Page 184 - Functional impairment and cues for rehabilitation of head and neck cancer patients -

182
Chapter 7
DISCUSSION
The aim was to develop a prediction model to estimate the risk of prolonged (> 90 days) feeding tube dependency, which can be helpful in deciding on proactive tube placement for patients receiving primary CRT for HNC. In univariable analysis, T4-stage tumor, BMI, weight loss, pain, dysphagia and FOIS below 7 were significant risk factors. The multivariable prediction model included T-stage, BMI, weight loss, and FOIS which resulted in a model with fair discriminative ability and adequate agreement between predicted and observed probabilities.
The model and corresponding nomogram include easily obtainable parameters and are therefore a practical tool for clinicians to estimate the risk of prolonged feeding tube dependency. This aids in shared decision making regarding proactive placement in high risk patients, preventing treatment interruption with benefits on tumor control (35). Identifying low risk patients can prevent unnecessary tube placements, reducing complications of placement (e.g., infection, bleeding, and perforation), reducing costs and preventing non-use atrophy of swallowing muscles (36). With a shift from authority-based medicine towards shared decision- based practice, the estimated risk can be used to inform patients and make educated decisions (37).
To assess the clinical net benefit of decisions based on the prediction model, decision curve analysis was performed (33). This analysis showed that a decision regarding proactive placement based on the model instead of providing all or no patients a prophylactic tube, has net benefit for probability thresholds between 35 and 80%. Since this falls within a plausible range of probability thresholds likely to be considered by clinicians and patients, we assume use of the model has clinical benefit in most cases. For patients or clinicians who would consider proactive placement at probability thresholds below 35% however, a treat all policy would yield the same net benefit (33).
Patients who receive a feeding tube before the start of CRT have an increased risk on prolonged feeding tube dependency. The vast majority (n = 55/59; 93%) of patients who received a feeding tube before or in the first week of CRT became prolonged feeding tube dependent. We therefore advise to directly place a PRG instead of an NGT in these patients, avoiding the disadvantages of an NGT including discomfort, prolonged feeding times, shorter tube lifetime and cosmetic grievances. This study also supports a reactive approach, since the later the placement, the lower the risk of prolonged dependency, thus optimal patient support to maintain oral intake along with preventive swallowing exercises seems to make earlier return to oral intake more likely.
Several studies have investigated predictive factors for prolonged feeding tube dependency in patients treated with CRT for advanced HNC (16, 18, 20, 22, 23, 25, 26). The predictors found in our study were broadly in accordance with these studies except for the predictors dysphagia, nodal stage and high dose bilateral neck irradiation. Firstly, pre-existing (subjective) dysphagia