Published online before print December 12, 2005, 10.1148/radiol.2381041823
(Radiology 2005;238:725.)
A more recent version of this article appeared on December 1, 2005
© RSNA, 2005
Interstitial Lung Disease: Effects of Thin-Section CT on Clinical Decision Making1
Zelena A. Aziz, MD,
Athol U. Wells, MD,
Eric D. Bateman, MD,
Susan J. Copley, MD,
Sujal R. Desai, MD,
Jan C. Grutters, MD,
David G. Milne, MD,
Gerrard D. Phillips, MD,
David Smallwood, MD,
John Wiggins, MD,
Margaret L. Wilsher, MD and
David M. Hansell, MD
1 From the Department of Radiology, Royal Brompton Hospital, Sydney St, London SW3 6NP, England. Received October 24, 2004; revision requested December 23; revision received February 1, 2005; accepted February 24; final version accepted April 27.
Address correspondence to D.M.H. (e-mail: d.hansell{at}rbh.nthames.nhs.uk).
 |
ABSTRACT
|
|---|
Purpose: To retrospectively quantify the change in the diagnosis and management of suspected interstitial lung disease when thin-section computed tomography (CT) is added to pretest probabilities.
Materials and Methods: The institutional review board does not require approval or patient informed consent for retrospective study of case records and CT studies. Six pulmonologists reviewed data sheets containing clinical information and results of pulmonary function tests and chest radiographs of 168 consecutive patients (86 women and 82 men; mean age, 59.8 years; age range, 22–86 years) suspected of having interstitial lung disease. Differential diagnoses and responses to specific questions regarding patient care were recorded before and after assimilation of thin-section CT findings. Both unweighted and weighted 
analyses were used to determine agreement between pulmonologists before and after CT.
Results: First-choice diagnosis changed in 520 (51%) of 1008 cases, and agreement on first-choice diagnosis increased from 0.47 to 0.72 after thin-section CT. In addition, confidence in the first-choice diagnosis increased, and there was a reduction in the number of differential diagnoses offered by all pulmonologists (P < .005 and P < .001, respectively). Agreement on diagnostic probabilities for individual disorders increased substantially, particularly for diagnoses of idiopathic pulmonary fibrosis (weighted = 0.58–0.89). With CT findings, pulmonologists changed their pre-CT responses regarding the use of bronchoalveolar lavage, transbronchial biopsy, and thoracoscopic biopsy in 242 (24.0%), 282 (28.0%), and 292 (29.0%) of 1008 cases, respectively. However, agreement for the use of these investigations was low both before and after CT. The request rate for thoracoscopic biopsy in patients in whom idiopathic fibrosis was diagnosed decreased from 48 of 179 (26.8%) to 26 of 233 (11.2%) after CT.
Conclusion: Thin-section CT resulted in a change in first-choice diagnosis in half the cases. Diagnostic confidence improved, and CT findings increased agreement between pulmonologists on diagnostic probabilities across a range of interstitial lung diseases.
© RSNA, 2005
|
INTRODUCTION
|
|---|
Thin-section computed tomography (CT) has become an integral part of the assessment of patients suspected of having interstitial lung disease; however, evaluation of all aspects of the performance of thin-section CT as a diagnostic test is incomplete. The majority of studies have focused on the diagnostic accuracy of thin-section CT and used histologic analysis as the reference standard (1–3). However, the perception of histologic analysis as the reference standard in the diagnosis of diffuse lung disease is changing (4). Several reports have described the problems of sampling error (5) and observer variation between histopathologists (6), as well as the limitations of histologic analysis in the prediction of survival rates (7,8). In addition to these limitations, surgical biopsy is increasingly reserved for patients in whom thin-section CT findings and clinical diagnoses are inconclusive. The current school of thought is that the reference standard should be an integrated process that draws on the opinions of clinicians, radiologists, and pathologists (9).
