Accurate communication between primary care and hospital is essential to the continuity of patient care. Despite considerable National Health Service (NHS) investment in information technology in recent years, paper discharge summaries remain the principal method of communication between most general practitioners (GPs) and hospitals. Discharge summaries have been shown to contain various types of discrepancies, which have been linked with a number of possible risks to patients, including medication errors and adverse drug events. ^{Reference Boockvar, Liu, Goldstein, Nebeker, Siu and Fried1,Reference Morcos, Francis and Duggan2}

The National Institute for Health and Clinical Excellence (NICE) has produced relevant guidance on medicine reconciliation on discharge ³ and supporting medicine adherence. ⁴ Electronic systems may replace written summaries in the longer term. Studies of electronic prescribing systems have shown however that there are complex interactions between these and their users, and that this can lead to further errors where there is inadequate testing, training and evaluation of such systems. ^{Reference Miller, Gardner, Johnson and Hripcsak5}

We used a completed cycle of clinical audit to improve the accuracy of medication information provided in discharge summaries at the Caludon Centre, an in-patient psychiatric unit in Coventry, UK.

Method

The audit's objectives were to improve the recording of in-patient medication in discharge summaries and to ensure GPs were advised on how long to continue as required (PRN) sedative medications. We reviewed discharge summaries of 100 patients consecutively discharged from the general adult wards in the last 5 months of 2004. Results were analysed in SPSS version 12 for Windows. Admissions without any in-patient prescription and those without discharge summaries were discounted from the final analysis. Results were then fed back at a postgraduate educational meeting in 2005.

Prior to the audit, the to-take-out prescriptions form was A5 size, a triplicate, no-carbon-required set of three sheets, one for pharmacy, one for filing in the records and one for the GP. Due to its small size it was often not easy to read or find in the notes. The GP would usually wait for discharge summaries to be posted or call the centre to clarify the discharge medications.

Results

For both audit cycles we identified 100 discharges within 6 months prior to the audit. We discounted 17 discharges in 2005: 4 were without discharge summaries and 13 without in-patient prescriptions. In 2007 we discounted 14 discharges: 4 were without summaries, 5 were without in-patient prescriptions and 5 individuals were transferred within the trust to be discharged elsewhere. Results for the first objective of the audit, to improve the recording of in-patient medication in discharge summaries, are shown in Table 1.

Table 1

Medication recording in discharge summaries

Audit	2005, n (%)1	2007, n (%)1
Summaries reviewed	83 (83)	86 (86)
Medications recorded correctly	53 (64)	70 (83)
Missing one or more medications	13 (16)	7 (8)
Wrong dosages recorded	6 (7)	6 (7)

Results for the second objective, to ensure GPs are advised on PRN sedative medication, are shown in Table 2.

Table 2

Advice on as required (PRN) sedative prescribing

Audit	2005, n (%)1	2007, n (%)1
In-patients prescribed PRN sedative	15 (18)	3 (3)
Advice to general practitioner on continuation at discharge	5 (33)	0 (0)

Discussion

The audit achieved one of its two original objectives. The results suggest the introduction of a guideline for discharge summaries increased the accuracy of medication recording (from 64 to 83%) and decreased the number of medications omitted from discharge summaries (from 16 to 8%), but it did not alter the frequency of the recording of wrong dosages (7% in both audits). However, we noticed that at re-audit the mistakes were mostly on physical medications, especially the errors about dosages. A study by Morcos et al ^{Reference Morcos, Francis and Duggan2} reported errors in the transfer of medication information at some stage of a completed admission in 91% of cases reviewed. Our results, although not strictly comparable with that study, confirm the impression that such errors are relatively common in routine clinical practice.

The number of in-patient PRN sedative prescriptions and the frequency of the recording of advice to GPs over these prescriptions both declined sharply (from 18 to 3% and from 33 to 0% respectively). It is possible that the first audit highlighted the extent of PRN sedative prescribing and doctors were more vigilant following this. The findings of improvements in record keeping after the introduction of written guidance or protocols are consistent with other audits, which have shown similar effects for the completion of risk assessments. ^{Reference Masson, Liew, Taylor and McGuigan6}

Clinical audit can be used to improve communication between hospitals and GPs. It is, however, labour intensive and sustaining the improvement can prove difficult given constant medical and administrative staff turnover or absences when staff are on leave or off sick. Computer-generated discharge summaries distributed via an online system may provide a more sustainable solution. Pilot studies of such systems have reported positive results, ^{Reference Kirby, Barker, Fernando, Jose, Curtis and Goodchild7,Reference Scullard, Iqbal, White, Olla and Thomson8} but they still appear a distant prospect for many NHS trusts.

The audit reviewed a limited range of prescribing information recorded in hospital records but did not attempt to investigate the accuracy of these records. Using prescription records meant there was no ambiguity about medication histories. Nevertheless, it is possible that the results are in part the product of variations in the audit samples. For example, more in-patients had no medications prescribed or were prescribed PRN sedatives in the first audit cycle and more patients were subject to transfer from the wards in the second audit. Explanations for these variations are beyond the scope of the audit.