures failed to yield a diagnosis. Pathologic criteria of HP included chronic inflammatory infiltrates along small airways, diffuse interstitial infiltrates of chronic inflammatory cells, and scattered, small, nonnecrotizing granulomas (23). In the case where a patient was classified as non-HP, a specific diagnosis was not mandatory, provided that HP was definitely excluded on the basis of BAL. Patients with inactive, late emphysematous, or fibrotic sequelae of HP were classified in the control group.
Four clinicians, one pathologist, and one radiologist were responsible for ascertaining the final diagnoses that were submitted by the clinical centers. The clinical committee first reviewed each case for data consistency with the investigator's final diagnosis and could request any additional information (such as data from the initial investigation or follow-up). Disagreement between the submitting center and the clinical adjudication committee led to a reassessment of the HRCT and, when available, biopsy material by a radiologist and a pathologist who were unaware of the submitted diagnosis. In addition to providing, when possible, a specific diagnosis from the HRCT or biopsy material, both classified each case as probable HP, indeterminate, or probable non-HP. The committee accepted the clinical diagnosis if (1) both the radiologist and the pathologist agreed with the clinical diagnosis or (2) the pathologist could provide the same specific diagnosis as the one submitted by the clinical center. The committee overturned the clinical diagnosis if (1) both the radiologist and the pathologist disagreed with the clinical diagnosis or (2) the pathologist could provide a specific diagnosis (different from the submitted one).
In this analysis, we sought to reduce the list of potential diagnostic criteria to those that would recognize a high proportion of patients with HP and exclude a high proportion of patients with other diseases (24). We first compared the clinical characteristics of the patients in the HP and non-HP groups using two-tailed Fischer's exact and unpaired t tests for dichotomous and continuous variables, respectively. From this analysis, we incorporated the variables found significant at the 0.05 level in a stepwise logistic regression model. Odds ratios and 95% confidence intervals (CI) for the multivariate predictors were identified. We then constructed a receiver operating characteristic (ROC) curve (that is a plot of the true-positive vs. false-positive rates of HP at various thresholds of probability) using the predicted probabilities of HP from the logistic regression analysis as diagnostic tests for HP. Also, we computed the area under the ROC curve, which indicates the probability that a random pair of HP and non-HP patients will be correctly classified as to their disease state (25).
We prospectively validated the rule using a separate cohort of patients that met the same inclusion criteria as those included in the development phase of the study. During the phase of validation, the investigators were kept blind to the criteria identified during the derivation phase and were asked to collect the same data. In this analysis, the model developed in the derivation phase of the study was evaluated by comparing the areas under the ROC curve (25). The hypothesis that the observed proportion of patients with HP was different from the predicted proportion in both the derivation and the validation cohorts was tested using the goodness-of-fit statistic described by Lemeshow and Hosmer (26). Finally, we constructed calibration curves, which are plots of the observed probability of HP compared with the predicted probability of HP ordered by the increasing probability of the disease (27).
In secondary analyses, to verify whether the rule can be applied in different settings, we cross-validated it by comparing, within each clinical site, the classification of each patient with their actual status. Also, to investigate the potential bias introduced by the use of information overlapping the criteria under study to define the presence or absence of HP, we constructed ROC curves using the subsets of patients who were submitted to surgical lung biopsy and those who were not.
The clinical prediction model yielded an equation expressing the probability of HP as a function of the statistically significant variables. From this equation, we constructed a table of probability for combinations of predictors.
