Adjustment for Propensity Score in Nonrandomized Clinical Studies: Comparison of Sivelestat Versus Conventional Therapy for Acute Lung Injury in Acute Respiratory Distress Syndrome

Satoru Fukimbara, Kouji Niibe, Michio Yamamoto, Takuhiro Yamaguchi

Research output: Contribution to journalArticle

1 Citation (Scopus)


Background: To confirm the effectiveness of sivelestat, a clinical trial was conducted comparing sivelestat with conventional treatment in an open, nonrandomized, multicenter study of patients with systemic inflammatory response syndrome (SIRS)–associated acute lung injury. The primary endpoint was ventilator-free days (VFD). Methods: This study adopted a “cluster entry” method to control for patient selection bias arising from the unblinded and nonrandomized clinical trial. Thus, all patients in the same hospital during the same entry period entered the same treatment arm, and entry periods did not overlap. In the primary analysis of VFD, adjusted mean VFD values were compared between groups using the inverse probability of treatment weighted (IPTW) method, based on propensity score, for control of confounding factors. Results: There were 374 patients in the sivelestat group and 168 in the conventional therapy group. The primary analysis confirmed that sivelestat was effective (between-group difference of adjusted mean was 3.5 [2-sided 95% confidence interval, 1.3-5.8]; P =.0022). Conclusions: In general, a study where all patients in the same cluster enter the same treatment arm has within-cluster correlations, which need to be considered in the study analysis. However, in analysis using the IPTW method, it is usual to use a robust variance estimator, the sandwich variance estimator, which is consistent regardless of whether the specification of within-cluster correlation structure is correct. Thus, in the analysis using the IPTW method, it was found that it was not necessary to adopt any other adjustment method for within-cluster correlation.

Original languageEnglish
Pages (from-to)89-99
Number of pages11
JournalTherapeutic Innovation and Regulatory Science
Issue number1
Publication statusPublished - Jan 1 2017
Externally publishedYes



  • cluster entry
  • inverse probability of treatment weighted
  • nonrandomized
  • postmarketing clinical study
  • propensity score

ASJC Scopus subject areas

  • Pharmacology, Toxicology and Pharmaceutics (miscellaneous)
  • Public Health, Environmental and Occupational Health
  • Pharmacology (medical)

Cite this