Accuracy and inter-observer agreement when reporting mammograms

A. Elinah, M. Nafikudin, Zahiah Mohamed, F. A. Meah, N. A. Sharifah, Abdul Aziz Jemain, Ahmad Mahir Razali, Wan Rosmanira Ismail, Annuar Zulfiqar

Research output: Contribution to journalArticle

Abstract

Aim: To compare the diagnostic accuracy of mammographic reports when interpreted by two radiologists. To determine the inter-observer agreement between the two radiologists. Method: Two radiologists were requested to report 30 sets of mammograms independently. Nineteen (63%) of the patients were asymptomatic patients who had screening mammography before commencing hormone replacement therapy, 7 (23%) had breast complaints and 4 (13%) were post-mastectomy patients. The radiologists were coded as R1 and R2 respectively. Sixteen lesions had proven histological diagnosis. Each observer was asked to give an impression of the mammogram findings. There were 5 possible responses, corresponding to an increasing likelihood of malignancy. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value and false negative rate were assessed by simple proportional indices. Inter-observer agreement was assessed by Kappa statistics. Results: R1 and R2 had the same diagnostic accuracy of 87.5%. Sensitivity was 50% for R1 and 75% for R2 (mean 62.5%). Specificity was 100% for R1 and 91.7% for R2 (mean 95.9%). Positive predictive value was 100% for R1 and 75% for R2 (mean 87.5%). Negative predictive value was 85.7% for R1 and 91.7% for R2 (mean 88.7%). The false negative rate was 12.5% for R1 and 6.3% for R2. The Kappa value for inter-observer agreement was 0.167 (p = 0.008) with 95% confidence intervals of 0.024-0.3. This Kappa value indicated poor strength of agreement between the two radiologists. Conclusion: The mammography reports of both radiologists had a good diagnostic accuracy of 87.5%. The false negative rates were within the acceptable values of 5-15%. The poor inter-observer agreement between the 2 radiologists was because there were 5 variables allowed in the structured reporting form.

Original languageEnglish
Pages (from-to)67-74
Number of pages8
JournalAsian Oceanian Journal of Radiology
Volume7
Issue number2
Publication statusPublished - 2002

Fingerprint

Mammography
Mastectomy
Hormone Replacement Therapy
Radiologists
Breast
Confidence Intervals
Sensitivity and Specificity
Neoplasms

Keywords

  • Accuracy
  • Inter-observer agreement
  • Mammography

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Elinah, A., Nafikudin, M., Mohamed, Z., Meah, F. A., Sharifah, N. A., Jemain, A. A., ... Zulfiqar, A. (2002). Accuracy and inter-observer agreement when reporting mammograms. Asian Oceanian Journal of Radiology, 7(2), 67-74.

Accuracy and inter-observer agreement when reporting mammograms. / Elinah, A.; Nafikudin, M.; Mohamed, Zahiah; Meah, F. A.; Sharifah, N. A.; Jemain, Abdul Aziz; Razali, Ahmad Mahir; Ismail, Wan Rosmanira; Zulfiqar, Annuar.

In: Asian Oceanian Journal of Radiology, Vol. 7, No. 2, 2002, p. 67-74.

Research output: Contribution to journalArticle

Elinah, A, Nafikudin, M, Mohamed, Z, Meah, FA, Sharifah, NA, Jemain, AA, Razali, AM, Ismail, WR & Zulfiqar, A 2002, 'Accuracy and inter-observer agreement when reporting mammograms', Asian Oceanian Journal of Radiology, vol. 7, no. 2, pp. 67-74.
@article{857408b1f11449898b52b8281fd9133f,
title = "Accuracy and inter-observer agreement when reporting mammograms",
abstract = "Aim: To compare the diagnostic accuracy of mammographic reports when interpreted by two radiologists. To determine the inter-observer agreement between the two radiologists. Method: Two radiologists were requested to report 30 sets of mammograms independently. Nineteen (63{\%}) of the patients were asymptomatic patients who had screening mammography before commencing hormone replacement therapy, 7 (23{\%}) had breast complaints and 4 (13{\%}) were post-mastectomy patients. The radiologists were coded as R1 and R2 respectively. Sixteen lesions had proven histological diagnosis. Each observer was asked to give an impression of the mammogram findings. There were 5 possible responses, corresponding to an increasing likelihood of malignancy. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value and false negative rate were assessed by simple proportional indices. Inter-observer agreement was assessed by Kappa statistics. Results: R1 and R2 had the same diagnostic accuracy of 87.5{\%}. Sensitivity was 50{\%} for R1 and 75{\%} for R2 (mean 62.5{\%}). Specificity was 100{\%} for R1 and 91.7{\%} for R2 (mean 95.9{\%}). Positive predictive value was 100{\%} for R1 and 75{\%} for R2 (mean 87.5{\%}). Negative predictive value was 85.7{\%} for R1 and 91.7{\%} for R2 (mean 88.7{\%}). The false negative rate was 12.5{\%} for R1 and 6.3{\%} for R2. The Kappa value for inter-observer agreement was 0.167 (p = 0.008) with 95{\%} confidence intervals of 0.024-0.3. This Kappa value indicated poor strength of agreement between the two radiologists. Conclusion: The mammography reports of both radiologists had a good diagnostic accuracy of 87.5{\%}. The false negative rates were within the acceptable values of 5-15{\%}. The poor inter-observer agreement between the 2 radiologists was because there were 5 variables allowed in the structured reporting form.",
keywords = "Accuracy, Inter-observer agreement, Mammography",
author = "A. Elinah and M. Nafikudin and Zahiah Mohamed and Meah, {F. A.} and Sharifah, {N. A.} and Jemain, {Abdul Aziz} and Razali, {Ahmad Mahir} and Ismail, {Wan Rosmanira} and Annuar Zulfiqar",
year = "2002",
language = "English",
volume = "7",
pages = "67--74",
journal = "Asian Oceanian Journal of Radiology",
issn = "0972-2688",
publisher = "Asian Oceanian Society of Radiology",
number = "2",

