BACKGROUND: While lagging subject enrollment in longitudinal clinical trials is a complex problem, the best recruitment strategy has not been established. Cumulative summation (CuSum) is a statistical process control procedure often applied in quality improvement efforts to detect trend shifts in highly variable serial data. OBJECTIVES: To describe changes in efforts to increase referrals and enrollment in a longitudinal quality-of-life breast cancer study, determine effects of changes in referral strategies on enrollment using a novel application of CuSum, and discuss implications of CuSum as a tool for prospectively managing the subject recruitment process. METHOD: Ten referrals and eight enrollments per month for a total of 31 months were estimated to meet study subject accrual requirements in the clinical trial. The estimates were used as standards in performing CuSum calculations. CuSum was applied to monthly referral and enrollment data and trend graphs were generated. Alterations in recruitment tactics and strategies were evaluated as to whether changes in trend occasioned such alterations. Unplanned changes in trend were noted. RESULTS: While monthly data were highly variable, an average of 8.42 referrals and 5.92 enrollments were realized during Months 1-12. Based on these figures, projected accrual for 31 months would have enrolled only 184 subjects, 66 subjects short of target. CuSum illustrated this trend. Subsequent shifts in enrollment trends were shown with improvements in referral. DISCUSSION: Indications for use of CuSum include (a) earlier detection of enrollment trend shifts, and (b) earlier discrimination between effective and ineffective recruitment. Thus, CuSum has implications for both evaluating the effects of planned and unplanned process changes and for managing the recruitment process.
BACKGROUND: While lagging subject enrollment in longitudinal clinical trials is a complex problem, the best recruitment strategy has not been established. Cumulative summation (CuSum) is a statistical process control procedure often applied in quality improvement efforts to detect trend shifts in highly variable serial data. OBJECTIVES: To describe changes in efforts to increase referrals and enrollment in a longitudinal quality-of-life breast cancer study, determine effects of changes in referral strategies on enrollment using a novel application of CuSum, and discuss implications of CuSum as a tool for prospectively managing the subject recruitment process. METHOD: Ten referrals and eight enrollments per month for a total of 31 months were estimated to meet study subject accrual requirements in the clinical trial. The estimates were used as standards in performing CuSum calculations. CuSum was applied to monthly referral and enrollment data and trend graphs were generated. Alterations in recruitment tactics and strategies were evaluated as to whether changes in trend occasioned such alterations. Unplanned changes in trend were noted. RESULTS: While monthly data were highly variable, an average of 8.42 referrals and 5.92 enrollments were realized during Months 1-12. Based on these figures, projected accrual for 31 months would have enrolled only 184 subjects, 66 subjects short of target. CuSum illustrated this trend. Subsequent shifts in enrollment trends were shown with improvements in referral. DISCUSSION: Indications for use of CuSum include (a) earlier detection of enrollment trend shifts, and (b) earlier discrimination between effective and ineffective recruitment. Thus, CuSum has implications for both evaluating the effects of planned and unplanned process changes and for managing the recruitment process.
Authors: Ann Boonzaier; Annabel Pollard; Maria Ftanou; Jeremy W Couper; Penelope Schofield; Linda Mileshkin; Michael Henderson Journal: Support Care Cancer Date: 2010-09-11 Impact factor: 3.603
Authors: Steffany J Fredman; Donald H Baucom; Tina M Gremore; Angela M Castellani; Theresa A Kallman; Laura S Porter; Jennifer S Kirby; E Claire Dees; Nancy Klauber-Demore; Jeffrey Peppercorn; Lisa A Carey Journal: Psychooncology Date: 2009-06 Impact factor: 3.894