Influence of infant/maternal characteristics on cross-method agreement of sleep measures: implications for assessing development

• In Event: Measurement Issues in Developmental Research

Abstract

Introduction: Developmental and sleep science have highlighted sleep as an important facilitator of development (e.g. Scher, 2008), impacting domains such as attention (Geva et al., 2016), memory (Friedrich et al., 2017) or temperament (Netsi et al., 2017). Given the importance of sleep for development, measuring it correctly is crucial. Measurements of sleep include subjective (e.g. parent-report questionnaire/diary) and objective measures (e.g. actigraphy). Agreement between those methods vary from study to study. Cross-method agreement has yet to be studied longitudinally. Moreover, it is unclear what drives the mixed findings with regard to concordance: a) differences across studies, b) infant or c) maternal characteristics? Here, we conduct a longitudinal study investigating the concordance between subjective and objective sleep measures and how infant (i.e. infant age) and maternal factors (i.e. stress levels/education) may impact cross-method concordance. Moreover, implications of the findings for studying development are discussed.
Methods: This study is part of a longitudinal study (up to 4 study visits/participant) exploring the relationship between sleep and early neurocognitive development. 76 typically developing, German (ethnicity: white Caucasian/Turkish) infants were studied (age: 4-14 months, 166 individual study visits). The subjective measures were the Brief Infant Sleep Questionnaire (Sadeh, 2004), the Sleep and Settle Questionnaire (Matthey, 2001) and a 7-days sleep diary and the objective measure was actigraphy (w-GT3X-BT, ActiGraph Corp.). Maternal and infant factors were assessed using an in-house demographics questionnaire and maternal stress measures were the State Trait Anxiety Inventory (Spielberger, 1999) and the Perceived Stress Scale (Cohen, 1994). Neurocognitive measures included eye-tracking (Gap-overlap, Elsabbagh et al., 2009) and the Infant Behaviour Questionnaire (Rothbart, 1982). Concordance between sleep measures was investigated using Bland-Altman plots and tests of equivalence, intra-class correlations (ICCs) and correlational analyses. Follow-up correlational analyses on z-scored maternal anxiety-stress measures/infant age/ maternal education were performed. Linear mixed models (LMMs) between sleep variables and eye-tracking and questionnaire measures were performed to investigate the impact of cross-methods concordance for studying sleep and development.
Results: Cross-method concordance overall was poor, as assessed by correlational analyses, ICCs, and equivalence testing. Moreover, Bland-Altman plots suggest presence of a systematic bias. Closer investigation of the bias showed that higher maternal stress/anxiety was associated with better cross-method agreement rates on night sleep parameters (r = 0.27, p<.05) and that older infants had better cross-method agreement for day sleep (r = .38, p<.001) but worse cross-method agreement for night wakening duration (r=.29, p<.001). Further we present results from LMMs on the relationship between subjective and objective sleep measures and measures of development (Eye-tracking/questionnaires). The latter show that different sleep methods yield different results in association with development.
Conclusion: Different sleep measures vary in their completion rates and agreement. Fragmentation yields better agreement between subjective and objective measures than duration. Agreement is potentially related to infant age and maternal stress/anxiety measures. Different sleep methods yield different results in association with development. Implications for studying neurocognitive development are discussed.

Authors

• Louisa Gossé, Centre for Brain and Cognitive Development
  Presenting Author

• Frank Wiesemann, Procter & Gamble
  Non-Presenting Author

• Clare Elwell, University College London
  Non-Presenting Author

• Emily J.H. Jones, Birkbeck, University of London
  Non-Presenting Author