Demographics
Of a total of 1,273 beds across the acute and community hospital, 118 patients were on the dysphagia caseload. 13 patients were receiving oral trials, 10 other patients had been on oral trials previously during their stay and six patients were nil by mouth and had never been on oral trials. No missing data was recorded during data collection. Patients were seen across a range of specialisms and presented with a variety of diagnoses (see table 1).
Comparisons between treatment groups (see table 2) show that patients on oral trials were significantly more likely to have a neurological diagnosis (78.3% vs 30.5%, p<0.001), be on a neurological ward (69.6% vs 25.3%, p<0.001), live independently (73.9% vs 42.1%, p=0.006) and were less likely to have dementia (17.4% vs 37.9%, p=0.027). There was a significant difference between age, with individuals in the oral trials group being younger on average (66.70 vs 77.14 years old, p=0.003). The UK uses a pay banding system to grade the levels of responsibilities of SLTs (range band 5-8), band 5 SLTs will usually be newly qualified and have up to 5 years’ experience. Individuals seen by Band 5 SLTs were more likely to be on full oral intake (36.8% vs 8.7%, p=0.009), whilst those seen by mixed SLTs were more likely to be on oral trials (65.2% vs 29.5%, p=0.001).
There was a considerable overlap between some variables measured. Younger individuals were significantly more likely to live independently (p<.001) and individuals with a neurological diagnosis were significantly more likely to be on a neurological ward (p<.001). Individuals with a neurological diagnosis were also younger on average than those admitted for other reasons (68.45 vs 79.51, p<0.001).
Hierarchical logistical regression analysis shows that when neurological diagnosis is included, age and being seen by mixed SLTs were no longer significant predictors of being on oral trials (see table 3). Having a neurological diagnosis was the key statistically significant contributor to the model, with individuals with a neurological diagnosis being over four times more likely to be on oral trials when controlled for age and treating SLT. The model as a whole explained between 17.0% (Cox and Snell R squared) and 27.2% (Nagelkerke R squared) of the variance in treatment group and correctly classified 79.7% of cases.