A prognostic model for priority public dental care.

Abstract

Excess demand over the service capacity of public dental services has resulted in ad hoc rationing of access to services. As a result a significant proportion of public dental service patients receive only same-day or priority dental care. However, many such appointments are of low urgency, an outcome which is facilitated by lack of a standardised method to determine access for those seeking priority dental care. Aim: This thesis aimed to develop, validate and trial the implementation of a prognostic model for urgent dental care. Methods: This was a multi-phased research study. The first phase, the Parent study, collected psychosocial data on people seeking emergency dental care in SA and NSW and dentists‟ assessment of urgency. Two sequential prognostic models for urgency of dental care were developed as preliminary research. As further development of the models was to occur only in SA, the sequential prognostic models were examined separately in SA and NSW. The next phase involved testing the validity of the models on patients accessing urgent dental care at two new SA sites over an eight month period in 2004. This led to the development of a single model, the Composite model, and testing of its accuracy. The Composite model was then tested in a 28-week intervention trial in four SA Dental Service clinics in 2006. Both quantitative outcomes and qualitative outcomes were assessed. Results: The Parent study included 839 people seeking urgent dental care. Two sequential logistic regression models were developed to predict dentists assessed urgency of < 48 hours vs 2+ days, and 2-7 days vs 8+ days. The models used responses to 11 psychosocial questions to predict with acceptable accuracy dentist assessed urgency. Small differences in the regression coefficients of predictor variables of urgency were evident in the models specified for SA and NSW separately. Model testing showed acceptable accuracy against clinically determined urgency of dental care when examined on a new set of 294patients from two clinics in SA and showed greater accuracy in assessing urgency than the traditional SA method of receptionist assessment. However, a single Composite model was preferred over the two separate models. The Composite model performed with higher sensitivity and specificity than reception staff and had higher reported AUC values indicating better fit of the model. The implementation trial involved 728 people pre- and 1013 people post-implementation making contact with one of four SA Dental Service clinics. People reporting they had received treatment fell after implementation of the Composite model (74.8 cf 65.6%). Of those who received care, a lower percentage received care from the SA Dental Service (70.2% cf 67.9%) and subsidised private treatment (22.2% cf 17.4%) and more from private dentists (7.6% cf 14.7%). Post-implementation, a lower proportion of people receiving treatment from the SA Dental Service reported experiencing pain at first contact. Within the SA Dental Service, a decrease in the percentage of staff time spent on urgent dental care occurred (60.2% cf 39.8%) with a matched increase in time spent time spent on general, preventive dental care. Focus group discussions showed that staff liked the transparency, support and equity the prognostic model provided. Staff remained concerned about those denied urgent dental care, but concern was ameliorated by the possibility in exceptional circumstances to over-ride the prognostic model. Conclusions: Further development of a prognostic model for urgent dental care led to a single composite model with acceptable accuracy in the clinical setting. The single prognostic model was more accurate in predicting dentist assessed urgency and was also found to be more transparent, consistent and equitable than relying on receptionist judgement. Implementation of the prognostic model rationed urgent dental care. The allocation of SA Dental Service staff time and resources moved away from urgent and towards general preventive care.