This radical shift in thinking requires a change in the evaluation of the usefulness of a diagnostic test to an approach that does not merely rely on verification of CT observations with histologic findings. Fryback and Thornbury (10) describe a six-tiered hierarchical model for the assessment of diagnostic imaging tests. These tiers are as follows: level 1, technical efficacy; level 2, diagnostic accuracy; level 3, effect on diagnostic thinking; level 4, therapeutic efficacy; level 5, effect on patient outcome; and level 6, societal efficacy. The effect of thin-section CT on diagnostic thinking in interstitial lung disease has been largely neglected, primarily because measurement of this effect is difficult. Nevertheless, one of the fundamental properties of a diagnostic test is the degree to which it alters the perception of a clinical diagnosis (11). In patients with interstitial lung disease, clinical history, physical examination findings, abnormal findings on chest radiographs, and laboratory investigations all contribute to an overall impression, from which a differential diagnosis is formulated. Thin-section CT findings may subsequently lead physicians to modify, refine, or radically change a presumptive diagnosis and treatment plan.
Thus, the diagnostic process proceeds with a series of consecutive estimations of the probability of a particular disease. Thus, it is essential to quantify, in isolation, the contribution of the different (ie, clinical, radiologic, and pathologic) components of the diagnostic process. This concept is fully understood in the evaluation of patients suspected of having a pulmonary embolism (12–14). Here, clinical guidelines promote the integration of results from different diagnostic tests, but this has come about only as a consequence of numerous studies that have been performed to separately evaluate the contribution of each test (ie, clinical probability, D-dimer analysis, ventilation-perfusion scanning, and CT pulmonary angiography) to the diagnosis of pulmonary embolism. The results of these studies have given clinicians an idea for how tests might be expected to change their diagnostic perceptions; however, to our knowledge, no comparable guidance exists in patients with interstitial lung disease. The specific aim of this study was to retrospectively quantify the change in the diagnosis and management of suspected interstitial lung disease when thin-section CT is added to pretest probabilities.
|
Materials and Methods
|
|---|
Patient Population
The study population consisted of 168 consecutive patients who were referred from June 1992 to July 1998 to a regional teaching hospital (Royal Brompton Hospital, London, England) for further diagnostic work-up because they were known to have or suspected of having interstitial lung disease (mean age, 59.8 years ± 16.8 [± standard deviation]; age range, 22–86 years; 86 women). This hospital was responsible for all interstitial lung disease referrals from three secondary hospitals and served a population equivalent to that of a medium-sized city (population, 1.1 million). All patients underwent pulmonary function tests, chest radiography, and thin-section CT within 1 month of each other.
Case records of the patients were obtained, and clinical data sheets were compiled for each patient by a single consultant respiratory physician (A.U.W.) with a specialist interest in interstitial lung disease and 18 years of experience. All relevant clinical information, including a comment on previous longitudinal behavior (if this information was available), was included. Details included in the data sheet are presented in the Appendix. Available pulmonary function indexes included forced expiratory volume in 1 second, forced vital capacity, forced expiratory volume in 1 second divided by forced vital capacity, total lung capacity, carbon monoxide transfer factor and coefficient, and 50% and 25% midexpiratory flow. These indexes were expressed as percentages of values predicted for patient age, sex, and height (15). Our institutional review board does not require its approval or patient informed consent for retrospective study of case records and CT studies.
CT Protocol and Image Evaluation
Prior to CT scanning, patients were placed in the supine position. Thin-section CT images were obtained with a single–detector row CT scanner (Somatom DR; Siemens, Iselin, NJ) and the following parameters: 1.5-mm collimation, 10-mm interspace, and breath holding at full inspiration. CT images were reconstructed with a high-spatial-frequency algorithm and photographed at window settings appropriate for viewing the lung parenchyma (window center, –550 HU; window width, 1500 HU).
Two experienced thoracic radiologists (S.J.C., S.R.D.) with specific training in the interpretation of CT images in patients with interstitial lung disease independently reviewed the CT images on hard-copy film. These radiologists had completed thoracic radiology fellowships 3 and 6 years previously, respectively. They had no knowledge of clinical findings or details of the patient population. They were asked to list their differential diagnosis, with no limit to the number of diagnoses, and to assign a likelihood (to the nearest 5%, totaling 100%) to each diagnosis. Radiologists were not given specific diagnostic criteria for the different interstitial lung diseases, but they used criteria based on current CT literature and their own experience. They were required to use the terminology of the American Thoracic Society and the European Respiratory Society (16) for the idiopathic interstitial pneumonias.