Between February 1998 and September 2001, 728 patients were enrolled. Of these, 67 (9%) were excluded from the analysis by the adjudication committee for the following reasons: unavailable or uninterpretable BAL (n = 49) or HRCT (n = 9); inconsistent final diagnosis that could not be ascertained (n = 7); and Stage 1 sarcoidosis (n = 2). Thus, 661 patients (56% women; mean age: 55 years; standard deviation: 14) contributed to the analysis. The data from the first 400 patients (116 HP, 284 non-HP) were used to develop the prediction rule, whereas those of the next 261 patients (83 HP; 178 non-HP) were used to validate it. The distribution of the final diagnoses is summarized in Table 1
No. of Patients
|Diagnosis||Derivation Cohort||Validation Cohort|
|Patients with HP||116||83|
|Pigeon breeder's/bird fancier's disease||71||61|
|Various exposures to fungi||11||8|
|HP of unknown origin||2||1|
|Idiopathic interstitial pneumonia*||132||94|
|Interstitial disease associated with collagen vascular disease||24||11|
|Drug-induced pulmonary disease||18||8|
|Bronchiolitis obliterans (with or without organizing pneumonia)||15||10|
|Unspecified interstitial lung disease†||15||11|
|Bronchoalveolar carcinoma/carcinomatous lymphangitis||0||4|
|Organic dust toxic syndrome||1||2|
|Lymphocytic interstitial pneumonia||1||1|
|Pulmonary edema (heart failure)||1||1|
From the 18 variables that reached the level of statistical significance in the univariate analyses comparing the HP and non-HP groups (Table 2)
Patients with HP
|Exposure to antigens||97%||33%||< 0.001|
|Tightness of chest||35%||20%||0.002|
|Weight loss||42%||25%||< 0.001|
|Chest (pleuritic) pain||10%||5%||0.040|
|Symptoms 4–8 h after exposure||27%||2%||< 0.001|
|Recurrent episodes of symptoms||44%||12%||< 0.001|
|Smoking status: current smoker||6%||20%||< 0.001|
|Supraclavicular or cervical adenopathies||3%||4%||0.76|
|Positive precipitins to known antigens||78%||31%||< 0.001|
|PO2, mm Hg||70||74||0.001|
|White blood cell count, ×109/L||7.7||7.9||0.61|
|Positive antinuclear antibodies||19%||23%||0.43|
|Pulmonary function testings, % predicted value|
|Carbon monoxide diffusion capacity||62%||68%||0.06|
|Functional residual capacity||96%||91%||0.14|
|Total lung capacity||82%||81%||0.61|
|Normal chest X-ray||14%||5%||0.007|
|Exposure to a known offending antigen||3.66||38.8||11.6–129.6|
|Positive precipitating antibodies||1.68||5.3||2.7–10.4|
|Recurrent episodes of symptoms||1.20||3.3||1.5–7.5|
|Symptoms 4–8 h after exposure||1.97||7.2||1.8–28.6|
Applying the rule to the 261 patients in the prospective validation cohort, the area under the ROC curve was 0.90 (CI: 0.87–0.94) and was not statistically different from the area obtained in the derivation phase (p = 0.32). The goodness-of-fit statistics for the derivation (χ2 = 10.04; degrees of freedom = 7; p = 0.2) and validation (χ2 = 4.30; degrees of freedom = 7; p = 0.7) data sets indicated that the observed proportion of patients with HP was similar to the predicted proportion in both the derivation and validation groups. The calibration curves for the derivation and validation data sets demonstrated good calibration of the prediction rule (Figure 3).
The rule was associated with similar operating characteristics across the seven clinical sites, the area under the curve ranging from 0.83 to 1.00 (median: 0.93). The accuracy of the rule was less in the subgroup of patients who were submitted to surgical lung biopsy (area under the curve: 0.83; CI: 0.78–0.98) than in the subgroup of those who were not (area under the curve: 0.95; CI: 0.92–0.97; p = 0.001).
The probability of HP computed for each of the 64 (26) combinations generated by the six significant predictors of HP is presented in Table 4
|Serum Precipitins||Serum Precipitins|
|Exposure to a Known
Offending Antigen||Recurrent Episodes
of Symptoms||Symptoms 4–8 h
After Exposure||Weight Loss||+||−||+||−|
Interstitial lung diseases often pose diagnostic challenges, even to expert clinicians. Recent studies emphasized that additional investigations (including surgical biopsy) are indicated in patients with interstitial diseases in whom the diagnosis remains unclear after initial assessment (30, 31). The results of the HP Study indicated that a simple clinical prediction rule may guide clinical practice by providing estimates of the probability of acute, subacute, or chronic progressive HP from noninvasive testing. The predictors of HP we identified do not apply to chronic and inactive forms of the disease.
