Skip to main content

Table 3 Source of variation in delay for persons residing at each level of models, obtained from Multilevel Logistic Regression Models (GLLAMM)

From: General practitioner characteristics and delay in cancer diagnosis. a population-based cohort study

  Origin of variation   Variation1 Proportion of
Delay type (level of model) No. of units (SE)2 total variation
All cancer types     
Patient delay Patients (level 1) 801 π2/3 100,00
  GPs (level 2) 266 0.00 (0.00) 0,00
  Practices (level 3) 187 0.00 (0.00) 0,00
Doctor delay Patients (level 1) 1234 π2/3 92,89
  GPs (level 2) 302 0.00 (0.00) 0,00
  Practices (level 3) 200 0.25 (0.20) 7,11
System delay Patients (level 1) 920 π2/3 94,29
  GPs (level 2) 283 0.00 (0.00) 0,00
  Practices (level 3) 194 0.20 (0.13) 5,71
After exclusion of gender-specific cancers    
Patient delay Patients (level 1) 545 π2/3 100,00
  GPs (level 2) 233 0.00 (0.00) 0,00
  Practices (level 3) 171 0.00 (0.00) 0,00
Doctor delay Patients (level 1) 819 π2/3 91,60
  GPs (level 2) 276 0.00 (0.00) 0,00
  Practices (level 3) 194 0.30 (0.30) 8,40
System delay Patients (level 1) 587 π2/3 100,00
  GPs (level 2) 250 0.00 (0.00) 0,00
  Practices (level 3) 182 0.00 (0.00) 0,00
  1. 1Level 1 variation was assumed to be π2/3 = 3.29, π2 = 3.1416 (Snijders and Bosker, 1999)
  2. 2SE, standard error