Pulmonologist Characteristics
Six pulmonologists (E.D.B., J.C.G., G.D.P., D.S., J.W., M.L.W.) from four continents participated in this study. These physicians ranged in age from 32 to 55 years, and their clinical experience ranged from 1 to 22 years. They all received previous training at an interstitial lung disease unit and developed an interest in interstitial lung diseases in their own practice. Two of the pulmonologists began their clinical practice before the advent of thin-section CT (ie, before 1990); the other four physicians completed their pulmonary fellowships during the development of thin-section CT.
Assessment by Pulmonologists
Each pulmonologist reviewed patient data sheets, results of pulmonary function tests, and chest radiographs of each patient. Prior to CT, pulmonologists completed a pro forma that included specific questions regarding the diagnosis and care of the patient (Appendix). The chest radiographs of each patient were digitized (Vidar Systems, Herndon, Va) at a resolution of 150 dpi, and the participating pulmonologists reviewed these images on compact disks with read-only memory. The pulmonologists were required to state their differential diagnoses and assign a likelihood (to the nearest 5%, totaling to 100%) to each one. A preselected list of possible diagnoses and specific diagnostic criteria was not given; the only stipulation was that terminology of the American Thoracic Society and the European Respiratory Society for the idiopathic interstitial pneumonias (16) should be used if an idiopathic interstitial pneumonia was diagnosed. Responses to the questions regarding the use of lavage, transbronchial biopsy, and thoracoscopic biopsy were graded as follows: 1, definitely yes; 2, on balance yes; 3, on balance no; and 4, definitely no. The terms on balance yes and on balance no were included to allow pulmonologists to indicate their level of confidence in their decision to treat patients and request additional tests. These terms were explained and understood by the pulmonologists prior to commencement of the study. When the pre-CT pro formas had been completed and submitted, the thin-section CT reports were sent to the pulmonologists. Reports for half of the cohort were from one radiologist, and reports for the other half were from another radiologist. The format of the thin-section CT report is shown in the Appendix. After assimilation of the thin-section CT report, pulmonologists completed a second pro forma that consisted of questions that were identical to those in the first pro forma. The pulmonologists could access their responses to the pre-CT questions when answering the post-CT questions.
Data and Statistical Analysis
After the pro formas were collected, and prior to analysis, diseases were classified into 13 categories (Fig 1). IPF is used in this article in accordance with the recent classification of the American Thoracic Society and the European Respiratory Society, which regards IPF as synonymous with the histologic pattern of usual interstitial pneumonia (16). All statistical analyses were performed by using Stata data analysis software (version 4.0; Computing Resource Center, Santa Monica, Calif).
View larger version (64K):
[in this window]
[in a new window]
|
Figure 1: Chart shows categories of disease that were used for statistical analysis. IPF = idiopathic pulmonary fibrosis.
|
|
Quantification of Change in Diagnosis
The frequency of change in the first-choice diagnosis after CT and the frequency of the post-CT diagnosis being in agreement with the first-choice diagnosis were calculated. Interobserver agreement on the first-choice diagnosis for the entire cohort before and after CT was quantified by using the unadjusted coefficient, as follows: = 0–0.20, poor; = 0.21–0.40, fair; = 0.41–0.60, moderate; = 0.61–0.80, good; = 0.81–1.00, excellent (17).
Interobserver agreement was quantified for IPF, nonspecific interstitial pneumonia, smoking-related interstitial lung disease, cryptogenic organizing pneumonia, sarcoidosis, hypersensitivity pneumonitis, interstitial pneumonias related to connective tissue disease, and other interstitial lung diseases by using the unadjusted coefficient.
The weighted coefficient of agreement was used to calculate interobserver agreement in the estimation of the probability of individual diseases for the entire cohort before and after CT. To do this, the percentage likelihood given to each diagnosis was assigned a grade representing clinically useful probabilities: grade 0, condition not included in the differential diagnosis; grade 1, low probability (5%–25% likelihood); grade 2, intermediate probability (30%–65% likelihood); grade 3, high probability (70%–95% likelihood); and grade 4, pathognomonic (100% likelihood). By placing percentages into numerical categories, it was possible to assess agreement over a range of clinically useful probabilities in individuals with interstitial lung disease. Weighted coefficient with quadratic weighting (18) was computed for IPF, nonspecific interstitial pneumonia, smoking-related interstitial lung disease, cryptogenic organizing pneumonia, sarcoidosis, hypersensitivity pneumonitis, and interstitial pneumonias related to connective tissue disease. Weighted values were calculated by paired observers and expressed as median values with ranges for the 15 possible combinations of six observers (eg, observer 1 vs observer 2, observer 1 vs observer 3). The paired t test was used to compare weighted values before and after CT. P values of less than .05 were considered to indicate statistically significant differences.
