(Radiology. 1999;211:529-533.)
© RSNA, 1999
Medical Radiologic Technologist Review: Effects on a Population-based Breast Cancer Screening Program1
Jon M. Tonita, BSc, MSc,
Joanne P. Hillis, MD, FRCPC and
Chong-Ha Lim, MD, FRCPC
1 From the Saskatchewan Cancer Agency, Epidemiology, Prevention and Screening, 4101 Dewdney Ave, Regina, Saskatchewan, Canada S4T 7T1. Received March 26, 1998; revision requested May 12; final revision received August 17; accepted October 26. Address reprint requests to J.M.T.
 |
Abstract
|
|---|
PURPOSE: To evaluate the effects of medical radiologic technologist review of mammograms in a population-based breast cancer screening program.
MATERIALS AND METHODS: A technologist review pilot project was incorporated into the Regina, Saskatchewan, Canada, reading center. Technologists received special training in mammographic interpretation. They reviewed all 27,863 mammograms obtained at the center from July 1995 to September 1996 that were reviewed by a radiologist and selected cases for second blind reading by another radiologist. When the two radiologists' readings were in agreement, the report was sent. When the readings differed, a third opinion was obtained from the program's consulting radiologist. Changes in the number of mammograms interpreted as abnormal and the number of cancers detected were assessed.
RESULTS: The technologist review was responsible for the detection of nine cancers missed at the first radiologist's interpretation. Technologists were very discriminating; only 391 cases (1.4%) were sent for double reading. The positive predictive value of screening did not change significantly (7.5% without review, 8.1% with review; P > .20).
CONCLUSION: A substantial number of cancers were found with the technologist review. The number of mammograms interpreted as abnormal was reduced slightly. The technologist review proved to be a cost-effective alternative to double reading by two radiologists.
Index terms: Breast neoplasms, diagnosis, 00.11, 00.30 • Cancer screening, 00.11, 00.30 • Diagnostic radiology, observer performance
|
Introduction
|
|---|
Our population-based breast cancer screening program, which was established in 1990, occupies two stationary centers, five satellite centers, and one mobile unit. The program invites women 50–69 years of age to biennial screening, with an annual invitation for those with a first-degree relative with a history of breast cancer. Women then book their own appointments. Two-view mammograms are interpreted once by a certified radiologist at a stationary center.
Double reading of mammograms, however, has been shown to increase the number of cancers detected (1,2). The additional cost of double reading by radiologists was considered prohibitive because our government-funded program operates on a fixed budget. We realized, however, that our medical radiologic technologists—because of their considerable experience, dedication, and specific training—were becoming adept at recognizing mammographic abnormalities. Research has shown that radiographers can be trained to identify mammographic abnormalities and that they can be used as second readers in screening mammography (3,4). We therefore proposed a reading format that involves a technologist review of screening mammograms. The purpose of this study was to determine the effects of a technologist review program on the number of cancers detected and the number of mammograms interpreted as abnormal.
|
MATERIALS AND METHODS
|
|---|
Three experienced medical radiologic technologists were involved in this pilot project. These technologists were experienced in both screening and diagnostic mammography and had received additional training in mammographic interpretation. Together they reviewed approximately 80% of the screening mammograms obtained in our program. The technologists reviewed 27,863 screening mammograms at our main stationary center in Regina, Saskatchewan, Canada, from July 1995 to September 1996. They organized and recorded our quality assurance sessions as well.
Each mammogram was reviewed by one of three technologists prior to interpretation by one of five radiologists (including J.P.H. and C.H.L.). The radiologist was blinded to this result. Technologists compared their interpretations with the radiology report. When the interpretations differed, the case was remounted on a light box for a second reading by another radiologist. He or she was blinded to the first interpretation. If both radiologists were in agreement, then the report was sent out. When there was a discrepancy, the case was submitted to the program's consulting radiologist (C.H.L.) for a third opinion. The consulting radiologist was aware that he or she was providing a third interpretation but was blinded to the previous interpretations.
