Development and Validation of an Algorithm to Objectively Quantify Adherence to Offloading Interventions

Diabetic foot ulcers require offloading for healing. However, adherence with removable offloading devices is often poor. Prior objective measures of adherence yielded important behavioral insights, despite substantial methodological limitations. This study sought to address these limitations by developing a method to track offloading adherence during walking and standing paired with tracking contralateral therapeutic footwear (shoe-lift) use.

Thirty healthy adults wore a removable cast walker (RCW) and contralateral shoe-lift. Accelerometers were placed on both and unilaterally on participants' thighs. Participants performed activities in the laboratory that were logged by investigators, after which monitoring continued for 24 h with participants logging footwear use in a diary. Walking and standing episodes were quantified via the thigh worn monitor. A custom algorithm was developed to classify RCW and shoe-lift adherence during walking and standing based on sample-to-sample variance thresholds. Algorithm accuracy was determined by comparing its adherence metrics with log entries.

The final algorithm resulted in > 99% accuracy in detecting steps and standing while using the RCW and shoe-lift during in-lab testing. In 24-h, community monitoring, algorithm-derived adherence showed excellent agreement with diary-based adherence (ICC > 0.96, p < 0.01). Error rates (misclassifications) were < 3%.

A method to objectively quantify RCW and contralateral shoe-lift adherence use was developed and validated. Strong agreement with lab observations and user diaries support its use for monitoring real-world offloading behavior. This method will allow for enhanced assessments of adherence in future trials and may eventually be incorporated into clinical practice for patient monitoring.