The best diagnostic strategy will then depend on the probability of HP. For instance, in a farmer presenting with recurrent episodes of respiratory symptoms and inspiratory crackles and testing positive for the corresponding precipitating antibodies, the probability of HP would be 81% (Table 4). Another patient presenting with progressive dyspnea and inspiratory crackles as the only criteria of HP would have a probability of HP of less than 1%. Further investigation would be mandated in the former. The typical findings of an alveolar lymphocytosis and/or bilateral ground-glass opacities on HRCT in the former patient would confirm the diagnosis of HP, without resorting to surgical lung biopsy. However, HP would be confidently ruled out in the latter and the investigation oriented elsewhere. We submit that a probability greater than or equal to 90% or less than or equal to 10% should be sufficient in most cases to, respectively, rule in or rule out HP, especially in areas of high or low prevalence of HP, respectively. However, the “test threshold” (that is the probability below which a clinician would dismiss the diagnosis and order no further test) and the “treatment threshold” (that is the probability above which a clinician could consider the diagnosis confirmed and would stop testing) are likely to differ according to the clinical implications of the diagnosis (32, 33). A clinician and his/her patient will be more likely to accept the diagnosis of bird fancier's disease when the offending antigen is a pet, even if the probability of HP is 75%. In such a case, antigen avoidance would be appropriate. Further investigation would be required only if the clinical course is unusual. On the other hand, a clinician and his/her patient will want to confirm the diagnosis of farmer's lung even if the probability of HP is around 90%, given that more than 50% will quit farming within 6 years of a diagnosis of farmer's lung (34).
Some of our findings deserve further attention. First, the enrollment of patients without interstitial disease stemmed from the inclusion criteria of the study: all patients presenting with a pulmonary syndrome for which active HP was considered were eligible. A typical example was the inclusion of patients (often farmers) who presented with recurrent episodes of dyspnea in whom the investigation ruled out HP and demonstrated asthma. Similarly, in four patients who presented with symptoms suggestive of HP, we could not demonstrate any lung disease. Second, the identification of a potential offending antigen is crucial for the clinical diagnosis of HP. Inquiry about occupational and home environmental exposures will most often uncover the cause of the disease. In rare instances (1.5% of our cohort), the diagnosis of HP was made without any identifiable offending antigen. In these cases, the diagnosis was supported by the careful exclusion of other causes of BAL lymphocytosis and typical findings of HP on lung biopsy. Third, although smoking is often thought of as having a protective effect against the development of HP (35), nonsmoking was not identified as an independent predictor of HP.
An important strength of the HP Study was its adherence to methodologic standards for both the derivation and the validation of the rule (14, 36). The patients were chosen in an unbiased fashion and represented a wide spectrum of diseases from a variety of institutions, hence increasing generalizability. We included all perceived important predictors, and those of significance were present in a large proportion of the study population. The predictors and outcomes were clearly defined. In the absence of a unique gold standard defining HP, every effort was made to properly classify each patient as either HP or non-HP independent of the criteria under study. The final diagnosis relied on the combination of typical findings on BAL and HRCT and the exclusion of competing diagnoses. Surgical lung biopsy was available in 35% of the patients. An adjudication committee reviewed every case for consistency. We submitted the difficult cases to further investigation, including clinical reassessment providing follow-up information and a blind and independent review of the HRCT and biopsy material when available. The accuracy of the rule being less in patients who were submitted to surgical lung biopsy than in those who were not does not necessarily indicate bias due to circularity. This may result from the fact that only the most difficult cases were submitted to surgical lung biopsy. Finally, considering an estimated annual incidence of interstitial lung disease of approximately 30 per 100,000 (37), the sample size was large and determined a priori. The derived rule made clinical sense because it matched the expected typical findings in HP.
Potential applications of prediction rules include periodic surveillance in high-risk workers or case finding in outbreaks of HP. Several rules have been developed for such purposes (38–40). Little information is available regarding their accuracy. Whether the rule we developed can be used for case finding of HP remains uncertain. Because “exposure to a known offending antigen” is the strongest predictor of HP in our cohort of patients presenting with a variety of interstitial lung diseases (Table 3), our rule may not retain its discriminative properties when used in a homogeneous population of workers, all of whom have been exposed to a common antigen.
