Using sensitivity analysis to evaluate the fracture benefit of parathyroidectomy

Area of Science:

• Endocrinology
• Surgical Oncology
• Bone Metabolism

Background:

• Observational studies suggest parathyroidectomy reduces fractures compared to nonoperative management for primary hyperparathyroidism.
• Concerns exist regarding unmeasured confounding in these observational studies.
• The Robustness of Inferences to Replacement (RIR) method assesses sensitivity to unmeasured confounding.

Purpose of the Study:

• To evaluate the robustness of observational data linking parathyroidectomy to reduced fracture risk.
• To assess the strength of evidence for parathyroidectomy versus nonoperative management in primary hyperparparathyroidism.

Main Methods:

• A systematic literature review identified studies comparing fracture rates after parathyroidectomy and nonoperative management.
• The Robustness of Inferences to Replacement (RIR) was calculated for each study.
• The Fraction of Effect Above the Statistical Significance Threshold (FEAST) was also computed.

Main Results:

• Six retrospective cohort studies (217,343 patients) were analyzed.
• Parathyroidectomy showed a 31%-59% reduction in any fracture and a 16%-62% reduction in hip fractures.
• Up to 90% of the observed effect would need to be due to unmeasured confounding to overturn the results.

Conclusions:

• Observational evidence strongly suggests parathyroidectomy provides a true benefit in reducing fractures.
• The findings indicate that parathyroidectomy is likely robust to unmeasured confounding.
• The RIR method effectively informs clinical decision-making regarding observational study validity.