Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2009 Jan;19(1):15-24.
doi: 10.1016/j.annepidem.2008.08.009. Epub 2008 Oct 4.

Use of Penalized Splines in Extended Cox-type Additive Hazard Regression to Flexibly Estimate the Effect of Time-Varying Serum Uric Acid on Risk of Cancer Incidence: A Prospective, Population-Based Study in 78,850 Men

Affiliations
Free PMC article

Use of Penalized Splines in Extended Cox-type Additive Hazard Regression to Flexibly Estimate the Effect of Time-Varying Serum Uric Acid on Risk of Cancer Incidence: A Prospective, Population-Based Study in 78,850 Men

Alexander M Strasak et al. Ann Epidemiol. .
Free PMC article

Abstract

Purpose: We sought to investigate the effect of serum uric acid (SUA) levels on risk of cancer incidence in men and to flexibly determine the shape of this association by using a novel analytical approach.

Methods: A population-based cohort of 78,850 Austrian men who received 264,347 serial SUA measurements was prospectively followed-up for a median of 12.4 years. Data were collected between 1985 and 2003. Penalized splines (P-splines) in extended Cox-type additive hazard regression were used to flexibly model the association between SUA, as a time-dependent covariate, and risk of overall and site-specific cancer incidence and to calculate adjusted hazard ratios with their 95% confidence intervals.

Results: During follow-up 5189 incident cancers were observed. Restricted maximum-likelihood optimizing P-spline models revealed a moderately J-shaped effect of SUA on risk of overall cancer incidence, with statistically significantly increased hazard ratios in the upper third of the SUA distribution. Increased SUA (>/=8.00 mg/dL) further significantly increased risk for several site-specific malignancies, with P-spline analyses providing detailed insight about the shape of the association with these outcomes.

Conclusions: Our study is the first to demonstrate a dose-response association between SUA and cancer incidence in men, simultaneously reporting on the usefulness of a novel methodological framework in epidemiologic research.

Figures

FIGURE 1
FIGURE 1
Estimated adjusted hazard ratios (solid line) with 80% (dark grey) and 95% (light grey) confidence intervals for the association of SUA (in mg/dL) and overall cancer incidence (n = 5189) in 78,850 male VHM&PP participants from REML-optimal, extended Cox-type additive hazard regression, adjusted for age, BMI, smoking status, occupational status, and year of examination. The nonlinear effect of SUA on risk of overall cancer incidence was modeled with the use of a penalized spline expansion, with SUA as a time-varying covariate. A SUA concentration of 4.5 mg/dL, as the mid-point of the laboratory reference range, was used as reference value for the calculation of hazard ratios. BMI = body mass index; REML = restricted maximum-likelihood; SUA = serum uric acid; VHM&PP = Vorarlberg Health Monitoring and Promotion Program.
FIGURE 2
FIGURE 2
Estimated adjusted hazard ratios (solid lines) with 80% (dark grey) and 95% (light grey) confidence intervals for the association of SUA (in mg/dL) with malignant neoplasms of (a) digestive organs; (b) respiratory system/intrathoracic organs; (c) bone, connective tissue, soft tissue, and skin; (d) genital organs; (e) urinary organs; and (f) lymphoid, hematopoietic and related tissue in 78,850 male VHM&PP participants from REML-optimal extended Cox-type additive hazard regression adjusted for age, BMI, smoking status, occupational status, and year of examination. The effects of SUA on risk of site-specific malignancies were modeled with a penalized spline expansion, with SUA as a time-varying covariate. A SUA concentration of 4.5 mg/dL, as the mid-point of the laboratory reference range, was used as reference value for the calculation of hazard ratios. For abbreviations, see legend to Figure 1.
FIGURE 3
FIGURE 3
Comparison of estimated adjusted logits of the predicted probability of cancer incidence according to SUA concentrations (in mg/dL) in 78,850 male VHM&PP participants from different extended Cox-type additive hazard regression models. (a) REML-optimal, AIC-optimal, and BIC-optimal model with time-varying SUA modeled as penalized spline and time-varying SUA modelled as cubic polynomial. (b) Time-varying SUA modeled as linear function and categorized variable (using quintiles) in comparison with the REML-optimal P-spline model. 95% CIs are shown for the REML-optimal P-spline model. All models are adjusted for age, BMI, smoking status, occupational status, and year of entry into the cohort. For abbreviations, see legend to Figure 1.

Similar articles

See all similar articles

Cited by 15 articles

See all "Cited by" articles

Publication types

Feedback