To Register      SMDM Homepage

Tuesday, 19 October 2004

This presentation is part of: Poster Session - Clinical Strategies; Judgment and Decison Making

NATURAL HISTORY OF CERVICAL INTRAEPITHELIAL NEOPLASIA: A META-ANALYSIS

Scott B. Cantor, PhD1, E. Neely Atkinson1, Marylou Cardenas-Turanzas, MD, MPH1, J. L. Benedet, MD2, Michele Follen, MD, PhD3, and Calum MacAulay, PhD4. (1) The University of Texas M. D. Anderson Cancer Center, Biostatistics & Applied Mathematics, Houston, TX, (2) University of British Columbia and BC Cancer Agency, Gynecologic Oncology, Vancouver, BC, Canada, (3) The University of Texas M. D. Anderson Cancer Center, Biomedical Engineering Center, Houston, TX, (4) British Columbia Cancer Research Center, Cancer Imaging, Vancouver, BC, Canada

Purpose: The objective of this study was to determine the probabilities of transition of stages in the natural history of cervical cancer by conducting a formal meta-analysis of published studies of the topic.

Methods: We identified health states of interest in the natural history of cervical pre-cancer, identified all possible papers from MEDLINE (years 1966-2002) that met selection criteria, developed relevance and acceptability criteria for inclusion, then thoroughly reviewed the selected studies. Four transitions were investigated in detail: (1) high-grade squamous intraepithelial lesions (HGSIL) to cancer, (2) low-grade squamous intraepithelial lesions (LGSIL) to HGSIL, (3) HGSIL to LGSIL, and (4) LGSIL to normal. We converted data to determine 6-month transition probabilities, as this is the time frame for follow-up after an abnormal Papanicolaou smear. To determine the transition probability data we used a random effects model that assumed the parameters were drawn from a gamma distribution.

Results: 28 studies were found that met acceptability and relevance criteria; not all studies provided data for all 4 health-state transitions. Our final analysis included 9 studies for HGSIL to cancer (follow-up 0-336 months), 12 studies for LGSIL to HGSIL (follow-up 0-76 months), 3 studies for HGSIL to LGSIL (follow-up 0-132 months), and 12 studies for LGSIL to normal (follow-up 0-77 months). The homogeneity test failed for each transition under investigation, i.e. the studies were not homogeneous. The 6-month mean predictive transition probability (95% confidence intervals with “prediction interval” in parentheses) for HGSIL to cancer was 0.0037 (0.00004, 0.03386); for LGSIL to HGSIL was 0.0362 (0.00055, 0.23220); for HGSIL to LGSIL was 0.0282 (0.00027, 0.35782); and for LGSIL to normal was 0.0740 (0.00119, 0.42672).

Conclusion: The transition probabilities between cervical cancer health states for 6-month intervals are small. However, there was significant variation in the probabilities for the various health-state transitions. The results of this study, when used in a decision-analytic model for cervical cancer screening, will have to undergo extensive sensitivity analysis.


See more of Poster Session - Clinical Strategies; Judgment and Decison Making
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)