Abstract
Mental illness is an important public health problem in Australia. It is common, with an adult 1-year prevalence for schizophrenia of 1% and major depression 3.7% [1]. The lifetime prevalence of schizophrenia is 1%, of depression 15.1% and of mania 0.9% [2]. Those affected are at increased risk of morbidity and mortality from various causes, particularly suicide, compared with the general population [3–4]. Many mental illness-related deaths occur in the period immediately after hospital discharge and are potentially preventable [3–4]. There is a large expenditure of resources in mental health care compounded by the high costs of lost productivity and patients' functional loss [5]. Consequently, the issue of mental health care is firmly on the Government's agenda [6–7].
Community care has largely replaced chronic hospital stay as active de-institutionalisation of patients has occurred, but the effectiveness of this change remains unclear. Groves comments that in Britain the ‘scandal of community care is that it has not been monitored adequately’ [8]. The effectiveness of community follow-up of patients with mental illness after hospital discharge has been poorly researched. Hospital readmissions have been used to monitor what happens after discharge [9–10], but with conflicting results; improved community care has both increased [11–12] and decreased hospital readmissions [13]. Few studies have considered patient wellbeing or quality of life when evaluating community mental health care. One study indicated that improved mental health decreased hospital admissions, improved employment and improved patient wellbeing [13]. Patients receiving intensive psychiatric community care after acute care hospital discharge in the United States achieved greater long-term clinical improvement than those who did not [14]. However, a British study concluded improving mental state seems in itself to be insufficient and that social interventions are likely to be crucial to improving social functioning for the seriously mentally ill [9].
Southern Tasmania (population 220 000) has 35 acute psychiatry beds at the Royal Hobart Hospital (RHH), and 156 long-stay beds elsewhere. De-institutionalisation is well advanced. There are four community mental health teams, each comprising a psychiatrist and other community workers, which cover defined geographical areas. Only about half of the patients discharged from the acute psychiatry unit maintain contact with formal psychiatric services [Jones I: unpublished data]. This is similar to UK data which indicate that over one-third of patients with long-term mental illness had no current contact with psychiatric services (although 90% had contact with their general practitioner [GP]) [15].
In this paper, we evaluate the current follow-up system for patients discharged from the acute psychiatry unit in southern Tasmania into the community, using the conceptual framework of an epidemiological surveillance system. ‘Epidemiological surveillance is the ongoing and systematic collection, analysis, and interpretation of health data in the process of describing and monitoring a health event, the information being used for planning, implementing, and evaluating public health action and programs’ [16]. We suggest that community follow-up fits ideally into this framework, with the health event being the mental health system's loss of contact with a patient with mental illness. Follow-up is designed to monitor patient health and wellbeing, and reduce adverse health events in an at-risk population. It is viewed as part of a healthcare service, and not as coercion of unwilling patients. We propose that this mental health surveillance system includes arrangement of patient follow-up appointments, use of discharge summaries to relay information, detection of appointment non-attenders, and re-engagement of non-attenders. The study focused specifically on short-term follow-up including attendance at the first follow-up appointment. We evaluated the surveillance system from the point of view of those implementing it (the acute psychiatry unit), and did not include service providers who were not included in the formulation of discharge and follow-up planning.
The specific aims of the study were to: (i) describe the policy, process, and current activities of the surveillance system; (ii) evaluate the content of and time delays in sending out discharge summaries; (iii) determine planned versus actual follow-up received by the cohort of patients; and (iv) evaluate the usefulness of the surveillance system in detecting and ‘re-engaging’ non-attenders.
Method
Prospective follow-up was performed on 100 patients consecutively discharged from the RHH acute psychiatry unit in 1995. From patient records, we collected basic demographic data, the patient's diagnosis [17], the source of patient referral, the patient's GP (if any), details of the recommended community follow-up, and discharge details. We did not contact patients.
