Print Email

BMJ Endgames

Kaplan-Meier Survival Analysis: Types of Censored Observations

Philip Sedgwick
Reader in medical statistics and medical education
Centre for Medical and Healthcare Education, St George’s, University of London, London, UK

Researchers investigated if a low dose of the angiotensin converting enzyme (ACE) inhibitor ramipril reduced the incidence of cardiovascular and renal events in patients with type 2 diabetes. A randomised, double blind, parallel group trial design was used. Intervention was ramipril (1.25 mg/day) or placebo for at least three years. Participants were 4912 patients with type 2 diabetes aged over 50 years who used oral antidiabetic drugs and had persistent microalbuminuria or proteinuria. Participants received their usual treatment in addition to their allocated intervention.

The primary outcome was cardiovascular death, non-fatal myocardial infarction, stroke, heart failure leading to hospital admission, or end stage renal failure. Participants were followed for a median of four years (range 3-6 years). Of 2443 participants randomised to ramipril, 362 (14.8%) experienced a cardiovascular or renal event, compared with 377 (15.3%) of the 2469 randomised to placebo.

Kaplan-Meier survival analysis was used to compare the treatment groups in the length of time after randomisation until first occurrence of the primary outcome. Patients who did not experience the primary outcome during follow-up had their survival times censored. This included those participants who were followed until the end of the study; their primary outcome was censored on the date of last follow-up. A further 878 (17%) participants withdrew and refused to be followed up subsequently or were lost to follow-up. Those patients who withdrew were censored at the time of discontinuation, while those lost to follow-up were censored at the time of last known contact.

It was reported that there was no significant difference between low dose (1.25 mg) ramipril and placebo in the time until the first primary outcome for patients with type 2 diabetes and albuminuria or proteinuria (hazard ratio 1.03; 95% confidence interval 0.89 to 1.20, P=0.65).



Click here to view the answer


Competing interests: None declared.

  1. Hull SC, Colloca L, Avins A, Gordon NP, Somkin CP, Kaptchuk TJ, et al. Patients’ attitudes about the use of placebo treatments: telephone survey. BMJ 2013;346:f3757.
  2. Sedgwick P. Bias in clinical trials. BMJ 2011;343:d4176.

Cite this as: BMJ 2013;347:f5265

 BMJ Logo

This post is based on an article published in "Endgames" an educational series from The BMJ.  Other Endgames articles are available here. Copyright BMJ Publishing Group 2013