Main findings

The main finding of this RCT of DSH patients after discharge was that the prevalence of contact with the GP related to psychosocial problems was significantly higher in the intervention group. The patients were significantly more satisfied with the treatment in general and reported that the GPs listened to their personal problems and involved them in decisions. The findings may contribute to improved and more sustainable aftercare of DSH patients.

Two important aspects related to suicidal behaviour in general practice are availability in time of crisis, and referral to and follow-up of treatment from other health care services, especially related to psychosocial problems. To achieve this, it is important that the GPs devote sufficient time and are interested in the patients’ problems. The findings in this study indicate that these premises were achieved.

The high degree of satisfaction with the care received in general practice in both groups is important and underlines the role of the GP in aftercare of DSH patients, especially given the prevalence of patients that are referred to aftercare with their GP [20, 25].

This is supported by findings that most self-poisoning patients considered their GP to be the most important contact post-discharge [17]. Further, findings from an English study indicated that DSH patients found it helpful and were satisfied when they discussed emotional problems and factors that contributed to the act during follow-up consultations with their GP [25].

Among those who had thoughts about DSH, 54.1 % discussed them with the GP. In addition, 13.6 % discussed thoughts about DSH even if they did not report experiencing such feelings. We do not know the reason for this, but one possibility is that the GP raised the question, perhaps because the patient seemed to be depressed without having active DSH thoughts. These data indicate that the GPs were aware of and motivated to address suicidal issues before the study intervention. Even so, it turned out to be possible for GPs to offer more regular follow-up and address psychosocial issues to a greater extent, which was also associated with more satisfaction among the patients.

The prevalence of discussing suicidal ideation during follow-up would ideally have been higher, especially in the intervention group. It is possible, however, that the climate for discussing such thoughts is strengthened if the patients perceive that the GP has sufficient time and shows an interest in psychosocial problems. Further, a consistent and personal relationship with the GP as the gatekeeper facilitates the communication of suicidal ideation [27]. Further research should explore why patients in the intervention group did not communicate suicidal thoughts to the GP, despite indicating that they were more satisfied with the way the GP listened to their personal problems.

Strengths and weaknesses

To our knowledge, this is the first time that the impact of structured aftercare in general practice has been examined among DSH patients with satisfaction as one of the outcomes. Further, the KUHR registry data increased the reliability of the number of contacts made with GPs and the specification of whether the contacts were related to psychosocial or somatic diagnoses. The use of registry data also reduced answering bias, because the response rate was lower than expected. These data also enabled us to describe the extent and length of the contact with GPs, which to our knowledge has not been included in previous trials.

Another strength of this trial was the data provided by several health care services on the type and place of treatment for patients, thus enabling us to compare and more clearly differentiate treatment given in the intervention and the control groups. These comparisons have not often been provided in similar RCT studies, but have been requested in reviews [7, 28].

The trial was performed in wards in Oslo and Akershus County, which are specialised in psychosocial assessment and the referral of patients to appropriate treatment after self-poisoning. The suicide motive was assessed, which might have strengthened the inclusion criteria in line with the definition of DSP.

The blinding at the time of assignment prevented bias with regard to the usual treatment recommended in follow-up plans. However the GPs were not blinded, and therefore the comparison group with GPs that did not wanted to participate emerged.

The use of the validated EUROPEP scale also strengthened the reliability of the measured satisfaction with GPs.

Limitations

There are several limitations to the current study. First, similar to Bennewith’s trial [21], we did not obtain information from the GPs about whether they had actually used the guideline. The patients’ retrospective self-report containing information regarding their treatment and especially the communication of suicidal thoughts should also have been verified by the form sent to the GPs. Further studies should gather more detailed information about how the intervention was carried out.

Nevertheless, the finding that a considerable number of patients did not talk about their DSH thoughts indicates that the guideline was not used consistently. On the other hand, the GPs were encouraged to raise the question of suicidal ideation, to invite the patients to follow-up more actively and to address their compliance with other treatment. This may have helped the GPs to structure the consultations with these patients.

Second, the dropouts and self-reporting response rates during follow-up biased the validity, although there were no significant differences with regard to background variables at baseline. However, the socio-demographic profile was similar to previous studies [14, 29] and strengthens the external validity of the study.