The initial and final interpretations were recorded for all cases selected for double reading. All abnormal cases in our screening program were followed up by obtaining the diagnostic assessments and surgical and pathology reports. True- and false-positive rates then were determined. All normal cases were checked against our population-based cancer registry every 2 weeks to check for interval cancers, or cancers detected in women who had a normal result at screening mammography but in whom breast cancer was later diagnosed prior to their next scheduled date for screening mammography. This allowed calculation of true- and false-negative screening rates.
A 95% CI for the proportion of cancers found as a result of the technologist review was used to determine significance, assuming a Poisson distribution for the cancers detected (5). The change in the number of abnormalities because of the review was noted. The influence of the technologist review on the number of cancers detected and the rate of abnormal interpretations combined was determined by calculating the difference between the positive predictive value of screening with and the positive predictive value of screening without the technologist review process. Significance for the difference in proportions was determined by using the two-tailed standardized normal deviate (6), and the 95% CI for the difference in proportions was calculated (5).
|
RESULTS
|
|---|
Figure 1 shows the results of the technologist review from July 1995 to September 1996. The technologists selected 192 of 25,779 mammograms originally interpreted as normal and 199 of 2,084 mammograms originally interpreted as abnormal for a second reading. The technologists were very discriminating—only 391 of the 27,863 mammograms (1.4%) were remounted.
View larger version (28K):
[in this window]
[in a new window]
|
Figure 1. Flowchart demonstrates the effect of technologist review on mammographic results. Rad = radiologist, Tech = technologist.
|
|
Of the 391 cases submitted for a second reading, 95 cases were interpreted as abnormal (originally interpreted as normal) and 161 cases were interpreted as normal (originally interpreted as abnormal) at the second reading. These 256 cases (95 + 161 cases) were submitted to the consulting radiologist for a third reading. There were 135 cases with interpretations that did not change from the first to the second reading.
As a result of the third reading, 86 of the original 192 normal mammograms picked for review were interpreted as abnormal and 144 of the original 199 abnormal mammograms chosen for review were interpreted as normal. Overall, the number of mammograms interpreted as abnormal was reduced by 58 cases (144 - 86 cases). To date, no interval cancers have been detected in the group with interpretations changed from abnormal to normal. As well, 51 of the 144 women with interpretations changed from abnormal to normal have returned for screening mammography, and 48 had mammograms that were subsequently interpreted as normal. The three women with abnormal mammograms were determined, after diagnostic follow-up, not to have cancer.
In the Table, the outcomes for the screening program with the technologist review are compared with the outcomes that would have occurred without the review. The number of mammograms interpreted as abnormal was reduced by 2.8%, and the number of cancers detected was increased by 5.8% (Table). The nine additional cancers detected represents 5.5% of all cancers detected during the study (95% CI: 2.0%, 8.9%). The combined effect of fewer abnormal interpretations and more cancer detections was a slight nonsignificant increase of 8.7% in the positive predictive value of the screening program (P > .40; 95% CI: -13.1%, 30.7%).
Figures 2–4 show examples of cancers detected because of the technologist review. Both views in Figure 2 from a first screening examination show a large spiculated mass in the left breast that was missed at the first reading but that was detected by the technologist. This was an invasive lobular carcinoma. Figure 3, from a second screening examination, shows a subtle spiculated lesion in the inferolateral part of the left breast. This is also visible on both views and was a lobular carcinoma. Figure 4b, an oblique view from a first screening examination, shows microcalcifications and distortion in the left upper outer quadrant; this was an invasive ductal carcinoma. This cancer is not visible on the craniocaudal view (Fig 4a).
