Detecting Faltering Growth in Children via Minimum Random Slopes

A child is considered to have faltered growth when increases in their height or weight starts to decline relative to a suitable comparison population. However, there is currently a lack of consensus on both the choice of anthropometric indexes for characterizing growth over time and the operational definition of faltering. Cole's classic conditional standard deviation scores is a popular metric but can be problematic, since it only utilizes two data points and relies on having complete data. In the existing literature, arbitrary thresholds are often used to define faltering, which may not be appropriate for all populations. In this article, we propose to assess faltering via minimum random slopes (MRS) derived from a piecewise linear mixed model. When used in conjunction with mixture model-based classification, MRS provides a viable method for identifying children that have faltered, without being dependent upon arbitrary standards. We illustrate our work via a simulation study and apply it to a case study based on a birth cohort within the Healthy Birth, Growth and Development knowledge integration (HBGDki) project funded by the Bill and Melinda Gates Foundation.