Speech recognition should be used as an opportunity to develop improved cross agency working, argues Tom Rothwell, MD of clinical correspondence specialists Medisec Software.
One of the longest-standing bugbears for the NHS has been the highly inefficient way in which secondary care practitioners communicate patient information with GPs.
The immediate patient discharge summaries from hospital are handwritten in haste on ‘carbonised’ paper and are frequently incomplete and/or illegible. The information they contain is very basic at best, and inaccurate at worst.
The more detailed discharge notes which follow from the consultant often do not arrive until weeks, and sometimes even months, after the patient has been discharged. As a result, GPs often have to take over the care of patients who have had a major illness with minimal information available in a timely manner.
The problem is not helped by hospital issues such as delays in clinical dictation, delays in medical secretaries typing and delays in delivering letters to the corresponding GP.
Recent advances in clinical correspondence technology have presented us with the opportunity to do much of this automatically and in a fraction of the time it takes to do it manually. We have some highly automated and joined up solutions which automatically generate the letter template and provide a managed process for delivering the letter electronically to the GP, leaving the secretary with the job of typing the body content of the letter only.
The powerful impact of speech recognition technology within the field of clinical correspondence would appear obvious. The ability to process speech into text automatically is reaching the accuracy of transcription service and has a naturally quicker turn around time. PC hardware will support the high processing demands of speech recognition software and general levels of everyday English translation work well.
However, just because information is generated much more efficiently using speech recognition technology does not mean that it will be used more effectively, especially if the solution is presented in isolation.
Giving speech recognition software to clinicians supplies an efficient means of generating the body content of the letter. But it equally creates a need to generate the patient details (name, address, GP, NHS number etc) to complete the letter. Clinicians are unlikely to be comfortable doing this administrative work.
The point is that you need a correspondence system to manage the speech-captured information in the first place. Otherwise – apart from the initial wow factor – dictating notes into a speech recognition machine saves very little time over dictating them into a Dictaphone. A medical secretary will still have to process them, add the relevant details, print them out, wait for the consultant to check them and put them in the queue for the post.
But let’s consider, for a moment, a classic ‘what if’ scenario. What if a consultant could bring up the discharge screen on the ward based PC, select the patient, dictate his notes into the voice recognition system, check they are correct, hit a ‘complete info’ button which automatically populates the notes with discharge summary information such as GP details, diagnostics and test results, medication and demographics from the central PAS, and then hit the ‘send’ button, delivering the summary in a secure GP area, ready for access by the GP the next day.
There are some exciting precursors already in and working. The Countess of Chester Foundation Trust Hospital in Cheshire has been using automated generation and management of clinical correspondence for many years and more recently the generation and dispatch of radiology reports.
Ordinarily, these documents would be reported by radiologists onto dictation tape for secretaries to type and print; radiologists would then proof read the printout, amend the paper copy and return for the secretary to amend on the computer system and then reprint prior to sign off on paper and on the computer system.
In the meantime, the GP, and the anxious patient, wait.
At the Countess of Chester, speech recognition software is being used to capture radiology reports, with radiologists subsequently proof reading the reports, alongside the image, after an hour to ensure that the proof reading is undertaken with a 'refreshed' set of eyes. The radiologist amends and signs off the report, which is automatically delivered to the relevant GP surgery the same day along with all other electronic clinical correspondence. The report is automatically imported via document management software into the GP's 'in box', ready for when the patient calls or visits.
This is a highly efficient managed process, providing a practical solution to an enduring problem.
Imagine what a difference could be made to patient care if the two technologies were combined on a wider scale to discharge notification forms, discharge summaries and clinic letters.
With integrated speech recognition and the top and tail of the letter automatically generated at the click of a button, clinicians could generate the whole letter themselves, proof read it and sign it off on the same day – cutting the average NHS turnaround time by several days if not weeks.
If integration is the ‘Holy Grail’ of the NHS and of NPfIT in particular, we need to take a wider contextual view of exciting new developments like speech recognition and use them as an opportunity to develop improved cross agency working.
Clever technology for its own sake will win admirers, but not lasting friends. It is less about how you capture the information, and far more about what you do with it afterwards that matters. As with all IT, it only becomes a viable solution when it makes peoples’ lives easer or directly improves standards of patient care.
