Abstract
Inflammatory markers including C-Reactive Protein (CRP) are increasingly used within research and clinical settings. Yet, varying methodologies for cleaning immunoassay data with out of range (OOR) samples may alter characteristic levels of CRP, thereby obscuring interpretation and reliability. This study investigated the influence of eight immunoassay OOR data treatment techniques on salivary CRP (sCRP) samples from at-risk adolescents. Participants from the 'Sleep and Education: learning New Skills Early' (SENSE) Study were 86 adolescents at-risk for depression (50 female), aged 14.29 years (SD = 1.04). ANOVA results showed no statistically significant differences in average morning (F(7, 590) = 1.24, p = .28) and evening (F(7, 599)=1.29, p = .25) values produced by each OOR data cleaning technique. However, varying techniques produced differences in the magnitude of Pearson's correlations between consecutive saliva samples (r's between 0.27-0.78), and influenced the significance of a sCRP diurnal pattern; two techniques produced statistically higher morning than evening sCRP levels (t(85) = 2.70, p = .01 and t(85) = 2.67, p = .01), whereas six techniques failed to find statistical differences between morning and evening sCRP levels (p's >.05). Varying techniques also produced statistically divergent associations between sCRP and age and depressive symptoms. Results from this study provide evidence for the temporal stability of sCRP among adolescents, show winsorization as an effective OOR data management technique, and highlight the influence of methodological decisions in cleaning salivary biomarker data and the need for consistency within the field.