0
Review | Clinician's Corner

Accuracy of Diagnostic Tests Read With and Without Clinical Information:  A Systematic Review FREE

Clement T. Loy, MBBS, FRACP; Les Irwig, MBBCh, PhD
[+] Author Affiliations

Author Affiliations: Screening and Test Evaluation Program, School of Public Health, University of Sydney, Sydney, Australia.

More Author Information
JAMA. 2004;292(13):1602-1609. doi:10.1001/jama.292.13.1602.
Text Size: A A A
Published online

Context Although it is common practice to read tests with clinical information, whether this improves or decreases the accuracy of test reading is uncertain.

Objective To determine whether diagnostic tests are more accurate when read with clinical information or without it.

Data Sources MEDLINE search (1966-December 2003) extended by search of reference lists and articles citing the articles retrieved (Web of Science, 1985-December 2003).

Study Selection All articles comparing the accuracy of tests read twice by the same readers, once without and once with clinical information, but otherwise under identical conditions. Only articles that reported sensitivity and specificity or receiver operating characteristic (ROC) curves were included.

Data Extraction Data were extracted by one author and reviewed independently by the other. When the data were difficult to interpret, differences were resolved by discussion.

Data Synthesis Sixteen articles met the inclusion criteria. Eleven articles compared areas under ROC curves for tests read with and without clinical information, and 5 compared only sensitivity and specificity. Ten articles used actual clinical information; 6 used constructed clinical information that was plausible. Overall, clinical information improved test reading accuracy although the effect was smaller in the articles using actual clinical information when compared with those using constructed clinical information. There were no instances in which clinical information resulted in significant reduction in test reading accuracy. In some instances, improved test reading accuracy came from improved sensitivity without loss of specificity.

Conclusions At least for the tests examined, the common practice of reading diagnostic tests with clinical information seems justified. Future studies should be designed to investigate the best way of providing clinical information. These studies should also give an estimate of the accuracy of clinical information used, display ROC curves with identified data points, and include a wider range of diagnostic tests.

Figures in this Article

Whether diagnostic tests should be read with clinical information has been much disputed since Schreiber1 suggested 40 years ago that clinical information improved the accuracy of chest x-ray reading. Quiz Ref IDAn argument favoring reading diagnostic tests with clinical information is that the accuracy of the read may be improved by the additional information. It may help at 2 stages: perception and interpretation. Consider a 78-year-old woman undergoing a ventilation-perfusion (V/Q) scan for left-sided chest pain that develops 10 days after a total hip replacement. Clinical information may focus the reader’s perception to the relevant area—the left lung. Knowing that the patient had a hip replacement 10 days ago, the reader may also consciously or unconsciously alter his/her interpretation of subtle cues or alter his/her overall level of suspicion for pulmonary embolism.

An argument favoring reading diagnostic tests without clinical information is that it may bias the reading and that clinical information should be incorporated into decision making only after an unbiased read.2 For example, clinical information may bias the reader into seeing perfusion defects that may not be present. In addition, the clinical information would be double-counted if it influenced both the initial test reading and subsequent interpretation of test result by the physician ordering the test. It is also conceivable that clinical information can defeat the purpose of a diagnostic test. For example, if the physician who ordered the V/Q scan thought that a pulmonary embolus was unlikely in this patient clinically and informed the radiologist, the radiologist might respond by raising the threshold for diagnosis. This is equivalent to increasing the V/Q scan’s specificity and lowering its sensitivity, which may make the exclusion of pulmonary embolism on a negative test result less definite because specific tests are less useful in excluding disease than sensitive ones.3

We therefore systematically reviewed the literature to determine whether a diagnostic test is more accurate when read with clinical information or without it.

Data Sources, Study Selection

The published literature was searched by one of the authors (C.T.L.) using MEDLINE (1966 to December 2003) and 2 strategies. For the first strategy, we searched MEDLINE through OVID’s explode function (EXP) using (EXP medical history taking and EXP sensitivity and specificity) followed by an extended search which included examination of reference lists of all included articles. Additional articles citing the articles thus identified were located and examined using Web of Science (1985-December 2003). Using Web of Science proved to be a useful strategy in identifying articles not easily retrievable on MEDLINE because the relevant studies came from a wide variety of disciplines and there were a few early seminal articles that were likely to have been quoted by subsequent authors. For the second strategy, we searched MEDLINE through PubMed using the word history and the specific version of the “Clinical Queries” filter diagnosis.

For each of the strategies, abstracts were read for all articles with relevant titles. We obtained full articles for abstracts that suggested that the articles were relevant.

Quiz Ref IDWe included all articles that compared the accuracy of tests against a reference standard with or without clinical information and that had been read twice by the same readers under otherwise identical conditions. We accepted all articles that measured sensitivity and specificity or that had receiver operating characteristic (ROC) curves.