The first-choice diagnosis of each patient was given a confidence rating of 1–3 on the basis of the original percentages assigned by the pulmonologists. A confidence rating of 1 was assigned if diagnostic likelihood was less than 70%, which translated to low confidence. A confidence rating of 2 was assigned if diagnostic likelihood was between 70% and 95%, which translated to high confidence. A confidence rating of 3 was assigned if diagnostic likelihood was 100%, which indicated findings were pathognomonic. These categories were based on those used to assess the clinical probability of pulmonary embolism in the Prospective Investigation of Pulmonary Embolism Diagnosis study (19). Confidence ratings of the first-choice diagnosis for the six observers were summed. Thus, the highest potential confidence rating for the six observers was 18, and the lowest potential confidence rating was 6. The confidence ratings before and after CT were compared by using the Wilcoxon signed rank test. The number of differential diagnoses stated before and after CT was compared with the Wilcoxon signed rank test.
Quantification of Change in Care
Responses to the questions regarding bronchoalveolar lavage, transbronchial biopsy, and thoracoscopic biopsy were compiled for the six pulmonologists, and the scores before and after CT were compared by using the Wilcoxon signed rank test. The agreement of the six pulmonologists on the use of each investigation before and after CT was quantified by using the unadjusted coefficient.
For each pulmonologist, decisions regarding their use of further tests before and after CT were compared. The percentage of decisions that were changed after CT for all six pulmonologists was calculated. Further analysis was performed on the most frequently encountered idiopathic interstitial pneumonias, IPF, to determine whether the percentage of thoracoscopic biopsy requests changed after CT. Interobserver agreement of the radiologists on the first-choice diagnosis was quantified by using the unadjusted coefficient.
|
Results
|
|---|
Quantification of Change in Diagnosis
The first-choice diagnosis after assimilation of CT findings by pulmonologists was changed in 520 (51.6%) of 1008 cases (Figs 2, 3). Individual percentage changes of each pulmonologist are shown in Table 1. After CT, pulmonologists disagreed with the radiologists' first-choice diagnosis in 251 (24.9%) of 1008 cases. The coefficient of agreement for the first-choice diagnosis was 0.47 (ie, moderate) before CT and 0.72 (ie, good) after CT.
View larger version (128K):
[in this window]
[in a new window]
|
Figure 2: Transverse thin-section CT image obtained at the level of the right main pulmonary artery shows linear opacities and extensive ground-glass opacification. Several low-attenuation focal areas are evident, which suggests air trapping (arrows). The first-choice diagnosis of radiologists was hypersensitivity pneumonitis. The first-choice diagnosis of pulmonologists before CT was either IPF or nonspecific interstitial pneumonia. After CT, all pulmonologists altered their diagnosis to hypersensitivity pneumonitis.
|
|
View larger version (131K):
[in this window]
[in a new window]
|
Figure 3: Transverse thin-section CT image at the level of the right hemidiaphragm. Ground-glass opacification in association with traction bronchiectasis and a fine reticular pattern are demonstrated. In this case, six different first-choice diagnoses were offered by the pulmonologists. The first-choice diagnosis of radiologists was nonspecific interstitial pneumonia. CT findings resulted in all pulmonologists changing their original diagnosis to nonspecific interstitial pneumonia.
|
|
Agreement for individual diseases offered as a first-choice diagnosis was moderate to excellent ( = 0.46–0.85) after CT and poor to moderate ( = 0.10–0.58) before CT. There were notable increases in observer agreement after CT for diagnoses of IPF, nonspecific interstitial pneumonia, cryptogenic organizing pneumonia, and smoking-related interstitial lung disease. Smaller increases in agreement were seen in patients with hypersensitivity pneumonitis, sarcoidosis, or interstitial pneumonias related to connective tissue disease. The coefficients for first-choice diagnoses are shown in Table 2.
