It is well established that patients admitted for attempted suicide or other kinds of deliberate self-harm have a significantly elevated risk of repetition and completion of suicidal behaviours, especially in the first year after discharge 1234. Among the few strategies that bear evidence of effectiveness in reducing this risk is a chain of care for these patients 56. A chain of care implies significant challenges, both with respect to delivering services to patients with multiple and complex problems and with respect to organising and coordinating various actors and care providers in services that are funded and organised differently and on different care levels. Many of these patients leave hospital without adequate suicide risk assessment after a suicide attempt 78 and they often do not attend or drop out of aftercare treatment 9. Although many experts in the mental healthcare field consider continuity of care and integrated care important, the research literature is weak with regard to the effectiveness of such interventions. A meta-analysis of studies on the effectiveness of community mental health teams found a clinically, but not statistically, significant reduction in deaths from suicide 10. Even if we draw on research literature from a broader area, such as psychiatric services or integrated care, irrespective of patient group, there is still little available systematic knowledge that could inform us about what a chain of care structure should contain, how the service should be organised and delivered and how sustainable it will be over time 11. The present study was designed to expand our knowledge on the factors at the hospital level that seem to be of particular importance for the maintenance of a chain of care structure at the municipality level by analysing longitudinal data on chain of care structures for suicide attempters in Norway.

In Norway, a main objective of the National Strategy for Suicide Prevention (1994) was to improve the quality and continuity of care for patients admitted to general hospitals following a suicide attempt 12. Supported by time limited governmental funding, about 30% of Norwegian general hospitals with an Emergency Department (ED)-implemented chain of care programmes for suicide attempt patients in the period from 1995 to 1999. Nearly all of the health services in Norway are publicly funded and these services are organised at 2 levels; hospital services are organised at the regional level (comprising 4 health regions with 55 general hospitals), and primary healthcare services (community health services) are provided at the municipality level (comprising 431 municipalities). This implies that a chain of care presupposes collaboration between a hospital and each of the many municipalities within the hospital's catchment area.

Based on experiences with the chain of care model and international recommendations 13141516, national quality standards for the clinical management of patients admitted to general hospitals after a suicide attempt were issued in 2001 by the Norwegian Board of Health 17. General hospitals with an ED were required to satisfy quality standards regarding suicide risk assessment, training of staff, coordination of monitoring and management of suicide attempting patients and procedures for systematic collaboration with aftercare providers (see Table 1 and 18 for details). The national standards do not clearly define what constitutes a chain of care within community health services (CHS) at the municipality level, but recommend that CHS has a team or a coordinator responsible for receiving referrals and interacting with the local hospital in rapid (within 1 week from discharge) and active aftercare. Follow-up usually consists of consultations, aimed at crisis intervention, with the individual and their family, pending other treatment to be established, and guidance and support in establishing or maintaining contact with other community based services (for example, health and social services, school) and/or specialist mental health services.

In three previous studies 181920 we have reported on assessments of the quality of hospital services provided for patients admitted after a suicide attempt. The results indicate that maintaining high-quality clinical services to suicide attempt patients seems dependent on clearly defined standards and routines described in written guidelines, a system to ensure that the guidelines are followed and training of staff with respect to clinical management and care of patients. These and other previous studies have been limited to the standards and services given at EDs in general hospitals and we have no systematic knowledge about the aftercare services provided in the CHS. In 2006, we therefore conducted an interview study of the CHS in a sample of randomly selected municipalities in Norway. The purposes of the present paper are: (a) to assess how common it was for Norwegian CHS to have a chain of care structure for suicide attempters in 2006, (b) to explore the capacity of quality indicators measured in 1999 and 2006 at the general hospital EDs to predict the presence of a chain of care structure in the CHS in 2006, and (c) to explore whether hospitals that had implemented a time-limited chain of care programme between 1995 and 1999 were more likely to have municipalities with a chain of care structure in their catchment area in 2006.

Although all municipalities in Norway are supposed to have a chain of care, only one-third of the CHS in this representative sample of municipalities reported that they in fact had such a structure. The observed discrepancy between recommendations given in national guidelines and the reality described is much in line with international research on general hospital services for suicide attempters. Studies from other countries have shown that a large proportion of self-harm patients attending accident and emergency departments do not receive a psychosocial assessment 726. Furthermore, substantial shortcomings have been found in staff training, interdisciplinary working and service planning, and availability of all-hours specialist psychosocial assessment 272829.

Several explanations could apply to the observed discrepancy between policy and reality; the number of suicide attempts treated in general hospitals each year may be low in small municipalities/rural areas and this may negatively influence their ability to maintain services and routines over time. On a more general level, factors such as organisational changes, reforms, downsizing, change of leadership or loss of key personnel may contribute to the difficulty of establishing and maintaining collaboration between different levels of service provision over time. The lack of interventions that show evidence of effectiveness in preventing suicide after a suicide attempt 30 may also explain suboptimal collaboration between ED's and CHS. However, several previous studies suggest that shortcomings in hospital services provided to suicide attempters are associated with an increased risk of repeated suicidal behaviour 263132. Furthermore, an ecological study by Cooper and colleagues 33 suggests that the availability of mental health resources (mental health treatment, crisis treatment, case management) in the county of residence is associated with a reduction of suicidal behaviour in the year following a suicide attempt.