Data Extraction

Extraction of data on study characteristics and test accuracy were performed by one author (C.T.L.). Difficult to interpret articles were reviewed by the other author (L.I.), and differences were resolved by discussion. Quality assessment was performedusing the Quality Assessment of Diagnostic Accuracy Studies tool,4 as well as additional criteria specific to methodological issues relating to this research question, which included whether clinical information was actual or constructed, whether a balanced design was used, what the amount of time was between reading sessions, and whether alternate ways of providing clinical information was considered.

Test accuracy can be quantified in terms of sensitivity and specificity and ROC curves. Measures such as percentage agreement and repeatability (eg, κ statistic) address internal consistency but not accuracy and, therefore, are not helpful in this context.

Sensitivity is the proportion of individuals with a disease who have positive test results for that particular disease while specificity is the proportion of individuals without a disease who have negative test results for that particular disease. When the threshold of a test is varied, sensitivity and specificity can be traded off, without necessarily altering the overall test accuracy. An ROC curve is formed when test accuracy estimates for a test at several thresholds are joined together. On an ROC curve, a change in threshold is represented by shifts along the curve. On the other hand, any improvement in the overall diagnostic test accuracy is represented by an upward and leftward shift of the curve. The accuracy of 2 tests can be compared by looking at the areas under their ROC curves.

We used the area under the curve in our systematic review, wherever possible, to assess whether reading a test with clinical information improved its overall accuracy. This allowed us to distinguish observed changes in sensitivity and specificity due to true changes in overall test accuracy from those due to threshold shift alone.

Whenever available, confidence intervals were calculated using SEs and variances reported in the articles. P values quoted in the articles are also listed with levels of significance set at P<.05. Data on sensitivity and specificity were only used when data on the area under the curve were not available.

Using the combined Ovid–Web of Science strategy, we identified 39 articles that studied the impact of clinical information on test reading. We excluded 23 articles: 12 because they only measured percentage change, agreement, or proportion correct516 ; 6 because different readers read the tests with and without clinical information1722 ; 3 because they only assessed the impact of different clinical information, without comparing that against a test reading without clinical information2325 ; 1 because previous films were included in the clinical information26 ; and 1 because the study’s readers received test reading training between the 2 reading sessions.27

Using the specific version of the “Clinical Queries” filter diagnosis and a broad search term history in PubMed, 3593 entries were found identifying only 1 of the 16 articles found with the first strategy. These 3593 entries were also examined, finding no additional articles meeting our inclusion criteria. Using the sensitive version of the same filter, we found 88 880 entries that identified only 12 of the 16 articles using our first strategy.

Of the 16 articles thus identified, 11 reported areas under the ROC curve,2838 and 5 reported only sensitivities and specificities.1,3942 Nine of these 16 reports were from 3 research groups: (1) Berbaum et al,2832 (2) Cooperstein et al33 and Good et al,34 and (3) Rickett et al41 and Tudor et al.42 Two pairs of articles used the same sets of radiographs—(1) Berbaum et al30,31 and (2) Cooperstein et al33 and Good et al340dash; but the radiographs were read either by readers from different specialties30,31 or were different readers.33,34

Details on the topic and design of these articles can be found in Table 1. Diagnostic tests examined included cytology of bronchial washings; radiographs of the chest, abdomen, and bones; mammograms; and computed tomography of the head.

Table Graphic Jump LocationTable 1. Details of Articles Included*

Assessment of these articles using the Quality Assessment of Diagnostic Accuracy Studies tool found that all 16 articles clearly described their selection criteria although only 7 of these were representative of patients receiving the test in practice. Four did not specify the reference standard used, and 1 did not provide sufficient information about the reference standard to enable quality assessment. The remaining 11 of the 16 articles all used well-described reference standards performed over the whole sample within a reasonable time. Long-term clinical follow-up was often used as the reference standard for controls. In that sense, different reference standards were used for participants with different results at the time of initial testing, but this should not produce a bias in study results. All tests were interpreted without knowledge of results of the reference standard although only 1 study specifically stated a procedure to ensure independence of the reference standard from clinical information and test.36 Fifteen of the 16 articles included uninterpretable or intermediate results.

In addition to general assessment of quality using the Quality Assessment of Diagnostic Accuracy Studies tool, we identified 4 other methodological issues that are important in assessing articles on the effect of clinical information: whether clinical information was actual or constructed, whether a balanced design was used, what the amount of time was between reading sessions, and whether alternative ways of providing clinical information was considered. How each article addressed these issues is shown in Table 1 and summarized below.

