Table 3

Summary of recommendations for reporting propensity score (PS) methods

Elements to be reportedMethodological recommendations
Variable selection strategy for PS model
  • Potential confounders

  • Select variables a priori

  • Optional: strong predictors of the outcome

Balance diagnostics
  • Standardised mean differences (threshold <10%)

  • Graphical representation of PS distribution*

  • Optional: variance ratio

Matching
  1. Ratio for matches

  2. Matching strategy

  3. Number of subjects and balance diagnostics pre-match and post- match

  4. Variance estimation

  • 1:1 or 1:2 matching is sufficient

  • Nearest-neighbour with callipers strongly preferred

  • 0.2 SD of the logit of the PS

  • Untreated subjects chosen with or without replacement

  • Without replacement—untreated matches chosen with greedy or optimal matching

  • Account for matched pairs in outcome model with clustering, stratification or regression

  • Account for matching with replacement

Inverse probability weighting
  1. Application of weights

  2. Extraneous values

  3. Variance estimation

  •  Throughout; otherwise, if heterogeneity in treatment effect expected, apply subgroup-specific weights

  •  Use stabilisation, trimming and truncation, if appropriate

  •  Non-parametric bootstrap method preferred

Stratification
  1. Number of strata

  2. Size of strata

  3. Combine estimates

  •  Five strata

  •  Equal-sized or unequal-sized strata

  •  Pool stratum-specific estimates using the proportion of subjects in each stratum

Direct adjustment
 Balance diagnostics
  •  Conditional standardised difference or quantile regression

Causal interpretations
1. Inclusion criteria
2. Target population
  1. ATE

  2. ATT

  3. ATU

  • Methods consistent with target population

  • Describe the inclusion criteria and:

  • Treatment effect in treated and untreated groups

  • Treatment effect in the treated subgroup only

  • Treatment effect in the untreated subgroup only

  • *Kernel density plots, histograms, cumulative distribution functions, quantile–quantile plots, side-by-side box plots, etc.

  • ATE, Average treatment effect; ATT, average treatment effect in the treated; ATU, average treatment effect in the untreated.