View larger version (81K):
[in this window]
[in a new window]
|
Figure 2a. (a) Craniocaudal and (b) oblique mammograms show a large spiculated mass (arrow) in a 65-year-old woman with invasive lobular carcinoma.
|
|
View larger version (84K):
[in this window]
[in a new window]
|
Figure 2b. (a) Craniocaudal and (b) oblique mammograms show a large spiculated mass (arrow) in a 65-year-old woman with invasive lobular carcinoma.
|
|
View larger version (122K):
[in this window]
[in a new window]
|
Figure 3a. (a) Craniocaudal and (b) oblique mammograms show a subtle spiculated mass (arrow) in a 51-year-old woman with lobular carcinoma.
|
|
View larger version (123K):
[in this window]
[in a new window]
|
Figure 3b. (a) Craniocaudal and (b) oblique mammograms show a subtle spiculated mass (arrow) in a 51-year-old woman with lobular carcinoma.
|
|
View larger version (78K):
[in this window]
[in a new window]
|
Figure 4a. (a) Craniocaudal and (b) oblique mammograms in a 51-year-old woman with invasive ductal carcinoma. Microcalcifications and distortion (arrow in b) are visible only in b.
|
|
View larger version (80K):
[in this window]
[in a new window]
|
Figure 4b. (a) Craniocaudal and (b) oblique mammograms in a 51-year-old woman with invasive ductal carcinoma. Microcalcifications and distortion (arrow in b) are visible only in b.
|
|
|
DISCUSSION
|
|---|
Double reading of screening mammograms by radiologists has been shown to increase the number of cancers detected and reduce the number of women called back for further assessment (1,2,7,8). Our screening program operates on a fixed budget; currently, radiology services represent 11% of total expenditures. If double reading were to take place, the program would be forced to reduce eligibility criteria. We have found that in our setting, a technologist review can increase our cancer detection rate without jeopardizing the universality of the program or adversely affecting the rate of abnormal interpretations.
Our system of technologist review may not be applicable to every program. A number of criteria are essential to a successful review process. A close working relationship between radiologists and technologists is paramount. Radiologists must have a positive attitude toward technologists' input. Limiting the number of radiologists (in our case, five) and technologists (in our case, three) involved greatly enhances cooperation. The radiologists should be accessible at a time of day when the technologists are available for discussion so that cases can be reviewed and questions answered.
Only experienced technologists should be involved in the review process. This experience should include the review of large numbers of mammograms, along with diagnostic follow-up when mammograms are abnormal. Our technologists obtain this experience by preparing and attending our monthly quality assurance sessions. At these meetings, we also review any false-negative cases we have become aware of by checking with our cancer registry. These sessions help foster good communication between radiologists and technologists and improve overall mammographic skills. Our technologists also receive additional training in mammographic interpretation by attending courses in mammography.
The radiologists involved with our screening program have Canadian certification in radiology. In 1995, radiologists who participated in this project each read an average of 4,300 screening mammograms from the program, in addition to the diagnostic evaluations they performed at hospitals and in their group practices. In total, these radiologists have read approximately 78,000 screening mammograms from our program since 1990. Their combined false-negative rate for the first 5 years was three cases per 10,000 mammograms, and their combined sensitivity was 95.3% (488 of 512 cancers) (Screening Program for Breast Cancer, unpublished data, 1995).
Our results are favorable in that our cancer detection rate increased without an increase in the rate at which mammograms were interpreted as abnormal. Other positive outcomes are less quantifiable but real. Having the technologists involved in interpretation made them more aware of the importance of good positioning and optimal mammographic quality. This also made them more interested in obtaining complete follow-up of abnormal cases and were very helpful in this regard. Radiologists became more aware of their reporting record and certainly strived to keep their "technologist-assisted" cases to a minimum by optimizing their interpretation of mammograms and detection of cancer.
Continuous follow-up of reviewed cases is essential. All cases in which results are changed because of the review must be followed up. We must know how many additional cancers were detected by changing the interpretation from normal to abnormal. If few additional cancers are detected, then such a program may not be justified. Follow-up of cases in which the interpretation changed from abnormal to normal, however, is equally important. We do not want to miss detectable cancers. To date, no false-negative cancers have been diagnosed in review cases in which the interpretations changed to normal. We have therefore decreased the number of false-positive cases. This in turn has resulted in fewer diagnostic follow-up evaluations and has reduced the anxiety of patients. However, given the medicolegal implications of false-negative cases, we are very careful not to overly reduce our false-positive rate.