Actual clinical information was used in 10 of the 16 articles, with sensitivity and specificity of the clinical information extractable from only 1 of them.39 Constructed but plausible clinical information was used in 6 of the 16 articles, with sensitivity and specificity of the clinical information used inferable in all 6 of them.2832,35

An article was categorized as having adopted a balanced design when half of the tests were read without and half with clinical information in each of the 2 reading sessions. This controls for any changes in reader ability or reporting practice that might have occurred between the reading sessions. Ten of the 16 articles had a balanced design.

The time between reading sessions was usually several months, making it unlikely that readers remembered their previous responses.

Only two28,32 of the 16 articles considered alternative ways of providing clinical information.

Of the 11 articles that measured areas under the ROC curve, 6 presented only the area or the modeled ROC curves.28,3236 The other 5 plotted individual points of sensitivity and specificity on the ROC curves.

Articles Reporting Areas Under the ROC Curve

Figure 1 shows the areas under the ROC curve measured for test reading with or without clinical information, listed in decreasing magnitude of the difference. Whenever available or calculable, confidence intervals are displayed. Pvalues quoted in the articles are also listed. When different subsets of diseases or readers in the articles were considered as different entries, 9 of 16 of the entries reported significantly improved areas under the ROC curve. None suggested a significantly smaller area under the ROC curve when tests were read with clinical information. Actual clinical information was used in 9 of the 16 entries and constructed clinical information in 7 of the 16 entries.

Figure 1. Areas Under the Receiver Operating Characteristic Curve for Diagnostic Tests Read With or Without Clinical Information
Graphic Jump Location

Articles are presented in order by the difference in point estimates. For entries for which data reported were insufficient for calculation of confidence intervals, no error bars are displayed. Reported P values are for differences in areas under the receiver operating characteristic (ROC) curve in diagnostic tests read with or without clinical information. Some articles reported areas under the ROC curve for different subsets of readers or films. We have listed these subsets as separate entries in the display with descriptions for the subsets.
*Not significant with significance level set at P<.05.

Figure 1 also demonstrates substantial variation in results from the 16 entries. One possible reason for this is the different types of clinical information used. Figure 1 also identifies which studies used actual clinical information and which used constructed but plausible information. Entries that used actual clinical information were less likely to show improvement than entries that used constructed clinical information: 3 of 9 that used actual clinical information reported significant improvement while 6 of 7 of the entries that used constructed clinical information reported significant improvement. Constructed clinical information may be more effective in improving test reading because it may have greater accuracy than actual clinical information. As a typical example, one of the fracture studies35 used the site of tenderness as the clinical information for all the abnormal cases, giving a sensitivity of 100%. Half of the normal cases were labeled with sites of tenderness, and the others not—thus yielding a specificity of 50%. The other entries that used constructed clinical information also had schemes that generated clinical information with high sensitivity. Depending on how exactly the clinical information is used to aid test reading, this may have a greater prompting effect than actual clinical information, which is rarely 100% sensitive. There was not sufficient quantitative data on the accuracy of the clinical information used to allow for further analysis of the extent to which this affects the improvement of test reading.

Three other features in the data may explain the variation observed. First, the largest improvements with clinical information occurred when test accuracy had the smallest areas under the ROC curve when read without clinical information. Small areas under the ROC curve, in some instances at least, may be due to selection of subtle rather than routine radiographs as study material. Six of 7 entries that used constructed clinical information selected subtle radiographs as study material (all from Berbaum et al2832 ) while none of the 9 entries that used actual clinical information did so. Second, 6 of 9 entries that used actual clinical information were from studies by Cooperstein et al33 and Good et al,34 both of which used comprehensive clinical histories rather than the short suggestive cues used in the other articles. These comprehensive clinical histories may have had a lesser prompting effect than the suggestive cues. Third, some readers may depend on clinical information more than others. For example, the increase in accuracy when provided with clinical information was greater for orthopedists than for radiologists reading the same set of bone radiographs with subtle fractures of the extremities and cervical spine.30,31

Articles Reporting Only Sensitivity and Specificity

Five articles only measured sensitivity and specificity with and without clinical information (Figure 2). All of these articles used actual clinical information. Four of these reported improvement in sensitivity without loss of specificity.1,39,41,42 This suggests improvement in overall test accuracy rather than a shift in diagnostic threshold. Interestingly, data points in some of the articles with ROC curves also suggested that clinical information improved sensitivity to a greater extent than specificity.2931,37

Figure 2. Sensitivity and Specificity for Diagnostic Tests Read With or Without Clinical Information
Graphic Jump Location

All 5 articles used actual clinical information.

Although the number of studies on which to base our conclusion is small, we found, overall, that clinical information improved test reading accuracy. Since providing clinical information is common practice, it would seem reasonable not to alter that process unless studies show that it is deleterious to reader accuracy, and we found no examples of that in the studies to date. On the basis of these results, the common practice of reading diagnostic tests with clinical information seems reasonable.

