Table 2 Information for tested models.

From: Changing trends in the prevalence of H. pylori infection in Japan (1908–2003): a systematic review and meta-regression analysis of 170,752 individuals

 

AIC

BIC

LogLik

Model 1:

Logit(P) = s(birth year) + r(study ID) + f(source of population) + f(diagnostic test) + f(ELISA kits) + f(research year)

1687.895

1880.004

792.0792 (df = 51.87)

Model 2:

Logit(P) = s(birth year) + r(study ID) + f(diagnostic test)

1702.257

1889.008

−800.7071 (df = 50.42)

Model 3:

Logit(P) = s(birth year) + r(study ID)

1702.936

1890.291

−800.8835 (df = 50.58)

  1. Abbreviations and definitions:
  2. AIC: Akaike’s information criterion;
  3. BIC: Bayesian information criterion;
  4. LogLik: Log-likelihood;
  5. P: prevalence;
  6. s: penalized cubic spline;
  7. r: random effect;
  8. f: fixed effect;
  9. df: degree of freedom.
Back to article page