For those patients who were given community follow-up appointments, the planned places of follow-up were contacted 3 months after discharge to determine whether the patient attended. We asked whether there was a mechanism in place at each clinic/practice to detect appointment non-attenders and whether any action was taken to re-engage patients into the system. For non-attenders, we recorded details of any re-engagement activity by interviewing the doctor involved and examining the clinic/practice records, and noted whether the activity was successful. For non-attenders and those patients not given follow up appointments, we contacted their GP, private psychiatrist or community clinic to determine whether they had used these or other healthcare services.
We also monitored hospital readmission data and mortality data for the 2 years after the initial 3-month follow-up period.
The policy framework was defined by analysing written policy documents about community surveillance of patients after discharge, and interviewing senior psychiatrists. The actual process of patient reengagement was evaluated against the formal surveillance policy.
The quantitative data were analysed using frequency distributions and, where appropriate, cross tabulations using SPSS for Windows 6.1 [18]. Logistic regression analysis was performed to determine which independent variables were related to patient readmission into hospital. Ethics approval was obtained from the RHH Ethics Committee.
Methods
Policy
There is no written policy about the follow-up of acute psychiatric inpatients post discharge. If patients fail to attend pre-arranged follow-up appointments, this is usually noted by the team or doctor involved. However, for those patients not given specific appointments, there is no mechanism to determine whether they attend. There is no policy which defines the action that should be taken to re-engage nonattenders into the system. Community teams are expected to use appropriate clinical judgement rather than a policy-driven decision.
There is no policy about the involvement of GPs or private psychiatrists in the follow-up system. None of the 34 GPs or 10 psychiatrists contacted had a system to monitor whether patients made their own follow-up appointments. Where appointments had been pre-arranged, only one psychiatrist actively contacted all non-attenders to make further appointments. The others said most patients ‘voted with their feet’ when they did not attend. In general, however, if a patient had a pre-arranged appointment and failed to attend, all the doctors involved would contact those patients ‘at risk’.
Policy states that discharge summaries should be hand written on the day of discharge, by the attending doctor, and a copy sent to the treating community psychiatrist(s) and the patient's GP. Discharge plans should be made in discussion with the community mental health teams or private psychiatrist, but the patient does not necessarily meet the community worker prior to discharge. There is no policy stating the GP should be contacted before discharge, and GPs are rarely involved in discharge plans. There is no policy about the role of family or carers in discharge arrangements. Follow-up arrangements should be recorded in the hospital notes and discharge letter, and given to the patient.
Patient characteristics
The cohort consisted of 35 men and 65 women, aged 15–87 years (mean 40, median 37). The most common diagnoses were depression (31), adjustment disorder (21), schizophrenia (15) and hypomania (10). Substance abuse was the main diagnosis in five patients and was considered to be a contributing factor in a further 10. Hospital stays ranged from 1 to 36 days (median 3).
The majority of patients (51) lived with others in a family relationship, 35 lived alone and 11 lived with others but not in a family relationship.
Frequency of source of initial referral to hospital and frequency with which the general practitioner's (GP) name is recorded in hospital notes
Interim discharge summaries
Twenty-seven summaries were dispatched on the day of discharge, 83 by 7 days, 96 by 14 days and all by 24 days. Only 35 were dated. The doctor writing the summary was identified in 90% of cases. Three patients visited their GPs before the summary arrived. Each of these three GPs commented that, although no patient came to harm, this had caused problems in patient management. The GPs had been unaware of the admission, follow-up arrangements, and discharge medications, and had to obtain this information directly from the hospital before providing appropriate care.
Community follow-up and hospital readmission
Community follow-up was deemed necessary for 97 patients, although 14 refused follow-up. Of the other 83, 25 were referred to their GP, 26 to a private psychiatrist, 26 to community mental health teams and six to other agencies. Only three patients were considered to not require follow-up.
A formal follow-up appointment (time and place) was recorded for 30 patients. Of these, 21 (70%) attended on time. Another three attended by 4 weeks. Three (out of four) patients who did not attend their appointment with the community teams were actioned and offered new appointments and all three subsequently attended. The other two non-attenders were not actioned for re-engagement. Thus, where formal follow-up was planned 27 of 30 (90%) of patients attended by 4 weeks.
