A Comparative Effectiveness Analysis of Alternative Strategies to Assess Hypothalamic-Pituitary-Adrenal Axis Function After Microsurgical Resection of Pituitary Tumors

Abstract

Abstract BACKGROUND: Observational methods can be used in conjunction with the comparative effectiveness research (CER) paradigm to inform decisions between alternative patient management strategies in real-world clinical settings. OBJECTIVE: To present a brief review of current observational research regarding 3 strategies for predicting normal hypothalamic-pituitary-adrenal (HPA) axis function after surgical resection of pituitary tumors and to apply the CER model to compare these management alternatives. METHODS: We designed and conducted 2 prospective observational studies involving 183 patients undergoing microsurgical resection for pituitary tumors. These investigations yielded a comprehensive database comprising longitudinal data from multiple clinical domains. We investigated 3 potential strategies to predict normal postoperative HPA axis function in this cohort, including preoperative adrenocorticotrophic hormone stimulation testing and measurement of serum cortisol levels immediately after surgery or on the first postoperative day. We performed a focused comparative effectiveness review to help inform the decision between the 3 potential clinical management strategies. RESULTS: This investigation illustrates the use of observational research methods in conjunction with CER methodology as one means of informing clinical management decisions. Of the 3 strategies for assessing postoperative HPA axis function studied, preoperative and immediate postoperative adrenocorticotrophic hormone stimulation testing had the highest sensitivity, accuracy, and positive predictive value for normal HPA axis function postoperatively. The preoperative strategy was also the most cost-effective approach (12% reduction vs benchmark). CONCLUSION: The decision to use any of the 3 strategies outlined requires attention to a specific patient's clinical situation, but this decision may be aided by the results of this CER analysis.