May 2006
Volume 47, Issue 13
ARVO Annual Meeting Abstract  |   May 2006
Does Exclusion of Early Failures in Persistence Analysis Affect Results? A Sensitivity Analysis of Patients New to Therapy
Author Affiliations & Notes
  • G. Reardon
    Informagenics LLC, Worthington, OH
  • G.F. Schwartz
    Glaucoma Consultants, Greater Baltimore Medical Center
  • S.N. Shah
    Pfizer Ophthalmics, New York, NY
  • Footnotes
    Commercial Relationships  G. Reardon, Pfizer Inc, C; G.F. Schwartz, Pfizer, Allergan, Santen, F; Pfizer, Merck, R; S.N. Shah, Pfizer Inc, E.
  • Footnotes
    Support  Research Support from Pfizer Inc
Investigative Ophthalmology & Visual Science May 2006, Vol.47, 4411. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      G. Reardon, G.F. Schwartz, S.N. Shah; Does Exclusion of Early Failures in Persistence Analysis Affect Results? A Sensitivity Analysis of Patients New to Therapy . Invest. Ophthalmol. Vis. Sci. 2006;47(13):4411.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Purpose: : A recent study (Wilensky et al., AJO 2006) introduced a method for measuring persistence on prostaglandin (PG) therapy that excludes patients from analysis who fail to refill their initial PG within 90 days after the start. The current study evaluates the impact of this exclusion rule on persistence findings.

Methods: : Data was derived from the Constella managed care database. Patients initiated therapy with bimatoprost (BIM), latanoprost (LAT), or travoprost (TRA) between 4/1/01 and 12/31/01. The methodology developed by Wilensky et al. was replicated, except that patients who failed to refill the index prescription within the first 90 days were included and patients were required to be enrolled for 1 year after therapy start. Persistence outcome measures were: 1) whether the patient had any index medication on hand on the last day of the first therapy year and 2) the total number of days in the first therapy year that the patient had a supply of index medication. Results were run for the set of all patients (inclusion) and by applying the 90–day refill exclusion rule.

Results: : 574 patients met the inclusion criteria (BIM, n=126; LAT, n=306; TRA, n=142). 320 (56%) were still retained after applying exclusion. In a regression analysis, compared to LAT, patients with BIM had 55 fewer days of therapy in the first year (p<0.001) with inclusion vs. 45 fewer days (p=0.005) with exclusion. TRA had 79 fewer days (p<0.001) with inclusion vs. 71 fewer days (p<0.001) with exclusion. Compared to patients who refilled their index therapy, patients who failed to refill had 130 fewer days of therapy (p<0.001, t–test), and were 44% less likely to have a glaucomatous–coded medical visit within the first year (p=0.004, logit z–test).

Conclusions: : Excluding patients who failed to refill their index medication within 90 days affected persistence findings. These patients had poor persistence with both medication and medical visitation within the first year. Choice of medication was strongly associated with persistence.

Keywords: clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical (human) or epidemiologic studies: treatment/prevention assessment/controlled clinical trials 

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.