Intended for healthcare professionals

CCBYNC Open access

Rapid response to:

Research

Evaluation of telephone first approach to demand management in English general practice: observational study

BMJ 2017; 358 doi: https://doi.org/10.1136/bmj.j4197 (Published 27 September 2017) Cite this as: BMJ 2017;358:j4197

Rapid Response:

Errors & omissions in calculating workload annul a key finding

My company GP Access Ltd provided just under half the subject practices for this study and all the operational data on consultations (extracted from GP clinical systems).

The public and policy impact of this study largely depends not on the detailed findings but on the headlines selected for wider media. Thus “65% of patients called within one hour” could have been the main story, and a quite extraordinary one compared with the current experience of most patients in seeking help from their GP. “56% found it more convenient vs 22% less convenient” shows a huge net approval rating.

Instead the headline writers have concentrated on the guarded & weak finding that consulting time may have increased by 8% (highly variable between practices). This has morphed into “GP workload increased” without the caveats.
It is therefore of great importance to understand the source of this data, what it measures and what it can’t measure.
The measure is a total of “record open to record close” time for all GP consultations, in SystmOne practices only, as identified by GP Access (not by Doctor First). It is therefore from around 20% of practices in the study, and the rest of the data was “imputed”. It is not clear what period the change relates to, whether this was restricted close to the launch date of the new system, or much longer.

We supplied the data knowing that it was not powered to measure workload, as has been implied in the study and subsequent commentary.
Workload consists of many factors only one of which is measured here, and some of the key omissions are:
1. Visits are a significant part of the GP day, and have not been measured as we have no reliable method to do so. Yet consistently GPs tell us that visits have fallen, and they say sometimes by half.
2. DNAs fall, typically by 80%, and we do measure this with reasonable quality as recorded by the system. This was supplied to the study but has not been quoted. With typical DNA rates dropping from 6% to 1%, an increase in “consulting time” might be expected and positive.
3. Unmet need falls dramatically. Pre-launch we run a 1 week 100% audit of demand at reception, recording the response given. Of those who get through, the average is 14% turned way “Nothing available, call another day”. Post launch this figure drops dramatically as increased capacity means all or nearly all patients can be helped. From this alone, no change in workload would be experienced with a 14% increase in throughput.

This data was offered to the study but not accepted. In our view all the datasets we collect are relevant to the intervention.
We have not presumed to measure total workload because of the number of factors involved and difficulty of consistent and comparable results. It is highly unsafe for the study to comment on workload in totality from the limited data we supplied.
A more useful measure would be efficiency. This can be calculated from data supplied, in terms of telephone and face to face durations, and the “resolve rate” ie proportion resolved by telephone only. Better measures of efficiency would look at the whole system.

Clearly, if the method increases efficiency it should be widely adopted. Good understanding and measurement is also required: telephone duration longer than half face to face duration, coupled with a resolve rate below 50%, would clearly be less efficient.

It is striking that while headlines cause alarm in some quarters about workload, many GPs are increasing their telephone consultations or adopting a telephone first policy without any help from us. We don’t know how well they are performing. We can safely say that they are doing so without increasing their overall workload, or they would quickly stop.
The study certainly adds to our knowledge of the system. I accept that no effect on A&E was found, and while my small study of pioneer practices showed 20% reduction, we could not reproduce this at scale and we stopped making any such connection or claims in 2013. We have never made specific claims on cost savings for GPs. The study and commentary should make clear distinctions between companies making such claims and GP Access Ltd. They are not the same.

I call upon the authors
1. To revisit their data and conclusions on the above evidence.
2. To measure efficiency rather than workload.
3. To highlight the patient benefit in reduced waiting time to receive help (data supplied to the study)
4. To investigate the high variability they found in terms of workload, concentrating on those practices with the best performance so that best practice can be disseminated.

General practices are under severe pressure of workload, and recruitment is more difficult than ever. This does not have to be the case. We welcome independent investigation, but we need it to concentrate on what works and to provide answers.

Competing interests: Founder & Chief Executive, GP Access Ltd

13 November 2017
Harry J A Longman
Engineer
GP Access Ltd
Leicester