Quiz Ref IDIn some circumstances, at least, improvement in test reading with clinical information came from improved sensitivity without loss of specificity. This reflects overall improvement in test reading accuracy without any evidence of threshold shift, which is consistent with the reader being alerted to particular additional features rather than merely altering the level of suspicion because of the clinical information.

Studies in this area will benefit from further improvements in design, conduct, and reporting.

Repeatability was used as the sole outcome measure in many excluded articles. Studies measuring accuracy by ROC curves or sensitivity and specificity are more useful. The recently published Standards for Reporting of Diagnostic Accuracy statement43 may be a useful guideline for the planning and conduct of diagnostic test evaluation studies. Because there are additional specific issues in assessing the impact of providing clinical information on test reading, we suggest that several items should be assessed and reported in future studies and discuss several issues that would most benefit from further study.

First, the variation of results found between articles using actual and constructed clinical information may be partly explained by the difference in accuracy of the clinical information used. A number of other articles found worsening test accuracy when constructed, but misleading information was used.23,24 It also seems intuitive that the accuracy of clinical information itself would determine its impact on test reading. Therefore reporting of future studies should include data on the accuracy of clinical information used. This can be done by measuring sensitivity and specificity of clinical information against the same reference standard. Ideally cross-tabulation of clinical information against test results within the diseased and nondiseased groups should also be provided.

Second, because test reading with clinical information was found to improve sensitivity without loss of specificity in some of the articles, it would be helpful to plot and label individual points of sensitivity and specificity on the ROC curves in all future studies to assess whether this is a constant finding.

Third, because data in this area of research are sparse, further high-quality studies could usefully be performed across a wider range of diagnostic tests and clinical situations to explore the reasons for the heterogeneity that we have observed. For example, is clinical information most helpful for test readings that are more subjective? Does clinical information that helps identify the site of a lesion have different effects on test reading compared with clinical information that alters a reader’s assessment of the prior probability of disease? Does reader perception of the value of clinical information predict improvements in accuracy?

Quiz Ref IDThe study design used in many of the articles, consisting of 2 reading sessions separated by a washout period and balanced for reading with or without clinical information across sessions is valid. However, it does require setting up a study in which experts may read tests differently from how tests are read in actual practice. Therefore, consideration should be given to designing studies of reading with or without clinical information integrated into routine testing. These studies should also measure the amount of time required to report tests under each of these conditions. Although it is possible that clinical information helps to shorten the time needed for test reading, reading with clinical information may also take longer. If reading with clinical information does improve accuracy, then a decision will need to be made as to whether the improvement is worth the extra time.

Fourth, available evidence is compatible with continuing the common current practice of test readers only issuing a report after being provided with clinical information. However, the best way of providing clinical information is unclear, which suggests a major area worthy of further investigation. To date, this has only been examined in 2 studies (both by Berbaum et al28,32). One study32 found higher test accuracy of x-ray reading when clinical information was consulted before rather than after reading the test but without the opportunity to reread the test. The other28 found improved test accuracy of x-ray reading when clinical information and tentative diagnoses were provided. The magnitude of improvement was similar whether the information was provided before an initial reading or after an initial reading with the opportunity to reread.

Quiz Ref IDConceptually, clinical information may influence test reading at 2 stages: perception and interpretation. Perception refers to the identification of abnormal areas and their features and may be divided into initial (first impression) or overall perception. Interpretation refers to the attribution of observed abnormalities to a disease process. It will be important to assess empirically the effect of providing clinical information in ways that impact at different stages of the perception-interpretation sequence.

Typically, clinical information is provided before a test is read. In this case, consciously or unconsciously, clinical information may assist the reader both in the perception and interpretation stages. A possible disadvantage of this approach is that perception may be biased by the clinical information (Table 2). On the other hand, a diagnostic test can be read, clinical information provided, and the test reread before final reporting. This may have the advantage of allowing unbiased initial perception yet retaining the opportunity to improve overall perception and it may have the advantage of aiding interpretation after consulting the clinical information. Although this option seems to offer the best of both worlds, it is conceivable that there may be benefits from clinical information that will only make a difference on initial perception. To understand how clinical information might most help, further studies are needed to investigate the impact of providing clinical information at different stages of the perception-interpretation sequence.

Table Graphic Jump LocationTable 2. Different Ways of Providing Clinical Information During Test Reading*

Corresponding Author: Les Irwig, MBBCh, PhD, Screening and Test Evaluation Program, School of Public Health, Room 301, A27, Edward Ford Bldg, University of Sydney, NSW 2006, Australia (lesi@health.usyd.edu.au).