Where patients were advised to seek follow-up (place advised but formal appointment not made), it occurred for 36 of 53 (68%) patients by 4 weeks. No more had attended by 3 months. None of the patients who did not attend was noted as a non-attender; thus none was actioned for re-engagement.
Of the 14 patients who refused follow-up eight (57%) visited their GP during the first 4 weeks after discharge. The others did not visit their GP, private psychiatrist or community clinic during the initial 3-month study period. The three patients who were considered not to require follow-up did not consult their GP, private psychiatrist or community clinic. Data were not available to determine whether patients had received care elsewhere.
During the first 3 months of the study, 71 patients had received community follow-up. Forty-two patients had been readmitted to hospital during this 3-month period. In the subsequent 2 years, 45 and 29 patients were admitted to hospital, respectively, each year.
Of the 23 patients who did not attend their formal or advised follow-up in the first 4 weeks after discharge, the surveillance system only picked up that four had not attended. Of these four, three were actioned and all three subsequently attended community follow-up.
Three months after discharge 11 patients had been lost to follow-up and had not received any medical care from the hospital, nominated GP, private psychiatrist or community clinic. Another 15 had not attended follow-up but remained in the system because they had been readmitted to hospital. Another seven patients returned into the mental health care system during the subsequent 2 years (six in the first year, one in the second) due to readmission. By the end of 2 years, three patients had died; none had died as a result of or in relation to their mental illness.
Not surprisingly, patients were more likely to attend for follow-up by 4 weeks if they were given a formal appointment rather than not given a formal appointment (relative risk=5.3 [95% confidence levels=1.4–29.6], X2=6.25, p=0.01). The presence of known substance abuse did not predict failure to attend for follow-up. Patients who were receiving parenteral medication were not more likely to attend for follow-up than those on oral or no medication.
Logistic regression analysis indicated that two variables were significantly related to patient readmission into hospital during the first 3 months of the study. Readmission was more likely if the patient lived alone: relative risk=2.04 (95% confidence limits=1.31–3.19), X2=8.34, p=0.004. Readmission was less likely if the patient attended for follow-up in the first month after discharge; relative risk=0.33 (95% confidence limits=0.17–0.64), X2=10.64, p=0.01. There was no association between patient diagnosis, or any other independent variable, and readmission.
Discussion
In this study we use the framework of a surveillance system to describe a system designed to decrease adverse health events in, and offer protection to, an at-risk population. We believe mental health services have a responsibility to offer patients re-engagement which is considerate and humane, but does not coerce or infringe human rights. Conceptually, we would argue that all patients discharged from hospital after an acute admission should have some follow-up, even if only by telephone, although of course patients are free to decline this offer. Previous researchers also have concluded that active re-engagement of psychiatric outpatient defaulters is required [19].
The surveillance system considered in this study failed to routinely detect and re-engage patients who did not attend for follow-up. However, for the few non-attenders who were detected, re-engagement generally succeeded. If non-attenders were identified, they could be contacted to determine whether further care is required. Currently, there is potential for patients to be lost to follow-up, only to reappear in a crisis. Given the diversity of patients and conditions treated in acute psychiatric services, it is likely that no single system of follow-up would be appropriate. A range of strategies may be needed including employing community nurses through Divisions of General Practice, creating discharge nurse positions in mental health (for liaison and planning) and discharge groups for patients and relatives.
Patients in our study were more likely to attend for follow-up if they were given a formal appointment rather than just being advised to seek follow-up, a finding consistent with previous research [20]. Perhaps if all patients were given formal appointments most would attend follow-up, and it would be easier to detect and re-engage non-attenders. Of interest, a diagnosis of substance abuse did not predict appointment non-attendance, and patients receiving parenteral medication were not more likely to attend for follow-up than those receiving other treatment.
There is some evidence that failure to attend follow-up or failure to engage patients is related to poorer health outcomes, both in mental health care and other areas of clinical medicine. For example, patients with serious mental illness who failed to attend follow-up were much more likely to need emergency services rather than a regular appointment as their next contact [21]. Similarly, a study of patients with diabetes implied that failed attendance at appointments increased patient morbidity [22], whereas appropriate follow-up and good treatment compliance were positively associated with remission in patients with asthma [23].