Much confusion still surrounds the classification of HP. Its clinical presentations have classically been defined as acute, subacute, and chronic (12). The distinction between acute and subacute HP is often difficult as both likely represent different manifestations of a single disease that may be related more to the pattern of antigen exposure than to the offending antigen itself. This statement is supported by our finding of considerable overlap in the clinical manifestations of patients with farmer's lung (usually considered as the prototype of acute HP) and those with pigeon breeder's or bird fancier's diseases (the prototypes of subacute and chronic HP, respectively; data not shown). Also, chronic HP may still be active and progressive. Others have suggested a classification that takes into account the progression of the disease (acute intermittent, acute progressive, chronic progressive, chronic nonprogressive) that can be assessed only retrospectively (1, 41). For practical purposes, we suggest that patients with HP be considered as having either active or residual disease, the latter representing late emphysematous or fibrotic sequelae of the disease in which the typical alveolar lymphocytosis of active HP has disappeared.
The potential implications of such a simple rule are numerous. These simple criteria may establish uniformity in the definition of HP. By quantifying the individual contributions of various components of the history, physical examination, and basic laboratory results, the rule may reduce the number of unnecessary invasive procedures (including BAL or surgical lung biopsy) in patients with typical presentation of the disease. It may also rule out HP with confidence in those with low clinical scores. However, the HP Study will prove useful only if it increases awareness about HP and it can be used by others in various settings with confidence in its accuracy (36).
Yves Lacasse, M.D., M.Sc., Yvon Cormier, M.D. (Centre de Pneumologie, Hôpital Laval, Université Laval, Quebec, Canada), Mark Schuyler (University of New Mexico School of Medicine, Albuquerque, New Mexico).
Mexico (212 patients)—Instituto Nacional de Enfermedades Respiratorias, Mexico DF: Moises Selman, M.D., Andrea Estrada, M.D., Mayra Mejia, M.D., Teresa Suarez, M.D., Miguel Gaxiola, M.D., Guillermo Carrillo, M.D., Julio Robledo, M.D. Germany (123 patients)—Ruhrlandklinik, Essen: Ulrich Costabel, M.D., Norio Satake, M.D., Ricarda Cziborra, M.D., Josune Guzman, M.D. Canada (98 patients)—Hôpital Laval, Université Laval, Quebec, Quebec: Yvon Cormier, M.D., Gétane Bédard, R.N., Diane Page, R.N., Evelyne Assayag, M.Sc. France (90 patients)—Centre Hospitalier Universitaire de Besançon, Besancon: Jean-Charles Dalphin, M.D., Gabriel Reboux, Ph.D. Japan (55 patients)—Kumamoto University School of Medicine, Kumamoto: Masayuki Ando, M.D., Hisato Yamasaki, M.D. Spain (49 patients)—Hospital Universitari Vall d'Hebron, Barcelona: Ferran Morell, M.D., Leonardo Reyes, M.D. Finland (34 patients)—Kuopio University Hospital, Kuopio: Riitta Erkinjuntti-Pekkanen, M.D., Ph.D.
Clinicians: Yves Lacasse, M.D., M.Sc., Yvon Cormier, M.D., Michel Laviolette, M.D. (Hôpital Laval, Université Laval, Quebec, Quebec, Canada), Ulrich Costabel, M.D. (Ruhrlandklinik, Essen, Germany). Radiologist: Nestor Müller, M.D., Ph.D. (University of British Columbia, Vancouver, British Columbia, Canada); Pathologist: Thomas V. Colby, M.D. (Mayo Clinic Scottsdale, Scottsdale, Arizona).
Yves Lacasse, M.D., M.Sc., Sylvie Martin, M.Sc. (Hôpital Laval, Université Laval, Quebec, Quebec, Canada).
Yves Lacasse, M.D., M.Sc., Gaétan Daigle, M.Sc. (Hôpital Laval, Université Laval, Quebec, Quebec, Canada).
Y.L. has no declared conflict of interest; M.S. has no declared conflict of interest; U.C. has no declared conflict of interest; J.-C.D. has no declared conflict of interest; M.A. has no declared conflict of interest; F.M. has no declared conflict of interest; R.E.-P. has no declared conflict of interest; N.M. has no declared conflict of interest; T.V.C. has no conflict of interest; M.S. has no conflict of interest; Y.C. has no conflict of interest.
