Unanticipated vaginal pathology after pelvic organ prolapse surgery: a retrospective cohort study and cost-effectiveness analysis

Technical Summary: Unanticipated Vaginal Pathology After Pelvic Organ Prolapse Surgery

Problem Statement
Pelvic organ prolapse (POP) affects approximately 10% of women, with over 300,000 corrective surgeries performed annually in the United States. Many of these procedures, such as colporrhaphy, vaginectomy, and colpocleisis, involve the excision of vaginal epithelium. While vaginal cancer and vaginal intraepithelial neoplasia (VaIN) are rare (0.6 and 0.2–2 cases per 100,000 women, respectively), routine pathologic assessment of excised vaginal tissue remains standard practice at many institutions. This practice is often driven by accreditation standards (e.g., The Joint Commission) and institutional policies rather than robust evidence of clinical utility. The variability in submission protocols across providers highlights a lack of data regarding the incidence of unanticipated malignant or premalignant findings in these specimens and the cost-effectiveness of routine screening.

Methodology
The study employed a two-pronged approach: a retrospective cohort analysis and a cost-effectiveness decision analytic model.

1. Retrospective Cohort Analysis:

   • Data Source: Pathology reports from vaginal epithelium specimens collected during POP surgery at Indiana University-affiliated hospitals between April 3, 2000, and June 15, 2015.
   • Inclusion/Exclusion: The study included 487 cases of benign POP repair performed by gynecologists. Cases were excluded if there was pre-surgical suspicion of malignancy, known cervical intraepithelial neoplasia, or if the specimen was non-vaginal/incomplete.
   • Classification: Specimens were categorized as benign, premalignant (dysplasia, atypia, or lichenoid reaction), or malignant (invasive disease). No slides were re-reviewed; classification relied on original reports.
   • Statistical Analysis: Frequencies and percentages were calculated with Wald 95% confidence intervals (CIs).

2. Cost-Effectiveness Model:

   • Structure: A decision analytic model was constructed using TreeAgePro 2023 software from a healthcare perspective, comparing "routine pathologic assessment" versus "no routine assessment" over a lifetime horizon (28.6 years).
   • Inputs: The model utilized incidence rates derived from the retrospective cohort. Costs were converted to 2023 USD, incorporating pathology fees (CPT codes 88305 and 88307), management costs for VaIN and vaginal cancer, and death costs.
   • Outcomes: Primary outcomes included total costs, quality-adjusted life years (QALYs), vaginal cancer cases, and cancer-related deaths.
   • Sensitivity Analysis: Univariate and probabilistic Monte Carlo simulations (300,000 iterations) were performed to test model robustness, assuming a willingness-to-pay threshold of $50,000.

Key Results

• Pathologic Findings: Out of 487 vaginal specimens, 483 (99.2%) were benign. Only 4 specimens (0.8%) were classified as premalignant: two cases of dysplasia, one case of atypia, and one case of lichenoid reaction. No malignant lesions were identified.
• Non-Vaginal Specimens: Among associated non-vaginal specimens, 4 cases (0.8% of the total cohort) showed premalignant findings (cervical atypia/dysplasia or endometrial hyperplasia), all of which had benign vaginal findings.
• Cost-Effectiveness: In a theoretical cohort of 300,000 patients:
  • Costs: The "no routine assessment" strategy cost approximately $18.76 million, compared to $126.80 million for routine assessment (a difference of ~$108 million).
  • Effectiveness: The "no routine assessment" strategy yielded 5,687,886 QALYs, slightly higher than the 5,683,419 QALYs for routine assessment.
  • Clinical Outcomes: The difference in vaginal cancer cases was minimal (305 vs. 300), and cancer-related deaths were nearly identical (94 vs. 93).
  • Dominance: Routine assessment was a dominated strategy (higher cost, lower effectiveness). Probabilistic sensitivity analysis confirmed that "no routine assessment" was the dominant strategy in 99.9% of simulations.

Key Contributions

• Incidence Data: This study provides the first published data quantifying the rate of unanticipated premalignant and malignant findings in vaginal epithelium specifically removed during POP surgery, finding a rate of 0.8% for premalignancy and 0% for malignancy.
• Economic Evaluation: It offers a comprehensive cost-effectiveness analysis demonstrating that routine pathologic evaluation of these specimens is not cost-effective, generating significant savings without compromising patient survival or cancer detection rates.
• Policy Implications: The findings challenge the necessity of universal submission protocols for vaginal specimens in POP surgery, suggesting that institutional policies could be revised to exempt these specimens from routine review.

Significance and Claims
The authors conclude that unanticipated malignant or premalignant pathologic findings in vaginal epithelium specimens at the time of surgical POP repair are rare. Consequently, routine pathologic evaluation of these specimens is neither clinically useful nor cost-effective. The study suggests that the marginal benefit of identifying a few additional dysplastic cases does not justify the substantial financial burden and the potential for unnecessary downstream interventions (such as vulvoscopy, laser ablation, and excision) that arise from false positives or the detection of lesions that might regress. The authors posit that their results should inform the development of internal institutional protocols and contribute to the standardization of specimen assessment guidelines, potentially reducing healthcare costs without adverse clinical outcomes.

Limitations
The authors acknowledge the retrospective nature of the study, the lack of complete patient history (e.g., Pap smear history), and the reliance on data from a single institution. Furthermore, the model required extrapolating transition probabilities for VaIN from cervical and vulvar cancer literature due to a scarcity of specific data on vaginal dysplasia progression. Despite these limitations, the large sample size spanning 15 years and the use of rigorous decision modeling support the study's conclusions regarding the low yield and high cost of routine assessment.