Access to Data: Dr Loy had full access to all of the data and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding/Support: This study was supported in part by grant 211205 from the National Medical Research Council to the Screening and Test Evaluation Program.

Role of the Sponsor: The study sponsor did not influence the design or conduct of the study; the collection, analysis, interpretation, or preparation of the data; or the preparation, review, or approval of the manuscript.

Schreiber MH. The clinical history as a factor in roentgenogram interpretation.  JAMA. 1963;185:399-401
PubMed   |  Link to Article
Griscom NT. A suggestion: look at the images first, before you read the history.  Radiology. 2002;223:9-10
PubMed   |  Link to Article
Sackett D, Straus S, Richardson W, Rosenberg W, Haynes RB. Evidence-Based MedicineEdinburgh, Scotland: Churchill Livingstone; 2000:74
Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews.  BMC Med Res Methodol. 2003;3:25
PubMed   |  Link to Article
Abbey LM, Kaugars GE, Gunsolley JC.  et al.  The effect of clinical information on the histopathologic diagnosis of oral epithelial dysplasia.  Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1998;85:74-77
PubMed   |  Link to Article
Berst MJ, Dolan L, Bogdanowicz MM, Stevens MA, Chow S, Brandser EA. Effect of knowledge of chronologic age on the variability of pediatric bone age determined using the Greulich and Pyle standards.  AJR Am J Roentgenol. 2001;176:507-510
PubMed   |  Link to Article
Dunn PM, Levinson W. The lack of effect of clinical information on electrocardiographic diagnosis of acute myocardial infarction.  Arch Intern Med. 1990;150:1917-1919
PubMed   |  Link to Article
Hatala RA, Norman GR, Brooks LR. The effect of clinical history on physicians’ ECG interpretation skills.  Acad Med. 1996;71:(suppl 10)  S68-S70
PubMed   |  Link to Article
Hatala R, Norman GR, Brooks LR. Impact of a clinical scenario on accuracy of electrocardiogram interpretation.  J Gen Intern Med. 1999;14:126-129
PubMed   |  Link to Article
Leslie A, Jones AJ, Goddard PR. The influence of clinical information on the reporting of CT by radiologists.  Br J Radiol. 2000;73:1052-1055
PubMed
Paradise JE, Winter MR, Finkel MA, Berenson AB, Beiser AS. Influence of the history on physicians’ interpretations of girls’ genital findings.  Pediatrics. 1999;103:980-986
PubMed   |  Link to Article
Simons M, Parker JA, Donohoe KJ, Udelson JE, Gervino EV. The impact of clinical data on interpretation of thallium scintigrams.  J Nucl Cardiol. 1994;1:365-371
PubMed   |  Link to Article
Van Gijn J, Bonke B. Interpretation of plantar reflexes: biasing effect of other signs and symptoms.  J Neurol Neurosurg Psychiatry. 1977;40:787-789
PubMed   |  Link to Article
Vogel HP. Influence of additional information on interrater reliability in the neurologic examination.  Neurology. 1992;42:2076-2081
PubMed   |  Link to Article
Whited JD, Horner RD, Hall RP, Simel DL. The influence of history on interobserver agreement for diagnosing actinic keratoses and malignant skin lesions.  J Am Acad Dermatol. 1995;33:603-607
PubMed   |  Link to Article
Williams GW, Lesser RP, Silvers JB.  et al.  Clinical diagnoses and EEG interpretation.  Cleve Clin J Med. 1990;57:437-440
PubMed   |  Link to Article
Bonke B, Koudstaal PJ, Dijkstra G.  et al.  Detection of lacunar infarction in brain CT-scans: no evidence of bias from accompanying patient information.  Neuroradiology. 1989;31:170-173
PubMed   |  Link to Article
Bonke B, Kwakernaak A, Van Den Berg M, Koudstaal PJ. Bias in the assessment of computed-tomography scans for lacunar infarction.  Cerebrovasc Dis. 1992;2:107-110
Link to Article
Doubilet P, Herman PG. Interpretation of radiographs: effect of clinical history.  AJR Am J Roentgenol. 1981;137:1055-1058
PubMed   |  Link to Article
Eldevik OP, Dugstad G, Orrison WW, Haughton VM. The effect of clinical bias on the interpretation of myelography and spinal computed tomography.  Radiology. 1982;145:85-89
PubMed
Gjorup T, Saurbrey N, Hermann N. Clinical estimation of the duration of pregnancy in legal abortion–are doctors biased by their knowledge of the duration of amenorrhoea?  Methods Inf Med. 1984;23:96-98
