Although it is statistically accurate to consider values outside the healthy reference interval (i.e., the 5% of data in the tail(s) of the distribution) as “abnormal”, or “extreme”, we need to consider whether or not the boundaries of the reference interval, are good boundaries for deciding whether or not somebody has a value of clinical concern. The answer here is not necessarily. In fact, using the boundaries of the reference interval for this purpose is probably too stringent because values in healthy people and values in people who genuinely have a clinical condition of concern are likely to overlap to some degree. Thresholds used to identify values of potential clinical concern are called “Clinical Decision Points” and they are often set at values that approach but do not reach the boundaries of the reference interval. For example, the upper boundary of the reference interval for measures of fasting blood glucose in healthy adults is ≥ 6 mmol/L (with some variations in exactly where this line falls due to age, sex and body mass index). However, the clinical decision point that is used as the clinically accepted value for triggering concern and diagnosing a patient with “prediabetes” is ≥ 5.6 mmol/L.
When it comes to swallowing, we don’t yet have enough data to set clinical decision points with confidence, but we believe that there is probably value in proposing clinical decision points at points along the distribution that are less extreme than the boundaries of the reference interval. Therefore, we propose using the term “typical” to describe values that fall in the central 50% of the healthy data distribution, and “atypical” for all values that fall outside that range, either below the 25th or above the 75th percentile. “Atypical” values include values that are truly extreme, in the tails of the distribution, and also values that are approaching those tails. If we are looking at a skewed parameter, would be clinically interested in only one side of the distribution rather than “atypical” values on both sides. In Figure 4 below, we use traffic light colour coding to represent typical values on a bell-shaped normal distribution in green-shaded areas and to divide the atypical ranges into red-coloured extreme values and amber-coloured values outside the typical range, but approaching the boundaries of the reference interval.
Figure 5 shows an analogous image for a positively skewed distribution:
Next: Frequently-asked-questions about the ASPEKT Method
Updated January 3, 2023