- Smart Healthcare, Wednesday 5 August 2009 09.00 BST
Sounds interesting: if a single department introducing speech recognition, others often follow. Photo: jiunlimited.com
When it comes to dictation systems, replacing analogue tape-based systems with digital dictation equipment is seen as a natural progression. It results in better sound quality, leading to fewer errors during the transcription process.
Moreover, because users speak into dedicated recording devices or PCs, which create a digital audio file that is transferred over a network for access by transcriptionists, the whole process is speeded up. There is also less risk of files being lost or mislaid as was often the case with tape.
But far fewer NHS organisations have taken the next step: adding a speech recognition layer, where words are immediately converted into text. This can streamline activities further, particularly if the technology is integrated with electronic healthcare record systems, as information can be entered directly into individual documents in real time, although it still needs to be checked for accuracy and edited.
The downside of introducing such technology, however, is that it significantly alters the way that clinicians work. This means that a large change management programme, which includes training, may be required.
In addition, many trusts are only just starting to deploy electronic healthcare record systems, and they tend to introduce the former before introducing speech recognition. Adoption is more widespread in specific departments such as radiology and cardiology however and, once implemented, such systems tend to spread to other departments.
One organisation that has bitten the bullet is East Kent Hospitals University NHS Foundation Trust. As part of its aim to move to a paperless environment, and because its existing analogue dictation technology was coming to the end of its life and becoming increasingly difficult to support, the trust's histopathology department undertook a review in 2007.
It decided on a three-phase plan to update existing facilities with the help of service provider SRC. Paul Williams, head biomedical scientist at the trust's department of cellular pathology, explains the rationale: "We felt that implementing the project in stages was the best way to do it as it's a quantum leap to move from writing to voice recognition in one step. Digital dictation is a useful way to start, but voice recognition is the icing on the cake for improving turnaround times." A significant justification in this context is meeting and exceeding national targets on cancer waiting lists.
The first stage of the £30,000 initiative started in January 2008 and led to the deployment of Winscribe's digital dictation workflow management system to allow pathologists to describe tissue specimens at the cut-up area using wireless headsets. The voice files were sent immediately to a back-end server, where they became immediately available to a team of transcriptionists, resulting in time savings of about 10%.
Going faster at traffic lights
The second phase of the scheme was implemented in June last year. It lets pathologists record microscopy reports at their workstations rather than needing to write them by hand.
The move led to time savings of 20% and also generates an audit trail for voice files, which makes it easier to track changes and find missing information. A traffic light system was provided so that users could see whether transcribed files had been fully, partially or not checked at all.
Then in November, Nuance Communications' Dragon NaturallySpeaking voice recognition software was integrated with both the WinScribe applications and the trust's iSoft Apex laboratory information management system to enable staff to enter information directly into Apex.
Although this approach means that there is no audit trail because voice files are not saved, it does result in same day report turnaround times – as opposed to the one to two weeks which were common in the days of tape, which led to backlogs.
One pathologist out of a team of 10 is currently using the voice recognition system, although another two are expected to start employing it soon. Two others have stated a preference for remaining with digital dictation, while a further two have fast typing skills and prefer to enter information directly. This means that only one is still employing the older analogue system, although this will be phased out by the end of the year.
"The technology side of things isn't a big deal, but what is, is getting people to change their mindset. As a result, training is vital as is following up maybe a month or so later to fill any gaps in people's knowledge," Williams says.
Another important consideration is engaging secretarial staff from the start, as the introduction of such technology inevitably leads to fears of job losses. East Kent dealt with the issue by training the team to the level of Band 2 healthcare support scientists, which means that they can now assist in cutting up specimens.
"The variety means that they now find their job more interesting," says Williams. It also means that the lab will recoup its investment within three years as it has been able to fill the equivalent of 2.5 healthcare support scientist positions without hiring more staff.
The next step is to roll the software out to cover additional applications such as emails and letters and to encourage GPs to access reports held in the LIM system using a web browser. This will enable them to see documents as little as 20 minutes after they have been authorised, rather than waiting for information to arrive on paper.
"The faster turnaround times will be reflected in cost savings in terms of patient treatment as it should result in a faster turnaround of beds because people will be treated more quickly. It's a hidden cost, but for the trust, it all represents savings," concludes Williams.