View this table:
[in this window]
[in a new window]
|
Table 2. Unweighted Values for the Pulmonologists' First-Choice Diagnosis of Individual Categories before and after CT
|
|
Weighted coefficients were calculated for the seven diagnoses, with sufficient prevalence to allow analysis (Table 3). Weighted coefficients increased considerably for all the idiopathic interstitial pneumonias after CT; the greatest increase was seen in IPF after CT (weighted value increased from 0.58 to 0.89). There were minimal increases in the weighted coefficient for hypersensitivity pneumonitis and interstitial pneumonias related to connective tissue disease; this increase was not significant for hypersensitivity pneumonitis. The change in weighted values was highly significant for IPF, nonspecific interstitial pneumonia, smoking-related interstitial lung disease, cryptogenic organizing pneumonia, sarcoidosis, and the interstitial pneumonias related to connective tissue disease (Table 3).
The confidence of the six pulmonologists for the first-choice diagnosis rose from 10.9 ± 3.0 (± standard deviation) to 13.6 ± 2.6 (1 = lowest confidence score, 3 = highest confidence score) (P < .001). Pulmonologists suggested fewer differential diagnoses after CT (median number of diagnoses, 2; range, 1–5) compared with the number of differential diagnoses suggested before CT (median number of diagnoses, 3; range, 1–7) (P < .001).
Quantification of Change in Care
After the assimilation of thin-section CT findings, the frequency of requesting lavage, transbronchial biopsy, and thoracoscopic biopsy decreased by (a) 3.3% (ie, from 686 [68.1%] to 653 [64.8%] of 1008 cases) for lavage, (b) 3.8% (ie, from 463 [45.9%] to 424 [42.1%] of 1008 cases) for transbronchial biopsy, and (c) 6.1% (388 [38.5%] to 327 [32.4%] of 1008 cases) for thoracoscopic biopsy. In all instances, this change was statistically significant (P < .001).
Agreement among the pulmonologists for all investigations was poor before CT. For lavage, the value was 0.17; for transbronchial biopsy, 0.26; and for thoracoscopic biopsy, 0.29. After CT, clinicians changed their pre-CT response regarding use of further investigations in (a) 242 (24.0%) (ie, lavage), (b) 282 (28.0%) (ie, transbronchial biopsy), and (c) 292 (29.0%) (ie, thoracoscopic biopsy) of 1008 cases. Despite the change in response, values remained almost identical for lavage ( = 0.14), transbronchial biopsy ( = 0.26), and thoracoscopic biopsy ( = 0.28).
The number of IPF cases stated as a confident first-choice diagnosis (ie, diagnostic probability of 70%–100%) increased by 9.4% (from 105 [10.4%] to 199 [19.8%] of 1008 cases) after CT. Prior to CT, the biopsy rate in patients with an IPF diagnosis was 48 (26.8%) of 179 cases; the biopsy rate decreased to 26 (11.7%) of 223 cases after CT, even though more IPF cases were identified after CT. The coefficient of agreement for first-choice diagnosis for the radiologists was 0.68 (ie, good agreement).
|
Discussion
|
|---|
In this study, the first-choice diagnosis was changed after the integration of CT findings in half of the cases. Additionally, thin-section CT findings resulted in an increase in the number of confident IPF diagnoses; this is a reflection of the influence of CT on the diagnosis of IPF. Although agreement between pulmonologists for the first-choice diagnosis was moderate before CT, it increased considerably after assimilation of CT findings. The fact that in a quarter of cases the post-CT diagnosis assigned by pulmonologists did not match the first-choice diagnosis assigned by radiologists indicates that CT was not used indiscriminately; rather, in certain cases, pulmonologists were more reliant on clinical information.
The use of the weighted coefficient to evaluate agreement in the estimation of diagnostic probabilities of individual diseases before and after CT was crucial to this study. In our study and in clinical practice, the diagnosis in certain cases may be a close call. For example, one pulmonologist may choose differential diagnoses of IPF and hypersensitivity pneumonitis with certainties of 45% and 55%, respectively, whereas another pulmonologist may favor differential diagnoses of IPF and hypersensitivity pneumonitis with certainties of 55% and 45%, respectively. The use of the unweighted coefficient in this circumstance would indicate complete disagreement between the two observers, despite the fact that the percentage probabilities assigned to the conditions by the two pulmonologists were very similar. By converting percentage probabilities into five numerical categories and then applying the weighted coefficient, we were better able to assess agreement across a range of probabilities. A further advantage of this analysis is that by using a category (grade 0) to represent cases in which a particular diagnosis was not included in the list of differential diagnoses, the weighted coefficient also reflects agreement for the exclusion of a specific diagnosis. Thus, the increase in the weighted coefficient from 0.58 to 0.89 for the diagnosis of IPF after CT is striking. The ability of CT findings to increase diagnostic convergence in determining the presence or absence of IPF is relevant in light of numerous findings that emphasize the poor prognosis and survival of patients with IPF compared with the other idiopathic interstitial pneumonias (7,20–23).