PubMed
Potchen EJ, Gard JW, Lazar P, Lahaie P, Andary M. The effect of clinical history data on chest film interpretation: direction or distraction.  Invest Radiol. 1979;14:404
Ashworth CS, Fargason CA Jr, Fountain K, Cutter G, Centor R. Impact of patient history on residents’ evaluation on child sexual abuse.  Child Abuse Negl. 1995;19:943-951
PubMed   |  Link to Article
Babcook CJ, Norman GR, Coblentz CL. Effect of clinical history on the interpretation of chest radiographs in childhood bronchiolitis.  Invest Radiol. 1993;28:214-217
PubMed   |  Link to Article
Flach SD, Canaris GJ, Tape TG, Huntley KM, Wigton RS. The effects of proficiency and bias on residents’ interpretation of the microscopic urinalysis.  Med Decis Making. 2002;22:318-325
PubMed
Quekel LG, Goei R, Kessels AG, van Engelshoven JM. Detection of lung cancer on the chest radiograph: impact of previous films, clinical information, double reading, and dual reading.  J Clin Epidemiol. 2001;54:1146-1150
PubMed   |  Link to Article
Grum CM, Gruppen LD, Woolliscroft JO. The influence of vignettes on EKG interpretation by third-year students.  Acad Med. 1993;68:(suppl 10)  S61-S63
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD.  et al.  Tentative diagnoses facilitate the detection of diverse lesions in chest radiographs.  Invest Radiol. 1986;21:532-539
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD, Barloon TJ. Influence of clinical history upon detection of nodules and other lesions.  Invest Radiol. 1988;23:48-55
PubMed   |  Link to Article
Berbaum KS, el-Khoury GY, Franken EA Jr, Kathol M, Montgomery WJ, Hesson W. Impact of clinical history on fracture detection with radiography.  Radiology. 1988;168:507-511
PubMed
Berbaum KS, Franken EA Jr, el-Khoury GY. Impact of clinical history on radiographic detection of fractures: a comparison of radiologists and orthopedists.  AJR Am J Roentgenol. 1989;153:1221-1224
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD, Lueben KR. Influence of clinical history on perception of abnormalities in pediatric radiographs.  Acad Radiol. 1994;1:217-223
PubMed   |  Link to Article
Cooperstein LA, Good BC, Eelkema EA.  et al.  The effect of clinical history on chest radiograph interpretations in a PACS environment.  Invest Radiol. 1990;25:670-674
PubMed   |  Link to Article
Good BC, Cooperstein LA, DeMarino GB.  et al.  Does knowledge of the clinical history affect the accuracy of chest radiograph interpretation?  AJR Am J Roentgenol. 1990;154:709-712
PubMed   |  Link to Article
Ehara S, Katsuragawa S. Influence of clinical information on the detection of wrist fractures in children.  Tohoku J Exp Med. 1999;189:147-153
PubMed   |  Link to Article
McNeil BJ, Hanley JA, Funkenstein HH, Wallman J. Paired receiver operating characteristic curves and the effect of history on radiographic interpretation: CT of the head as a case study.  Radiology. 1983;149:75-77
PubMed
Raab SS, Oweity T, Hughes JH.  et al.  Effect of clinical history on diagnostic accuracy in the cytologic interpretation of bronchial brush specimens.  Am J Clin Pathol. 2000;114:78-83
PubMed   |  Link to Article
Song KS, Song HH, Park SH.  et al.  Impact of clinical history on film interpretation.  Yonsei Med J. 1992;33:168-172
PubMed
Elmore JG, Wells CK, Howard DH, Feinstein AR. The impact of clinical history on mammographic interpretations.  JAMA. 1997;277:49-52
PubMed   |  Link to Article
Kinnunen J, Gothlin J, Totterman S.  et al.  The effects of clinical data on radiological diagnostic performance in middle-face trauma.  ROFO Fortschr Geb Rontgenstr Nuklearmed. 1987;147:143-146
PubMed   |  Link to Article
Rickett AB, Finlay DB, Jagger C. The importance of clinical details when reporting accident and emergency radiographs.  Injury. 1992;23:458-460
PubMed   |  Link to Article
Tudor GR, Finlay D, Taub N. An assessment of inter-observer agreement and accuracy when reporting plain radiographs.  Clin Radiol. 1997;52:235-238
PubMed   |  Link to Article
Bossuyt PM, Reitsma JB, Bruns DE.  et al.  Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative.  BMJ. 2003;326:41-44
PubMed   |  Link to Article