}

TY - JOUR

T1 - Accuracy and inter-observer agreement when reporting mammograms

AU - Elinah, A.

AU - Nafikudin, M.

AU - Mohamed, Zahiah

AU - Meah, F. A.

AU - Sharifah, N. A.

AU - Jemain, Abdul Aziz

AU - Razali, Ahmad Mahir

AU - Ismail, Wan Rosmanira

AU - Zulfiqar, Annuar

PY - 2002

Y1 - 2002

N2 - Aim: To compare the diagnostic accuracy of mammographic reports when interpreted by two radiologists. To determine the inter-observer agreement between the two radiologists. Method: Two radiologists were requested to report 30 sets of mammograms independently. Nineteen (63%) of the patients were asymptomatic patients who had screening mammography before commencing hormone replacement therapy, 7 (23%) had breast complaints and 4 (13%) were post-mastectomy patients. The radiologists were coded as R1 and R2 respectively. Sixteen lesions had proven histological diagnosis. Each observer was asked to give an impression of the mammogram findings. There were 5 possible responses, corresponding to an increasing likelihood of malignancy. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value and false negative rate were assessed by simple proportional indices. Inter-observer agreement was assessed by Kappa statistics. Results: R1 and R2 had the same diagnostic accuracy of 87.5%. Sensitivity was 50% for R1 and 75% for R2 (mean 62.5%). Specificity was 100% for R1 and 91.7% for R2 (mean 95.9%). Positive predictive value was 100% for R1 and 75% for R2 (mean 87.5%). Negative predictive value was 85.7% for R1 and 91.7% for R2 (mean 88.7%). The false negative rate was 12.5% for R1 and 6.3% for R2. The Kappa value for inter-observer agreement was 0.167 (p = 0.008) with 95% confidence intervals of 0.024-0.3. This Kappa value indicated poor strength of agreement between the two radiologists. Conclusion: The mammography reports of both radiologists had a good diagnostic accuracy of 87.5%. The false negative rates were within the acceptable values of 5-15%. The poor inter-observer agreement between the 2 radiologists was because there were 5 variables allowed in the structured reporting form.

AB - Aim: To compare the diagnostic accuracy of mammographic reports when interpreted by two radiologists. To determine the inter-observer agreement between the two radiologists. Method: Two radiologists were requested to report 30 sets of mammograms independently. Nineteen (63%) of the patients were asymptomatic patients who had screening mammography before commencing hormone replacement therapy, 7 (23%) had breast complaints and 4 (13%) were post-mastectomy patients. The radiologists were coded as R1 and R2 respectively. Sixteen lesions had proven histological diagnosis. Each observer was asked to give an impression of the mammogram findings. There were 5 possible responses, corresponding to an increasing likelihood of malignancy. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value and false negative rate were assessed by simple proportional indices. Inter-observer agreement was assessed by Kappa statistics. Results: R1 and R2 had the same diagnostic accuracy of 87.5%. Sensitivity was 50% for R1 and 75% for R2 (mean 62.5%). Specificity was 100% for R1 and 91.7% for R2 (mean 95.9%). Positive predictive value was 100% for R1 and 75% for R2 (mean 87.5%). Negative predictive value was 85.7% for R1 and 91.7% for R2 (mean 88.7%). The false negative rate was 12.5% for R1 and 6.3% for R2. The Kappa value for inter-observer agreement was 0.167 (p = 0.008) with 95% confidence intervals of 0.024-0.3. This Kappa value indicated poor strength of agreement between the two radiologists. Conclusion: The mammography reports of both radiologists had a good diagnostic accuracy of 87.5%. The false negative rates were within the acceptable values of 5-15%. The poor inter-observer agreement between the 2 radiologists was because there were 5 variables allowed in the structured reporting form.

KW - Accuracy

KW - Inter-observer agreement

KW - Mammography

UR - http://www.scopus.com/inward/record.url?scp=0036302099&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036302099&partnerID=8YFLogxK

M3 - Article

VL - 7

SP - 67

EP - 74

JO - Asian Oceanian Journal of Radiology

JF - Asian Oceanian Journal of Radiology

SN - 0972-2688

IS - 2

ER -