Although CT did increase agreement for the diagnosis of sarcoidosis, the good agreement prior to CT (w = 0.68) reflects the fact that pulmonologists often diagnose sarcoidosis without use of thin-section CT. In a statement on sarcoidosis, thin-section CT was not recommended as part of the diagnostic work-up (24), except when the presentation is atypical. It is evident that in cases of hypersensitivity pneumonitis in which there is an obvious precipitant, CT adds little to pretest diagnostic perception; the modest rise in the weighted value for hypersensitivity pneumonitis after CT was not significant. Similarly, a minimal increase in agreement was observed after CT for the interstitial pneumonias related to connective tissue disease. Features of connective tissue disease are often evident at presentation.
Thin-section CT caused pulmonologists to change their decisions regarding the necessity of lavage, transbronchial biopsy, and thoracoscopic biopsy in a quarter of cases. However, the proposed use of these tests after CT decreased marginally (albeit significantly). Additionally, our findings emphasize a striking lack of agreement between pulmonologists in deciding which patients required further testing both before and after CT. This unexpected disparity reflects the inherent subjectivity of management decisions. In a diverse group of pulmonologists, disagreement on the level of testing required to achieve acceptable diagnostic certainty is inescapable. The considerable heterogeneity of opinion that exists in this area suggests the need for specific guidelines with a greater emphasis on management decisions than those that currently exist (16).
When the whole cohort was analyzed, the thoracoscopic biopsy request rate decreased by only 6%. When IPF was the first-choice diagnosis, however, the perceived need for biopsy fell from 27% to 12% after CT. This finding is in keeping with the findings of two studies that showed that biopsy may not be needed in patients in whom there is a confident clinical and imaging diagnosis of IPF (25,26). Both studies also suggested that biopsy is most useful in patients who do not have IPF or when there is clinical or radiologic ambiguity. In our study, the high biopsy rates for the entire cohort—even with CT information—suggests that, apart from the IPF group, diagnostic biopsy is favored by pulmonologists. Indeed, CT information may increase the biopsy rate in non-IPF cases.
In 1994, Grenier et al (27) evaluated the integration of CT into clinical diagnosis. Bayesian analysis was used to quantify the relative contributions of clinical assessment, chest radiography, and CT to clinical diagnosis. The overall diagnostic sensitivity was 80% when all the available information was integrated; the addition of CT increased diagnostic sensitivity considerably, raising it from 54% when only clinical and radiographic data were considered. However, a statement of sensitivity does not fully encapsulate the effect of thin-section CT on the diagnostic process. We used end points relevant to management and clinical decision making. In the context of the Fryback and Thornbury (10) model for the evaluation of diagnostic tests, this approach is a logical progression in the evaluation of CT.
A strength of our study is that cases were not limited to those with a histologically proved diagnosis; lung biopsy tends to be performed in only select cases (ie, those in which confident diagnoses are not otherwise possible). A study confined to biopsy-proved cases is inherently biased. Our study differs from clinical practice in that radiologists were not given access to clinical information. It was fundamental that the influence of CT findings to the diagnostic process was determined in isolation. This method is not at odds with the view that diagnoses should be reached jointly by clinicians and radiologists; rather, it is a necessary prerequisite to the adoption of a multidisciplinary approach.
One of the limitations of this study was that the reliability of the anonymous radiologists (apart from the knowledge that they were experienced thoracic radiologists) was not known to the pulmonologists. It is certainly possible that reports from radiologists with whom the pulmonologists were familiar might have had an effect on their diagnostic perception. An area where the design deviated from clinical practice was in the presentation of the radiology report. The radiologists were asked to provide a percentage likelihood of their differential diagnoses, which was conveyed to the pulmonologists. Radiology reports do not normally contain numerical expressions of probability; however, an important consideration for pulmonologists in the interpretation of a diagnostic report is the confidence with which conclusions are made. In theory, reports given with a high degree of confidence add greater weight to the posttest probability of a diagnosis being correct; therefore, they should have a greater effect on clinical decision making (28). In a study designed to evaluate changes in diagnostic perception, conveying the confidence of radiologists was crucial.