There are some drawbacks to such a program. There is an increase in the time required for the technologist to review mammograms, remount mammograms on the light box, and document findings. At the stationary center, one technologist will dedicate 2–3 h/d for these tasks, which are now part of the technologists' regular duties. This increases the cost of the screening program slightly because more technologist time is required. This increase, however, is substantially less than the cost of double reading by radiologists. Some additional radiologist time is also required for the second and third readings. The technologists were very discriminating, however, in remounting only 1.4% of all mammograms. Currently, our radiologists do not receive payment for these readings, but this is under review. The review process also delays our mailing of final reports of cases reviewed. Family physicians have indicated, however, that quality screening that involves double reading is preferred over single reading, even if there is a slight delay involved (9). Most reports are mailed out within a week from the mammography date, which we believe is reasonable.
These results represent the experience from a population-based screening situation, not from testing or training conditions. In our program, the technologist review has been a positive experience. The review is now ongoing and has expanded to include our second stationary center. The technologist review is now an integral part of our quality assurance program. The review has helped us achieve our goal of maximum cancer detection while maintaining a reasonable rate of interpretation of mammograms as abnormal.
|
Acknowledgments
|
|---|
The authors thank the screening program radiographers, Nancy Baumann, RTR, and Barbara Layton, RTR, AC, for recording and maintaining the records necessary for this project. We also thank Mary Johnson for her assistance in preparing the manuscript.
|
Footnotes
|
|---|
Author contributions: Guarantors of integrity of entire study, J.P.H., C.H.L., J.M.T.; study concepts, C.H.L.; study design, J.P.H., C.H.L., J.M.T.; definition of intellectual content, J.P.H., J.M.T.; literature research, J.P.H., J.M.T.; data acquisition and analysis, J.M.T.; statistical analysis, J.M.T.; manuscript preparation, J.P.H., J.M.T.; manuscript editing, J.P.H., C.H.L., J.M.T.; manuscript review, C.H.L.
|
References
|
|---|
-
Thurfjell EL, Lernevall KA, Taube AA. Benefit of independent double reading in a population-based mammography screening program. Radiology 1994; 191:241-244.[Abstract/Free Full Text]
-
Warren RM, Duffy SW. Comparison of single reading with double reading of mammograms, and change in effectiveness with experience. Br J Radiol 1995; 68:958-962.[Abstract]
-
Bassett LW, Hollatz-Brown AJ, Bastani R, Pearce JG, Hirji K, Chen L. Effects of a program to train radiologic technologists to identify abnormalities on mammograms. Radiology 1995; 194:189-192.[Abstract/Free Full Text]
-
Pauli R, Hammond S, Cooke J, Ansell J. Radiographers as film readers in screening mammography: an assessment of competence under test and screening conditions. Br J Radiol 1996; 69:10-14.[Abstract]
-
Gardner MJ, Altman DG. Calculating confidence intervals for proportions and their differences. In: Gardner MJ, Altman DG, eds. Statistics with confidence. London, England: British Medical Journal, 1993; 28-30.
-
Armitage P, Berry G. Statistical inference. In: Armitage P, Berry G, eds. Statistical methods in medical research. 3rd ed. Oxford, England: Blackwell Science, 1994; 128-129.
-
Brown J, Bryan S, Warren R. Mammography screening: an incremental cost effectiveness analysis of double reading versus single reading of mammograms. BMJ 1996; 312:809-812.[Abstract/Free Full Text]
-
Anttinen I, Pamilo M, Soiva M, Roiha M. Double reading of mammography screening films: one radiologist or two?. Clin Radiol 1993; 48:414-421.[Medline]
-
Slanetz PJ, Moore RH, Hulka CA, et al. Physicians' opinions on the delivery of mammographic screening services: immediate interpretation versus double reading. AJR 1996; 167:377-379.[Abstract/Free Full Text]
TOP
Abstract
Introduction