In our study, we did not seek evidence as to whether failure to attend follow-up resulted in adverse outcomes, as this was not a research question. We evaluated the mental health follow-up/surveillance system as it currently operated to determine whether those patients considered to require followup actually received it. If follow-up did not occur and was undetected, it suggests failure of the surveillance system and should be of concern to the clinician. In our study, it is possible that some patients had moved from the area or contacted other practitioners. However, where the system was unable to determine what had happened to patients who did not attend their follow-up appointments, it was, by definition, ineffective. These are limitations of the surveillance system rather than of the study.
Hospital readmission rates have been used to measure the success of community mental health care, with the implication that the higher the readmission rate the less successful the care [9–10]. However, in certain circumstances readmission may be an appropriate decision resulting in better patient care. In our study, 42% of patients were readmitted to hospital during the first 3 months after discharge. Possible explanations include failure of the surveillance system, inadequate services, and patient discharge before being fully able to manage in the community. We were unable to determine accurately whether the readmissions in our study were actually a part of planned management strategy. Our data suggest that patients living alone were more likely to be readmitted, perhaps because they have fewer social supports, manage less well in the community or are more severely ill. Of interest, patients who attended followup appointments in the first 4 weeks after discharge were less likely to be readmitted. Possibly those patients who do not attend follow-up are too unwell to do so and are subsequently re-hospitalised.
The source of initial referral to hospital for a third of patients was through DEM. Although most patients had a nominated GP, few patients were referred to hospital by their GP. Perhaps GPs play a minor role in the acute treatment of mental health ‘emergencies’ which require hospital admission.
Hospital policy about discharge summaries is consistent with the Australian Council on Healthcare Standards (ACHS) guidelines [24]. Discharge summaries should be completed for each patient at the time, or within 14 days, of separation. This occurred for 96% of our patients. The guidelines further state that discharge summaries must accompany patients returning to the care of another doctor; if this includes patients returning to their GP or a community psychiatrist, this occurs for only a minority of patients. Further, summaries should contain details of follow-up arrangements. Although details of the place of follow-up are recorded for most patients, appointment details (time and place) are recorded for less than half. It is likely that some patients who are expected to make their own appointments are too unwell to comply. Problems in patient management may occur if summaries do not arrive promptly or do not contain details of the recommended follow-up.
The study has some limitations. Our patient sample may not be typical of the rest of Australia. The number of appointment non-attenders was too small to allow statistical analysis of the re-engagement process. Although we contacted all known health care providers for each patient it is likely that we under-reported health care utilisation. The surveillance system was not sensitive enough to determine whether patients had accessed alternative healthcare providers. Ideally we could have asked patients and their relatives for additional information, but ethics constraints and the time frame of the study did not permit patient contact.
Conclusions
Community follow-up of patients with mental illness fits well into a surveillance system framework and can be appropriately evaluated using the framework for evaluating such systems.
The surveillance system under evaluation failed to detect and re-engage patients who did not attend for follow-up and is therefore not properly able to monitor patients after discharge.
There is no policy about the role of community practitioners in follow-up.
Appropriate policy exists about the content and timing of discharge summaries, but few summaries are sent out on the day of discharge.
Patients are more likely to attend for follow-up if they are given a formal appointment rather than just being advised to seek follow-up without a formal appointment.
Recommendations
Research, using appropriate outcome measures, is needed to determine whether increased community follow-up of patients with mental health problems improves patient outcomes.
If formal surveillance is appropriate, guidelines should be established including defined roles for community practitioners and agencies, and targets against which evaluation can occur.
A system should be developed which integrates hospital-, community- and practice-based expertise to allow detection and re-engagement (where appropriate) of appointment non-attenders.
Patients should be given specific appointment times after discharge because they are more likely to attend follow-up and it would be easier to detect and re-engage those who do not attend.
Footnotes
Acknowledgements
We would like to thank Dr Lin Arias and Professor Saxby Pridmore for their help and advice, and all the study participants. We also thank the Commonwealth Department of Human Services and Health for the General Practice Evaluation Program scholarship that supported this research.