In our study, we evaluated thin-section CT from a broader perspective than the traditional ambit of diagnostic accuracy. However, future studies that will address the important issue of the effect of CT on patient health outcomes–levels 4, 5, and 6 of Fryback and Thornbury's model—are now required. Assessment of CT at these levels will be a considerable challenge because the conventional methods of evaluating the effect of a test on health outcomes are not readily applicable to patients with the heterogeneous entity of "interstitial lung disease."
In conclusion, thin-section CT substantially influenced pulmonologist behavior, changing diagnostic impressions in half the cases, resulting in a more confident first-choice diagnosis and increasing agreement among pulmonologists. These findings should facilitate the integration of clinical and thin-section CT information in an era when diagnostic decisions in interstitial lung disease are increasingly taken collectively.
|
Appendix
|
|---|
The clinical data sheets for each patient included both general and detailed information (Fig A1). Prior to and after CT, pulmonologists completed a pro forma that included specific questions regarding the evaluation and care of the patient (Fig A2). The format of the thin-section CT report is also shown (Fig A3).
View larger version (41K):
[in this window]
[in a new window]
|
Figure A1: Clinical data sheets, which were completed for each patient, included both general and detailed information.
|
|
View larger version (52K):
[in this window]
[in a new window]
|
Figure A2: Pro forma completed by pulmonologists before and after CT. ATS = American Thoracic Society, CXR = chest radiography, ERS = European Respiratory Society.
|
|
View larger version (46K):
[in this window]
[in a new window]
|
Figure A3: An example of the format in which the radiology report was conveyed to the pulmonologists. HRCT = thin-section CT.
|
|
|
FOOTNOTES
|
|---|
Abbreviations: IPF = idiopathic pulmonary fibrosis
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, D.M.H.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, Z.A.A.; clinical studies, E.D.B., S.R.D., J.C.G., D.G.M., G.D.P., D.S., J.W., M.L.W., D.M.H.; statistical analysis, Z.A.A., A.U.W.; and manuscript editing, Z.A.A., A.U.W., G.D.P., M.L.W., D.M.H.
|
References
|
|---|
- Mathieson JR, Mayo JR, Staples CA, Muller NL. Chronic diffuse infiltrative lung disease: comparison of diagnostic accuracy of CT and chest radiography. Radiology 1989;171:111–116.[Abstract/Free Full Text]
- Padley SP, Hansell DM, Flower CD, Jennings P. Comparative accuracy of high resolution computed tomography and chest radiography in the diagnosis of chronic diffuse infiltrative lung disease. Clin Radiol 1991;44:222–226.[CrossRef][Medline]
- Nishimura K, Izumi T, Kitaichi M, Nagai S, Itoh H. The diagnostic accuracy of high-resolution computed tomography in diffuse infiltrative lung diseases. Chest 1993;104(4):1149–1155.[Abstract/Free Full Text]
- Wells AU. Histopathologic diagnosis in diffuse lung disease: an ailing gold standard. Am J Respir Crit Care Med 2004;170(8):828–829.[Free Full Text]
- Flaherty KR, Travis WD, Colby TV, et al. Histopathologic variability in usual and nonspecific interstitial pneumonias. Am J Respir Crit Care Med 2001;164:1722–1727.[Abstract/Free Full Text]
- Nicholson AG, Addis BJ, Bharucha H, et al. Inter-observer variation between pathologists in diffuse parenchymal lung disease. Thorax 2004;59(6):500–505.[Abstract/Free Full Text]
- Flaherty KR, Thwaite EL, Kazerooni EA, et al. Radiological versus histological diagnosis in UIP and NSIP: survival implications. Thorax 2003;58(2):143–148.[Abstract/Free Full Text]
- Latsi PI, Du Bois RM, Nicholson AG, et al. Fibrotic idiopathic interstitial pneumonia: the prognostic value of longitudinal functional trends. Am J Respir Crit Care Med 2003;168(5):531–537.[Abstract/Free Full Text]
- Flaherty KR, King TE Jr, Raghu G, et al. Idiopathic interstitial pneumonia: what is the effect of a multi-disciplinary approach to diagnosis? Am J Respir Crit Care Med 2004;170(8):904–910.[Abstract/Free Full Text]
- Fryback DG, Thornbury JR. The efficacy of diagnostic imaging. Med Decis Making 1991;11(2):88–94.[Abstract/Free Full Text]
- Tsushima Y, Aoki J, Endo K. Contribution of the diagnostic test to the physician's diagnostic thinking: new method to evaluate the effect. Acad Radiol 2003;10(7):751–755.[CrossRef][Medline]
- Fedullo PF, Tapson VF. Clinical practice: the evaluation of suspected pulmonary embolism. N Engl J Med 2003;349(13):1247–1256.[Free Full Text]
- British Thoracic Society Standards of Care Committee Pulmonary Embolism Guideline Development Group. British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 2003;58(6):470–483.[Free Full Text]
- American College of Emergency Physicians Clinical Policies Committee; Clinical Policies Committee Subcommittee on Suspected Pulmonary Embolism. Clinical policy: critical issues in the evaluation and management of adult patients presenting with suspected pulmonary embolism. Ann Emerg Med 2003;41(2):257–270. [Published correction appears in Ann Emerg Med 2003;42(2):288.]
