LOCF Approach to Handling Missing Data Overestimates the Pain Score Improvement of Drop-Outs

Abstract

The results of Shaibani et al's 2 Shaibani A. Fares S. Selam J.L. Arslanian A. Simpson J. Sen D. Bongardt S. Lacosamide in painful diabetic neuropathy: An 18-week double-blind placebo-controlled trial. J Pain. 2009; 10: 818-828 Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar clinical trial comparing lacosamide to placebo in painful diabetic peripheral neuropathy (DPN) are potentially misleading as presented. The authors defined the primary endpoint as the “within-patient change in the mean daily pain score from baseline to the mean value over the last 4 weeks” of the maintenance period. In order to perform an intention-to-treat analysis incorporating all patients who took study medication, some method of incorporating patients with missing data must be used. The authors of this study used a last-observation-carried-forward (LOCF) method of handling missing data. This approach, which has been previously employed in chronic neuropathic-pain clinical trials 1 O'Connor A.B. The need for improved access to FDA reviews. JAMA. 2009; 302: 191-193 Crossref PubMed Scopus (48) Google Scholar and was also used in the other published phase III clinical trial of lacosamide in painful DPN, 4 Wymer J.P. Simpson J. Sen D. Bongardt S. Efficacy and safety of lacosamide in diabetic neuropathic pain: An 18-week double-blind placebo-controlled trial of fixed-dose regimens. Clin J Pain. 2009; 25: 376-385 Crossref PubMed Scopus (110) Google Scholar overestimates treatment benefit because the pain scores of patients unable to tolerate the treatment are carried forward to the endpoint. In a chronic-pain clinical trial studying an analgesic that is not known to modify the underlying disease, pain scores would be projected to return to baseline levels in patients who stop taking the analgesic; for this reason, a baseline-observation-carried-forward (BOCF) strategy of handling missing data would provide a more accurate estimation of the true pain score during the endpoint period (in this case, weeks 14–18) (See Fig 1). This is an especially important issue in this lacosamide trial, in which more than half of the patients who received 400 mg/day or 600 mg/day dropped out (145/262, 55%), the majority of them because of adverse events (88/145, 61%). Using the authors' endpoint definition and LOCF, patients unable to tolerate lacosamide for the duration of the trial are indistinguishable in the analysis from patients who successfully complete the trial with comparable pain scores. ReplyThe Journal of PainVol. 11Issue 5PreviewTo the Editor: Full-Text PDF