Table 3

Results latent class model (LCM) distinguished four classes of respondents

LCM (responder type)Latent class 1
All attributes (except safety) balanced
Latent class 2
Emphasis on efficacy, safety and patient′s preference, economic aspects less relevant
Latent class 3
All attributes except cost-effectiveness balanced
Latent class 4
Emphasise on efficacy, patient dislike and costs
Average class probability†43.2%22.6%23.1%11.1%
 Moderate responseReference levelReference levelReference levelReference level
 Good response1.98***3.65***2.32***1.74***
 Very rareReference levelReference levelReference levelReference level
Patient’s preference
 Favours treatment0.33***2.35***−0.37**0.53
 NeutralReference levelReference levelReference levelReference level
 Disfavours treatment−0.84***1.53***−1.34***−3.17***
Cost-effectiveness (10 000 EUR/quality adjusted life year)−0.29***−0.28*−0.03−0.07
Overall medication costs (1000 EUR/year)−0.13**−0.02−0.11***−0.09***
Constant 0.02−1.64***−0.22**−0.34
Class probability model—cluster 2‡1.41***2.09***1.16*
  • *significant at 10%, **significant at 5%, *** Significant at 1%.

  • †Average class probability=probability that individual respondent chooses drug treatments according to respective responder types (all 559 responses from 12 countries included in the analysis).

  • ‡Parameter estimates indicate that rheumatologists belonging to country cluster 2 (consisting of Romania, France, Portugal and Italy) compared with country cluster 1 (reference) are significantly more likely to be in classes 2, 3 or 4.

  • EUR, Euro.