ive thought about doing a post about 'the thing' off and on for a while and always thought the better of it. i mean does anyone really want to know? well, apparently some of ya do! Writing up a precis is a good chance for me to conslidate, as i still have my thesis abstract to write. You're it, my guinea pigs, so if its all gobbledygook, let me know. Its pretty hard to spice up mathematical modelling and public health, but i'll give it my best shot and try and explain why its recieved the attention it has. Its also pretty hard to stand back and not be too scientific and dogmatic about how i tell the story. Im so used to systematic and technically correct (scientific) writing!
Umm, background. My field is oral epidemiology; epidemiology is the science of looking at the determinants and distribution of diseases (who gets what and why). By its nature, epidemiology concerns itself not with the biology of disease but the public health context of illness/wellness.
The research question i set out to answer was ' can we develop an index by which the relative urgency of an individuals need for dental care can be determined?'. The reason i was interested in doing this research is that in the public sector, receptionists were making decisions about who got to see a dentist and who didnt and it appearred that in a lot of cases they were giving emergency priority to people who were not really urgent. Also, by tradition, people were given appointments on a first come first served basis, not urgency of need.
This work has been about 10 years in the making and it started with a very long series of questions asked to a large random selection of persons presenting to public dental clinics in two states of australia for both emergency and general dental care. These questions covered just about everything to do with an individuals oral health; hygiene practices, sugar consumption, dental visiting behaviours, medications taken, pain perceptions, sociodemographic characteristics, past dental treatments, etc. We asked pretty much everything about everything. These persons were then tracked through the dental system and all treatment provided to these people was recorded. Two dentists also provided independent opinion about the urgency of each patients dental problem.
Then i used statistics to match up question responses with dentists urgency categories to see if any questions were highly associated with urgency. This analysis resulted in 10 very predictive questions.
You still with me?
These questions were then tested on the computer on a pretend set of patients to see if the same questions would still be useful in predicting urgency and it wasnt just chance that led us to these 10 questions. They were. This is about 4 years of work so far. Research is like that. it takes ages when you deal with human populations. Give me a test tube! Ok!.. the questions work well in predicting urgency when i use a fake set of people. what about real people?
So then i took these 10 questions and had receptionists use them on the phone on real patients. I had dentists again rate the patients urgency. This was the first test of the questions on a new set of real people. After more analysis, the questions still matched up well with dentists ratings. What really 'Wowed' us was that the 10 questions worked better than receptionists at figuring out who needed most urgent care and was more like the dentists rating of urgency. Im getting excited now. It looks like i have found a systematic way to determine urgency for dental treatment.
Soo, then the public dental service gets all excited about my results and decides to let me do a full trial using my predictive model. They give me 4 clinics to run the trial and we actually say 'yes' or 'no' to people using responses to my series of questions. New computer interfaces for the dental clinic databases were made and installed and my special algorithm runs quietly in the background and calculates scores to peoples responses. It then tells receptionists the urgency category of each person calling and how long until an appontment should be made.
I do even more analysis and the questions still identify people with the most urgent need. It gets even better. Gasp, 'how can that possibly be'? i hear you all crying! This new way of determining need also lets the system better manage its bookings and over a year of using my system, these 4 clinics manage to change their service around from being mostly an emergency system;extracting rotten teeth which hurt, and start giving people check ups as well; looking after teeth, providing fillings and helping people keep their teeth healthy.
Before i have finished my dissertation, the system is rolled out in every clinic in the state and now all clinics are able to provide routine check ups for public dental patients, not just emergency dental services. It was a very thorough and scientifically rigourous piece of health services research using 'gold standard' design, process and analysis AND its had a great public health result. Hence its award nomination. Wheeeeee!