Stability over time on quality indicators at the EDs was the best predictor of whether or not community healthcare services had a chain of care structure in 2006, both on the level of single quality indicators and in the multivariate analyses. This study highlights several factors that could potentially contribute to a chain of care structure in the CHS in the hospital's catchment area. Maintaining a structured collaboration with aftercare providers and having a team or a coordinator in the hospital over time were both indicators associated with a relatively high prediction of the presence or absence of a chain of care structure in CHS in a representative sample of municipalities. A significantly higher likelihood of having a chain of care structure was also observed for CHS collaborating with hospitals that had written guidelines with a quality assurance system, training and systematic supervision of staff and routinely gave patients information about available aftercare providers/services during discharge in both 1999 and 2006. One common denominator of these variables is that they require the hospital to allocate human resources to the maintenance and implementation of local guidelines and practices both within the hospital system and towards aftercare providers.

Having implemented a chain of care programme at general hospitals in 1995 to 1999 significantly predicted the presence of a CCS in CHS 7-10 years later. This finding is in line with the results from another study conducted at our centre: hospitals that had implemented a chain of care programme in the latter half of the 1990s had standards and routines more in accordance with the recommendations than hospitals without such programmes in both 1999 and 2006 18. The chain of care programmes were of limited duration and funding. Even though an observational study such as the present work cannot directly measure the effectiveness of a chain of care programme, it is nevertheless noteworthy to observe that this limited intervention was associated with higher standards of care at another level of service provision as much as a decade later.

The hospital study was conducted at two time points in a nationwide fashion with remarkably high response rates. The sample size in the CHS study was relatively small. However, in light of the selection procedure and the fact that the CHS in the municipalities actually covered one-third of the entire Norwegian general population, we believe our findings are reasonably generalisable to the CHS in other parts of Norway and may have relevance to CHS in many developed countries. Due to low statistical power, there could be other variables that are important but did not reach statistical significance. The results reported in this paper have considerable overlap with the findings from the longitudinal nationwide survey of hospitals in Norway in 1999 and 2006 18, which supports the associations observed. The main limitations of this study lay inherent within the study design and the limited conclusions that can be drawn from an observational study based on self-reported data gathered from different levels of service provision. We do not know if the informants' accounts of the quality of care delivered to patients correspond with the patients' points of view. Nevertheless, we still believe this type of research may provide valuable information to advise policy makers and clinicians in their implementation and prioritisation of different types of intervention and routines for management and follow-up of suicide attempters.

This study has measured selected structural elements in the CHS as indicators of the presence of a chain of care for patients discharged from general hospitals following a suicide attempt. There are a number of other ways the chain of care model could be defined or operationalised. Selecting structural elements as indicators of a chain of care model is one way of managing the considerable heterogeneity in population size, population density and organisation and content of the services delivered across municipalities and service providers.

Is this what is needed to strengthen the continuity of care for persons with suicide attempts? Currently, we do not know what system of aftercare service provision may best serve this mixed patient group. It could be that the patients will be best served by anchoring the chain of care in the general practitioner/family doctor. However, it is not clear that all general practitioners will have the interest or ability to provide active follow-up of these patients.

The provision of aftercare services for patients treated after a suicide attempt at the community level is markedly below the recommendations of the national standards in Norway. It is likely that this has a negative effect on the patients' prognosis. The results of this study suggest that interventions aimed at establishing and maintaining structured collaboration with aftercare providers and a team or coordinator at the hospital level might be important aspects in fostering the presence of a CCS in the CHS. Regular training and supervision of staff in the assessment and psychosocial care and aftercare of patients admitted to EDs of general hospitals following a suicide attempt, and establishing local guidelines with a system for quality assurance, also seem to be important factors in maintaining a chain of care model. More thorough studies addressing programme content, structural elements, theoretical underpinnings and the effects of different chain of care programmes are needed to get a further grasp of what works and for whom. This study represents a first step giving a bird's eye view.

In a national context, clinical standards and practices should be monitored and evaluated against national recommendations at regular intervals. Because most of the healthcare system in Norway and some other developed countries are publicly funded, the health authorities can and should take advantage of this fact in demanding structured collaboration between the different levels of service provision to strengthen the continuity of care.

The authors declare that they have no competing interests.

LM designed the study. EM, LM, EAF and IR wrote the protocol. EM drafted the first manuscript and performed the statistical analysis. EM and EAF participated in the acquisition of data. All authors participated in the interpretation of data, revised it critically for important intellectual content and have read and approved the final manuscript.