Figures

Figure 1. Areas Under the Receiver Operating Characteristic Curve for Diagnostic Tests Read With or Without Clinical Information
Graphic Jump Location

Articles are presented in order by the difference in point estimates. For entries for which data reported were insufficient for calculation of confidence intervals, no error bars are displayed. Reported P values are for differences in areas under the receiver operating characteristic (ROC) curve in diagnostic tests read with or without clinical information. Some articles reported areas under the ROC curve for different subsets of readers or films. We have listed these subsets as separate entries in the display with descriptions for the subsets.
*Not significant with significance level set at P<.05.

Figure 2. Sensitivity and Specificity for Diagnostic Tests Read With or Without Clinical Information
Graphic Jump Location

All 5 articles used actual clinical information.

Tables

Table Graphic Jump LocationTable 1. Details of Articles Included*
Table Graphic Jump LocationTable 2. Different Ways of Providing Clinical Information During Test Reading*

References

Schreiber MH. The clinical history as a factor in roentgenogram interpretation.  JAMA. 1963;185:399-401
PubMed   |  Link to Article
Griscom NT. A suggestion: look at the images first, before you read the history.  Radiology. 2002;223:9-10
PubMed   |  Link to Article
Sackett D, Straus S, Richardson W, Rosenberg W, Haynes RB. Evidence-Based MedicineEdinburgh, Scotland: Churchill Livingstone; 2000:74
Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews.  BMC Med Res Methodol. 2003;3:25
PubMed   |  Link to Article
Abbey LM, Kaugars GE, Gunsolley JC.  et al.  The effect of clinical information on the histopathologic diagnosis of oral epithelial dysplasia.  Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1998;85:74-77
PubMed   |  Link to Article
Berst MJ, Dolan L, Bogdanowicz MM, Stevens MA, Chow S, Brandser EA. Effect of knowledge of chronologic age on the variability of pediatric bone age determined using the Greulich and Pyle standards.  AJR Am J Roentgenol. 2001;176:507-510
PubMed   |  Link to Article
Dunn PM, Levinson W. The lack of effect of clinical information on electrocardiographic diagnosis of acute myocardial infarction.  Arch Intern Med. 1990;150:1917-1919
PubMed   |  Link to Article
Hatala RA, Norman GR, Brooks LR. The effect of clinical history on physicians’ ECG interpretation skills.  Acad Med. 1996;71:(suppl 10)  S68-S70
PubMed   |  Link to Article
Hatala R, Norman GR, Brooks LR. Impact of a clinical scenario on accuracy of electrocardiogram interpretation.  J Gen Intern Med. 1999;14:126-129
PubMed   |  Link to Article
Leslie A, Jones AJ, Goddard PR. The influence of clinical information on the reporting of CT by radiologists.  Br J Radiol. 2000;73:1052-1055
PubMed
Paradise JE, Winter MR, Finkel MA, Berenson AB, Beiser AS. Influence of the history on physicians’ interpretations of girls’ genital findings.  Pediatrics. 1999;103:980-986
PubMed   |  Link to Article
Simons M, Parker JA, Donohoe KJ, Udelson JE, Gervino EV. The impact of clinical data on interpretation of thallium scintigrams.  J Nucl Cardiol. 1994;1:365-371
PubMed   |  Link to Article
Van Gijn J, Bonke B. Interpretation of plantar reflexes: biasing effect of other signs and symptoms.  J Neurol Neurosurg Psychiatry. 1977;40:787-789
PubMed   |  Link to Article
Vogel HP. Influence of additional information on interrater reliability in the neurologic examination.  Neurology. 1992;42:2076-2081
PubMed   |  Link to Article
Whited JD, Horner RD, Hall RP, Simel DL. The influence of history on interobserver agreement for diagnosing actinic keratoses and malignant skin lesions.  J Am Acad Dermatol. 1995;33:603-607
PubMed   |  Link to Article
Williams GW, Lesser RP, Silvers JB.  et al.  Clinical diagnoses and EEG interpretation.  Cleve Clin J Med. 1990;57:437-440
PubMed   |  Link to Article
Bonke B, Koudstaal PJ, Dijkstra G.  et al.  Detection of lacunar infarction in brain CT-scans: no evidence of bias from accompanying patient information.  Neuroradiology. 1989;31:170-173
PubMed   |  Link to Article
Bonke B, Kwakernaak A, Van Den Berg M, Koudstaal PJ. Bias in the assessment of computed-tomography scans for lacunar infarction.  Cerebrovasc Dis. 1992;2:107-110
Link to Article
Doubilet P, Herman PG. Interpretation of radiographs: effect of clinical history.  AJR Am J Roentgenol. 1981;137:1055-1058
PubMed   |  Link to Article
Eldevik OP, Dugstad G, Orrison WW, Haughton VM. The effect of clinical bias on the interpretation of myelography and spinal computed tomography.  Radiology. 1982;145:85-89
