In the period following discharge from hospital, psychiatric patients are at high risk of readmission. Within the first 6 months, readmission occurs for between 20 and 40% of patients (Reference Caton, Koh and FleissCaton et al, 1985; Reference Boydell, Malcolmson and SikerbolBoydell et al, 1991). In selected groups of patients the figure is higher; over 50% of patients were readmitted within 6 months of a course of electroconvulsive therapy (Reference Robertson and EaglesRobertson & Eagles, 1997). The peak period of risk for readmission is within the first month (Reference Naji, Howie and CameronNaji et al, 1999). For long-stay psychiatric patients a similar pattern obtains, with likelihood of readmission exhibiting a decaying curve over time, albeit with a lower initial rate of returning to in-patient care (Reference Rothbard, Kuno and SchinnarRothbard et al, 1999).

Suicide data tend to mirror those for readmission. Rates of suicide are high in the year after discharge, notably within the first 28 days (Reference Goldacre, Seagroatt and HawtonGoldacre et al, 1993; Reference Geddes and JuszczakGeddes & Juszczak, 1995; Reference Geddes, Juszczak and O'BrienGeddes et al, 1997; Reference Sohlman and LehtinenSohlman & Lehtinen, 1999). Noting an increase in rates of post-discharge suicide among women from 1968 to 1992, Geddes and Juszczak (Reference Geddes and Juszczak1995) made a link with decreasing numbers of in-patient beds. The National Confidential Inquiry into Suicide and Homicide (Scottish Executive, 2001) found a peak of post-discharge suicides within the first 2 weeks, when 8% of all suicides by community psychiatric patients occurred. Eighty per cent of this group died before their first follow-up contact. These findings gave rise to the authors' recommendations that all patients should be followed up within 1 week of discharge (within 48 hours for patients ‘who have been at high risk’), and that discharge should be preceded routinely by a joint case review between in-patient and community teams, with this review including an assessment of risk.

Inter-professional communication

As reflected in the second confidential inquiry recommendation above, it is often held that poor communication, notably between healthcare professionals, is responsible for problems that arise around the time of discharge. Certainly, with respect to the communication that hospital specialists have with general practitioners (GPs), this criticism is probably well founded.

If GPs are to implement continuity and changes in care following admission then they require information, accurately and promptly, following a patient's discharge. Orrel and Greenberg (Reference Orrel and Greenberg1986) found that only 26% of GPs had received a brief communication about an in-patient stay within 2 weeks of discharge. While it is straightforward to tailor information to suit GPs' preferences by altering the format of the hand-written discharge letter (Reference Walker, Boehlhoff and EaglesWalker et al, 1998), this information still has to reach the GP. Once fears about confidentiality have been allayed, it is to be hoped that electronic transmission will usually be used. Meanwhile, we rely on patient transmission by hand. Although this can be augmented by posting a copy of the hand-written discharge summary (Reference Curran, Gilmore and BeringerCurran et al, 1992), patients are, perhaps surprisingly, usually quite reliable in relaying this letter to their GP (Reference Colledge, Smith and LewisColledge et al, 1992; Reference Naji, Howie and CameronNaji et al, 1999). GPs are keener to be telephoned about their patients at the time of discharge than hospital specialists might think (Reference SagarSagar, 1990; Reference Walker and EaglesWalker & Eagles, 1994).

Poor information transfer at discharge does appear to increase the likelihood of readmission (Reference Olfson and WalkupOlfson & Walkup, 1997) and one study found that, after discharge, an alarming 90% of elderly patients were receiving different medication regimes at home from those they had been prescribed in hospital (Reference Cochrane, Mandal and Ledger-ScottCochrane et al, 1992). Are efforts to improve communication helpful in a patient's post-discharge care? McInnes et al (Reference McInnes, Mira and Atkin1999) found that pre-discharge visits to the frail elderly improved GP-hospital collaboration, were associated with increased patient satisfaction and gave rise to greater use of community resources. It is perhaps doubtful that this would transfer cost-effectively to psychiatric settings. A randomised trial in Aberdeen (Reference Naji, Howie and CameronNaji et al, 1999) of standard discharge procedure v. a package of enhanced communication (GPs were telephoned; patients' appointments were arranged with GPs before discharge; discharge letters were posted as well as hand-delivered) indicated marginal benefit only. There was a trend towards lower rates of readmission and patients had more consultations about psychiatric issues with their GPs after discharge.

Clinical care at discharge

Various clinical interventions have sought to ameliorate patients' vulnerability in the post-discharge period. As with efforts to enhance inter-professional communications, there is little evidence that these have been successful.

The UK 700 trial recruited patients with psychosis, either at the time of discharge from hospital, or when living in the community but having been admitted during the preceding 2 years. The patients were randomly assigned to standard or intensive case management, the latter being similar in format to the Care Programme Approach. Intensive case management had no impact on suicidality (Reference Walsh, Harvey and WhiteWalsh et al, 2001), nor on either clinical status or social functioning (Reference Burns, Creed and FahyBurns et al, 1999). There was no impact on likelihood of readmission (Reference Burns, Creed and FahyBurns et al, 1999). However, a similar study in London (Reference Tyrer, Morgan and Van HornTyrer et al, 1995) found that closely monitored community-based patients spent significantly longer in hospital. Tyrer et al (Reference Tyrer, Morgan and Van Horn1995) did find that loss to follow-up was less common in the closely monitored group.

Few studies have focused more specifically on the post-discharge period. Sullivan and Bonovitz (Reference Sullivan and Bonovitz1981) found that subsequent out-patient attendance was improved by offering the first appointment within 3 days of discharge. A nurse discharge coordinator had no positive effect on readmission rates, on post-discharge well-being or on patient satisfaction ratings (Reference Walker, Naji and HowieWalker et al, 2000). As in Roy's (Reference Roy and Lester2001) recent review, there have been no intervention studies of representative cohorts of discharged patients to determine whether suicidality can be influenced. Psychological autopsy studies, with all their inherent flaws, can perhaps yield pointers towards clinical practices that may reduce suicidality. King et al (Reference King, Baldwin and Sinclair2001) found that discontinuity of contact was associated with post-discharge suicides in Wessex. However, rates of ‘key personnel on leave or leaving’ were said to be 1% in the control group and 5% in the suiciding patients. Given that the average consultant psychiatrist is on leave for some 15% of the time, this strongly suggests incomplete and selective recording.

Comments and conclusions

Currently, we know that psychiatric patients are vulnerable in the post-discharge period, but we have no good evidence to direct our efforts to improve the situation. Attempts to enhance inter-professional communication have the advantage of being very cheap (Reference Naji, Howie and CameronNaji et al, 1999), which probably makes them worth pursuing despite the tenuous evidence of effectiveness. The same cannot be said for clinical packages of care in the post-discharge period, such as the pre-discharge meetings and rapid follow-up espoused by the National Confidential Inquiry (Scottish Executive, 2001). As others have pointed out (Reference MarshallMarshall, 1996; Reference GeddesGeddes, 1999), it is probably premature to introduce such policies without an adequate evidence base. It seems much more logical to conduct good research studies to determine whether patients' vulnerability in the post-discharge period can indeed be ameliorated and to design appropriate policies thereafter.