INDIVIDUAL MEMORY CHANGE AFTER ANTERIOR TEMPORAL LOBECTOMY - A BASE-RATE ANALYSIS USING REGRESSION-BASED OUTCOME METHODOLOGY

Purpose: To characterize patterns of base rate change on measures of v erbal and visual memory after anterior temporal lobectomy (ATL) using a newly developed regression-based outcome methodology that accounts f or effects of practice and regression towards the mean, and to comment on the predictive utility of baseline memory measures on postoperativ e memory outcome. Methods: Memory change was operationalized using reg ression-based change norms in a group of left (n = 53) and right (n = 48) ATL patients. All patients were administered tests of episodic ver bal (prose recall, list learning) and visual (figure reproduction) mem ory, and semantic memory before and after ATL. Results: ATL patients d isplayed a wide range of memory outcome across verbal and visual memor y domains. Significant performance declines were noted for 25-50% of l eft ATL patients on verbal semantic and episodic memory tasks, while o ne-third of right ATL patients displayed significant declines in immed iate and delayed episodic prose recall. Significant performance improv ement was noted in an additional one-third of right ATL patients on de layed prose recall. Base rate change was similar between the two ATL g roups across immediate and delayed visual memory. Approximately one-fo urth of all patients displayed clinically meaningful losses on the vis ual memory task following surgery. Robust relationships between preope rative memory measures and nonstandardized change scores were attenuat ed or reversed using standardized memory outcome techniques. Conclusio ns: Our results demonstrated substantial group variability in memory o utcome for ATL patients. These results extend previous research by inc orporating known effects of practice and regression to the mean when a ddressing meaningful neuropsychological change following epilepsy surg ery. Our findings also suggest that future neuropsychological outcome studies should take steps towards controlling for regression-to-the-me an before drawing predictive conclusions.