- Quanjer PH. Standardized lung function testing. Eur Respir J 1993;6(suppl):1–100.
- American Thoracic Society, European Respiratory Society. American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias. Am J Respir Crit Care Med 2002;165(2):277–304. [Published correction appears in Am J Respir Crit Care Med 2002;166(3):426.]
- Coblentz CL, Babcook CJ, Alton D, Riley BJ, Norman G. Observer variation in detecting the radiologic features associated with bronchiolitis. Invest Radiol 1991;26(2)115–118.[Medline]
- Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York, NY: Wiley, 1981; 212–236.
- The PIOPED Investigators. Value of the ventilation/perfusion scan in acute pulmonary embolism: results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED). JAMA 1990;263:2753–2759.[Abstract]
- Bjoraker JA, Ryu JH, Edwin MK, et al. Prognostic significance of histopathologic subsets in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 1998;157(1):199–203.
- American Thoracic Society. Idiopathic pulmonary fibrosis: diagnosis and treatment—international consensus statement. Am J Respir Crit Care Med 2000;161(2):646–664.[Free Full Text]
- Nicholson AG, Colby TV, Du Bois RM, et al. The prognostic significance of the histologic pattern of interstitial pneumonia in patients presenting with the clinical entity of cryptogenic fibrosing alveolitis. Am J Respir Crit Care Med 2000;162:2213–2217.[Abstract/Free Full Text]
- Daniil ZD, Gilchrist FC, Nicholson AG, et al. A histologic pattern of nonspecific interstitial pneumonia is associated with a better prognosis than usual interstitial pneumonia in patients with cryptogenic fibrosing alveolitis. Am J Respir Crit Care Med 1999;160(3):899–905.[Abstract/Free Full Text]
- Hunninghake GW, Costabel U, Ando M, et al. ATS/ERS/WASOG statement on sarcoidosis. American Thoracic Society/European Respiratory Society/World Association of Sarcoidosis and other Granulomatous Disorders. Sarcoidosis Vasc Diffuse Lung Dis 1999;16(2):149–173.[Medline]
- Hunninghake GW, Zimmerman MB, Schwartz DA, et al. Utility of a lung biopsy for the diagnosis of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2001;164:193–196.[Abstract/Free Full Text]
- Raghu G, Mageto YN, Lockhart D, et al. The accuracy of the clinical diagnosis of new-onset idiopathic pulmonary fibrosis and other interstitial lung disease: a prospective study. Chest 1999;116(5):1168–1174.[Abstract/Free Full Text]
- Grenier P, Chevret S, Beigelman C, et al. Chronic diffuse infiltrative lung disease: determination of the diagnostic value of clinical data, chest radiography, and CT and Bayesian analysis. Radiology 1994;191(2):383–390.[Abstract/Free Full Text]
- Wells AU. High-resolution computed tomography in the diagnosis of diffuse lung disease: a clinical perspective. Semin Respir Crit Care Med 2003;24:347–356.[CrossRef][Medline]
TOP
ABSTRACT
INTRODUCTION
