Clinical experience and previous studies have shown that unwanted incidents after discharge of patients from hospital to their homes are common, especially related to discrepancies in drug regimen [4
]. The present study has investigated the frequency and character of unwanted incidents observed by the AT during the transition of frail elderly patients from a geriatric hospital ward to their homes. It was shown that nearly 60% of the patients experienced unwanted incidents during transition and during the first four weeks after discharge. The most frequent incidents were mismatches between assistance appointed and assistance actually provided by the primary health care, errors regarding drug regimen and errors related to exchange of information.
Despite specific guidelines and accepted routines for transition from hospital to home, most patients in the present study experienced one or more unwanted incidents. This may have multiple explanations. Firstly, the study group was frail, of high age, often living alone, and having disabilities, including cognitive impairment, making them dependent of assistance in many daily life situations and vulnerable for experiencing incidents.
Secondly, although both the GEMU and the primary health care continuously work on improving communication within and between the systems, the large number of helpers around the patient requires that information about the patient is exchanged between different professions and employees several times and thereby increase the likelihood of unwanted incidents.
Thirdly, the skills and focus of the GEMU and the primary health care staff were different. The interdisciplinary staff of the GEMU emphasises optimal treatment for disorders precipitating somatic and psychiatric conditions, functional problems and the social situation of relevance for the actual hospital stay, while the main focus for the primary health care is to make it practicable and possible for the patient to stay at home as long as possible. Furthermore, the GEMU staffs was specialized in geriatric medicine while most of the employees in the primary health care had no specialised training in treating the actual patient group, many were unskilled. This may to some extent also explain why the patients did not receive help that was specifically appointed by the primary health care (Table ).
However, the study also revealed that some decisions made in the discharge-planning meeting in the GEMU were not relevant after discharge (Table ). Such errors were not registered for discharge-planning meetings arranged in the patients' home where also a representative from the home services who knew the patient beforehand was often present, demonstrating that the context of this meeting is important.
In concordance with other studies [5
] we found unwanted incidents related to drug regimen to occur frequently. We did not evaluate whether these errors constituted a risk for the patients' health, but according to clinical experience this could be the case in some occasions, while in other situations these errors probably had no immediate impact on health status.
As shown in Figure , GPs are not routinely involved in follow-up of the patient during or shortly after discharge from our hospital. Hansen et al. have shown that patients recently discharged from hospital are not medically stabilised [11
]. Their study uncovered problems during the first 3 weeks after discharge resulting in changes of medical or social treatment plans in the majority of patients. To further improve practice for these frail patients, a closer follow-up from the general practitioner shortly after discharge from hospital need to be highlighted. The length of hospital stay is constantly squeezed down, making the involvement of the general practitioner increasingly important.
It was shown that the patients in the present study during follow-up needed consultations from GPs, new hospital and nursing home admissions and one patient even died. However, the study was not designed to evaluate if these events were consequences of the unwanted incidents registered and thus could be prevented. In a controlled study it would be possible to evaluate if improvement in the transition from hospital to home could improve patient related outcomes.
Studies have shown that post-discharge visits in the patient's home by competent professionals can reveal important and potentially reversible clinical problems and unwanted incidents [13
] and there are indications that functional status improves after a time-limited supported discharge [26
]. The large number of incidents registered in the study could indicate that the AT in general should be more active in the management of the patients, such as following the patient home at discharge to make sure that assistance take place as appointed [43
], perform treatment in the patients homes that the home services are not competent to provide [31
], or working closely with the general practitioner during the transition to avoid medical incidents [31
Strength of our study was the comprehensive registration of a diversity of incidents related to the transitional process from hospital to the patients' homes and not only disagreements related to drug regimens. Most likely this has resulted in a higher number of incidents than registered in earlier studies. We argue that this broad perspective has given important knowledge about aspects of the patients' situation to further improved discharge planning and follow-up. Though, it is still not known, though to what extent the registered incidents in our study influence the patients' and the caregivers' quality of life and function, hospital readmissions, nursing home placement or death. This needs to be highlighted in future studies.
The study has some limitations. Firstly, the practice for the cooperation between GEMU, AT and the primary health care described in the present study is unique for the city of Trondheim and the external validity may be questioned. However, we believe that the challenges observed are recognisable also for frail patients in general when focusing on transitional care. Secondly, we were not able to find standardised questionnaires for registration of relevant unwanted incidents. Therefore an incidence form was developed and revised through a pilot study; however it was not tested for reliability. Filling in the form was based on consensus between the GEMU, the AT and the primary health care which should reduce the possibility for systematic bias. Thirdly, according to the pragmatic design of the study, the AT had a double role being involved both in the planning of the transitional care during the hospital stay, but also in uncovering and resolving unwanted incidents of importance for the patient's situation, as well as registering unwanted incidents. This might have reduced the number of incidents, indicating that the incidence rate could have been even higher than registered without an AT.