The authors would like to thank Dr. Gordon Guyatt for his comments on early drafts of the manuscript and the other members of the American Thoracic Society Workshop on Hypersensitivity Pneumonitis (Drs. Cecile Rose, Hal Richerson, and David Lynch) who provided useful suggestions for the design and implementation of the study.
|1.||Selman M. Hypersensitivity pneumonitis. In: Schwarz MI, King TE Jr, editors. Interstitial lung disease, 3rd ed. Hamilton, BC: Decker Inc.; 1998. p. 393–422.|
|2.||Schwartz M, Patterson R. Hypersensitivity pneumonitis—general considerations. Clin Rev Allergy 1983;1:451–467.|
|3.||Hodgson MJ, Parkinson DK, Karpf M. Chest X-rays in hypersensitivity pneumonitis: a meta-analysis of secular trends. Am J Ind Med 1989;16:45–53.|
|4.||Cormier Y, Bélanger J. The fluctuant nature of precipitating antibodies in dairy farmers. Thorax 1989;44:469–473.|
|5.||Burrel P, Bylander R. A critical review of the role of precipitins in hypersensitivity pneumonitis. Eur J Respir Dis 1981;62:332–343.|
|6.||Cormier Y, Bélanger J, Leblanc P, Laviolette M. Bronchoalveolar lavage in farmer's lung disease: diagnostic and physiological significance. Br J Ind Med 1986;43:401–405.|
|7.||Semenzato G, Bjermer L, Costabel U, Haslam PL, Olivieri D. Clinical guidelines and indications for bronchoalveolar lavage (BAL): report of the European society of pneumology task group on BAL: extrinsic allergic alveolitis. Eur Respir J 1990;3:945–946.|
|8.||Cormier Y, Bélanger J, Beaudoin J, Laviolette M, Beaudoin R, Hébert J. Abnormal bronchoalveolar lavage in asymptomatic dairy farmers: study of lymphocytes. Am Rev Respir Dis 1984;130:1046–1049.|
|9.||Lacasse Y, Fraser RS, Fournier M, Cormier Y. Diagnostic accuracy of transbronchial biopsy in the diagnosis of acute farmer's lung. Chest 1997;112:1459–1465.|
|10.||Matar LD, McAdams PH, Sporn TA. Hypersensitivity pneumonitis. Am J Roentgenol 2000;174:1061–1066.|
|11.||Terho EO. Diagnostic criteria for farmer's lung disease. Am J Ind Med 1986;10:329.|
|12.||Richerson HB, Bernstein IL, Fink JN, Hunninghake GW, Novey HS, Reed CE, Salvaggio JE, Schuyler MR, Schwartz HJ, Stechschulte DJ. Guidelines for the clinical evaluation of hypersensitivity pneumonitis. J Allergy Clin Immunol 1989;84:839–844.|
|13.||Schuyler M, Cormier Y. The diagnosis of hypersensitivity pneumonitis. Chest 1997;111:534–536.|
|14.||Laupacis A, Sekar N, Stiell IG. Clinical prediction rules: a review and suggested modifications of methodological standards. JAMA 1997;277:488–494.|
|15.||Lacasse Y, Selman M, Costabel U, Dalphin JC, Morell F, Ando M, Erkinjuntti-Pekkanen R, Müller N, Colby T, Schuyler M, et al. for the HP Study Group. Clinical prediction rule for the diagnosis of hypersensitivity pneumonitis (HP): the HP Study. Am J Respir Crit Care Med 2002;165:A198.|
|16.||American Thoracic Society. Standardization of spirometry—1994 update. Am J Respir Crit Care Med 1995;152:1107–1136.|
|17.||Quanjer PhH, Tammeling GJ, Cotes JE, Pedersen OF, Peslin R, Yernault JC. Lung volumes and forced ventilatory flows: official statement of the European respiratory society. Eur Respir J 1993;6(Suppl 16):5–40.|
|18.||American Thoracic Society. Single-breath carbon monoxide diffusing capacity (transfer factor): recommendations for a standard technique—1995 update. Am J Respir Crit Care Med 1995;152:2185–2198.|
|19.||Chi-Lem G, Perez-Padilla R. Gas exchange at rest during simulated altitude in patients with chronic lung diseases. Arch Med Res 1998;29:57–62.|
|20.||Reboux G, Piarroux R, Mauny F, Madroszyk A, Millon L, Bardonnet K, Dalphin JC. Role of molds in farmer's lung disease in Eastern France. Am J Respir Crit Care Med 2001;163:1534–1539.|
