Health Service Focus

The Friends & Family Test – a gold mine of patient feedback

How will the NHS handle millions of individual comments? Jim Ward, director at Good4HEALTH Ltd, gives his view.

In April this year, following new guidelines from the Department of Health, hospitals started surveying patients within 48 hours of discharge.

Known as the Friends and Family Test (FFT), the survey was announced by Prime Minister David Cameron in May 2012. Following a staged implementation, it is expected that by 2015 every patient, following treatment at any NHS service, will be offered an opportunity to respond to the FFT along with accompanying questions. One of these questions will be a request to provide a comment on the service received and will enable every patient to give their views and provide an effective way of identifying common concerns that could relate to service improvement. 

Not only does giving participants an opportunity to give a verbatim response signify that their opinions are cared about, the NHS is also able to garner information that would otherwise be lost in the straitjacket of a closed response type. However, with the average hospital discharging around a quarter of a million patients a year, it is easy to see how the task of coding and classifying large volumes of text could appear to be unworkable. 

Using Intelligent Discourse Analysis (IDA) it is possible to automate the process of interpreting verbatim responses in a fraction of the time it would take a human coder and with the same or better coding accuracy. 

IDA doesn’t merely assist with the task of coding by making an enhanced digital process out of the traditional coding methods, nor does it rely on a ‘text mining’ method of searching for words or certain text within the responses; rather it uses machine learning to understand the code frame and fully automate the response analysis. 

Unlike some products that ‘text mine’ or search for words or phases in order to attempt to categorise responses, IDA extracts linguistic patterns from each response in order to code it. By looking at different levels – from individual words, noun phrases, verb phrases and entire sentences to semantic and sentiment related content – IDA builds a pattern from the response and in turn determines the codes that should apply. 

Traditionally the price paid for including verbatim feedback opportunities in surveys is that collating and theming the responses for use as quantitative information takes time and requires a human coder to develop a code frame and go through the responses. 

The Synapta platform offers an industrial-strength software system for automatically coding open-ended survey responses. 

The system is based on a learning metaphor, where automatically generated verbatim coders learn (from a user-provided sample of manually coded verbatims) the characteristics that new, uncoded verbatims should have in order to be attributed the codes in the code frame. 

In plain terms, the function currently handled by humans – reading a comment, then pigeon-holing each comment into multiple frames – has been replicated by a sophisticated learning system that you teach over a short period of time. Once taught how to handle the tasks, it will work at machine-speed to segregate comments into multiple codes. 

In the NHS, managers are interested in comments relating to key areas relating both to transactional and relational aspects of care. IDA can effectively support the NHS to interrogative data, produced automatically, from millions of individual comments. The payback is authoritative system improvement based on evidence.

 A launch within the NHS will soon be presented by NHS Patient Experience consultants Inspiration NW and FFT providers Good4HEALTH along with our technical provider The 3rd Degree.