PubMed
Gjorup T, Saurbrey N, Hermann N. Clinical estimation of the duration of pregnancy in legal abortion–are doctors biased by their knowledge of the duration of amenorrhoea?  Methods Inf Med. 1984;23:96-98
PubMed
Potchen EJ, Gard JW, Lazar P, Lahaie P, Andary M. The effect of clinical history data on chest film interpretation: direction or distraction.  Invest Radiol. 1979;14:404
Ashworth CS, Fargason CA Jr, Fountain K, Cutter G, Centor R. Impact of patient history on residents’ evaluation on child sexual abuse.  Child Abuse Negl. 1995;19:943-951
PubMed   |  Link to Article
Babcook CJ, Norman GR, Coblentz CL. Effect of clinical history on the interpretation of chest radiographs in childhood bronchiolitis.  Invest Radiol. 1993;28:214-217
PubMed   |  Link to Article
Flach SD, Canaris GJ, Tape TG, Huntley KM, Wigton RS. The effects of proficiency and bias on residents’ interpretation of the microscopic urinalysis.  Med Decis Making. 2002;22:318-325
PubMed
Quekel LG, Goei R, Kessels AG, van Engelshoven JM. Detection of lung cancer on the chest radiograph: impact of previous films, clinical information, double reading, and dual reading.  J Clin Epidemiol. 2001;54:1146-1150
PubMed   |  Link to Article
Grum CM, Gruppen LD, Woolliscroft JO. The influence of vignettes on EKG interpretation by third-year students.  Acad Med. 1993;68:(suppl 10)  S61-S63
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD.  et al.  Tentative diagnoses facilitate the detection of diverse lesions in chest radiographs.  Invest Radiol. 1986;21:532-539
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD, Barloon TJ. Influence of clinical history upon detection of nodules and other lesions.  Invest Radiol. 1988;23:48-55
PubMed   |  Link to Article
Berbaum KS, el-Khoury GY, Franken EA Jr, Kathol M, Montgomery WJ, Hesson W. Impact of clinical history on fracture detection with radiography.  Radiology. 1988;168:507-511
PubMed
Berbaum KS, Franken EA Jr, el-Khoury GY. Impact of clinical history on radiographic detection of fractures: a comparison of radiologists and orthopedists.  AJR Am J Roentgenol. 1989;153:1221-1224
PubMed   |  Link to Article
Berbaum KS, Franken EA Jr, Dorfman DD, Lueben KR. Influence of clinical history on perception of abnormalities in pediatric radiographs.  Acad Radiol. 1994;1:217-223
PubMed   |  Link to Article
Cooperstein LA, Good BC, Eelkema EA.  et al.  The effect of clinical history on chest radiograph interpretations in a PACS environment.  Invest Radiol. 1990;25:670-674
PubMed   |  Link to Article
Good BC, Cooperstein LA, DeMarino GB.  et al.  Does knowledge of the clinical history affect the accuracy of chest radiograph interpretation?  AJR Am J Roentgenol. 1990;154:709-712
PubMed   |  Link to Article
Ehara S, Katsuragawa S. Influence of clinical information on the detection of wrist fractures in children.  Tohoku J Exp Med. 1999;189:147-153
PubMed   |  Link to Article
McNeil BJ, Hanley JA, Funkenstein HH, Wallman J. Paired receiver operating characteristic curves and the effect of history on radiographic interpretation: CT of the head as a case study.  Radiology. 1983;149:75-77
PubMed
Raab SS, Oweity T, Hughes JH.  et al.  Effect of clinical history on diagnostic accuracy in the cytologic interpretation of bronchial brush specimens.  Am J Clin Pathol. 2000;114:78-83
PubMed   |  Link to Article
Song KS, Song HH, Park SH.  et al.  Impact of clinical history on film interpretation.  Yonsei Med J. 1992;33:168-172
PubMed
Elmore JG, Wells CK, Howard DH, Feinstein AR. The impact of clinical history on mammographic interpretations.  JAMA. 1997;277:49-52
PubMed   |  Link to Article
Kinnunen J, Gothlin J, Totterman S.  et al.  The effects of clinical data on radiological diagnostic performance in middle-face trauma.  ROFO Fortschr Geb Rontgenstr Nuklearmed. 1987;147:143-146
PubMed   |  Link to Article
Rickett AB, Finlay DB, Jagger C. The importance of clinical details when reporting accident and emergency radiographs.  Injury. 1992;23:458-460
PubMed   |  Link to Article
Tudor GR, Finlay D, Taub N. An assessment of inter-observer agreement and accuracy when reporting plain radiographs.  Clin Radiol. 1997;52:235-238
PubMed   |  Link to Article
Bossuyt PM, Reitsma JB, Bruns DE.  et al.  Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative.  BMJ. 2003;326:41-44
PubMed   |  Link to Article
CME


You need to register in order to view this quiz.
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 57

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Topics
PubMed Articles