|21.||Sacher RA, McPherson RA. Widmann's clinical interpretation of laboratory tests, 10th ed. Philadelphia, PA: FA Davis; 1990.|
|22.||The BAL Cooperative Group Steering Committee. Bronchoalveolar lavage constituents in healthy individuals, idiopathic pulmonary fibrosis, and selected comparison groups. Am Rev Respir Dis 1990;141:S169–S202.|
|23.||Coleman A, Colby TV. Histologic diagnosis of extrinsic allergic alveolitis. Am J Surg Pathol 1988;2:514–518.|
|24.||Bloch DA, Moses LE, Michel BA. Statistical approaches to classification: methods for developing classification and other criteria rules. Arthritis Rheum 1990;33:1137–1144.|
|25.||Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36.|
|26.||Lemeshow S, Hosmer DW. A review of goodness-of-fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982;115:92–106.|
|27.||Swensen SJ, Silverstein MD, Ilstrup DM, Schleck CD, Edell ES. The probability of malignancy in solitary pulmonary nodules: application to small radiologically indeterminate nodules. Arch Intern Med 1997;157:849–855.|
|28.||Katzenstein AA, Meyers JL. Idiopathic pulmonary fibrosis: clinical relevance of pathologic classification. Am J Respir Crit Care Med 1998;157:1301–1315.|
|29.||Hansen JE, Casaburi R. Patterns of dissimilarities among instruments models in measuring po2, pco2 and pH in blood gas laboratories. Chest 1998;113:780–787.|
|30.||Hunninghake GW, Zimmerman MB, Schwartz DA, King TE Jr., Lynch J, Hegele R, Waldron J, Colby T, Muller N, Lynch D, et al. Utility of a lung biopsy for the diagnosis of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2001;164:193–196.|
|31.||Raghu G, Mageto YN, Lockhart D, Schmidt RA, Wood DE, Godwin JD. The accuracy of the clinical diagnosis of new-onset idiopathic pulmonary fibrosis and other interstitial lung disease: a prospective study. Chest 1999;116:1168–1174.|
|32.||Jaeschke R, Guyatt G, Sackett DL. Users' guides to the medical literature: III. How to use an article about a diagnostic test: B. What are the results and will they help me in caring for my patients? JAMA 1994;271:703–707.|
|33.||Pauker SG, Kassirer JP. The threshold approach to clinical decision making. N Engl J Med 1980;302:1109–1117.|
|34.||Bouchard S, Morin F, Bedard G, Gauthier J, Paradis J, Cormier Y. Farmer's lung and variables related to the decision to quit farming. Am J Respir Crit Care Med 1995;152:997–1002.|
|35.||Dalphin JC, Debieuvre D, Pernet D, Maheu MF, Polio JC, Toson B, Dubiez A, Monnet E, Laplante JJ, Depierre A. Prevalence and risk factors for chronic bronchitis and farmer's lung in French dairy farmers. Br J Ind Med 1993;50:941–944.|
|36.||McGinn TG, Guyatt GH, Wyer PC, Naylor CD, Stiell IG, Richardson WS. Users' guides to the medical literature. XXII: how to use articles about clinical decision rules. JAMA 2000;284:79–84.|
|37.||Coultas DB, Zumwalt RE, Black WC, Sobonya RE. The epidemiology of interstitial lung diseases. Am J Respir Crit Care Med 1994;150:967–972.|
|38.||Sullivan PA, Odencrantz JR, Petsonk EL, Fox JL, Trout D. Development and validation of a hypersensitivity pneumonitis surveillance questionnaire [abstract]. Am J Respir Crit Care Med 1997;155:A946.|
|39.||Fox J, Anderson H, Moen T, Gruetzmacher G, Hanrahan L, Fink J. Metal working fluid associated hypersensitivity pneumonitis: an outbreak investigation and case control study. Am J Ind Med 1999;35:58–67.|
|40.||Dangman KH, Cole SR, Hodgson MJ, Kuhn C, Metersky ML, Schenck P, Storey E. The hypersensitivity pneumonitis diagnostic index: use of non-invasive testing to diagnose hypersensitivity pneumonitis in metalworkers. Am J Ind Med 2002;42:150–162.|
|41.||Boyd D, McSharry CP, Banham SW, Lynch PP. A current view of pigeon fancier's lung: a model for pulmonary extrinsic allergic alveolitis. Clin Allergy 1983;12:53–59.|