The validity is also to some extent limited by the participation bias from the patients who declined to participate. Although their number was lower than in similar studies, the characteristics of the non-participants should preferably be described in future studies. However, this is difficult in Norway because usually, Regional Ethics Committees do not approve such studies. We did not obtain approval from the Ethics Committee to describe demographic and clinical factors in patients that did not consent to participate in the trial. In a Norwegian study of the same population, although with a different design, the researchers were allowed to obtain information about those who did not participate. They found no significant difference according to gender, but a higher age in the non-participation group [30].

Third, the findings cannot be generalised as applying to patients with higher risk of further suicidal behaviour, such as patients admitted to psychiatric inpatient care [29], those with no fixed abode [30], and DSH patients using violent methods [31–33].

Fourth, the number of participants in the two groups might have been more equal if we had adjusted for the non-participating GPs, although this was difficult to estimate beforehand.

Fifth, the GPs in the control group were not blinded. This might have contributed to performing bias as they could follow their patients more frequent than ususal. As such, the results from the comparison group would therefore be more valid, because the GPs that declined to participate in the pre-trial invitation were not informed again when one of the patient on their list later was included in the trial. However, the comparison group also consisted of GPs that were telephoned during the trial when a patient on their list was randomised into intervention and then declined. Further it can be questioned whether it is ethically correct to blind the participating GPs in a RCT, and therefore all the GPs in the current trial were given information and an opportunity to consent.

Finally, factors that might influence the outcome are not described. Attitudes and personal factors among the GPs that may affect the doctor–patient relationship, how the various problems are addressed, and whether the patients feel that the GP endeavours to understand the individual’s unique history and life circumstances, are all important. Sensitivity to such factors and the ability to handle interpersonally challenging encounters with patients may influence the outcome [34]. However, a comparison of satisfaction between the patients whose GP declined participation and the control group yielded no significant differences in satisfaction or prevalence of contact with the GP during the follow-up. Because all the GPs in the intervention group agreed to participate, it is difficult to specify the number of GPs in the control group who declined because we only knew the number of GPs who declined beforehand by responding negatively to the pre-trial invitation. Given the premise that the responding GPs were more inclined and dedicated to the treatment of suicidal patients, an additional analysis of the GPs in both groups who returned the form after 3 months was performed. No significant differences between the control and the intervention group were found in the outcome variables. However, the rather low sample size in outcome variables from self-reporting increases the risk of type 2 errors, whereas the outcomes from the registry data did not show any differences.

Findings in relation to other studies

The finding of higher satisfaction in the intervention group is consistent with results from trials in which patients received interpersonal therapy [12] and experimental social work services [35]. In another RCT, where Hawton and colleagues compared outpatient counselling with aftercare in general practice, there were no significant differences in repeated suicidal behaviour or with regard to whether the patients perceived the treatment as helpful [22].

Assertive outreach for poorly compliant patients may be a necessary component in maximising the delivery of any effective treatment [6]. In the present guidelines, suggestions were made for GPs to encourage the patient to follow planned treatment, although the dropout rates did not differ between the intervention and control groups. These findings are in contrast to previous trials with community follow-up of patients who did not attend outpatient appointments [18], and home treatment where substantially increased rates of treatment take-up were found [36].

Availability and assertive outreach after an episode of DSH have shown promising results in decreasing suicidal behaviour [37]. Diverging results were found in two trials of assertive outreach in Denmark; one quasi-experimental study showed a significant reduction in repeated suicidal acts [14], while the replicated randomised trial did not support these results [15]. These trials were based on the Baerum model from Norway, where suicide prevention teams followed patients after discharge from hospitals. However, no significant decline in repeated suicide attempts was found [13].

Bennewith and colleagues were moreover unable to demonstrate any reduction in repeated DSH in one British RCT in general practice, in which they sent a guideline to the GPs [21]. This may indicate that even though it is possible to improve follow-up from GPs, interventions that are more extensive may be necessary to reduce DSH. Results from the current trial where outcome measures are of suicidal symptoms and behaviour will be reported in another paper.

Clinical implications

The findings related to increased contact with GPs are consistent with one RCT [16] in which the researchers pointed out that their intervention was useful for patients who had never received psychiatric care before their index suicide attempt, and before they were offered contact with health professionals.

Patients that are not already in a treatment programme can be followed by their GP until they receive other scheduled treatment. Improvement is needed, as many patients do not receive follow-up [38, 39]. Active outreach is also important, because almost a third of patients did not report follow-up despite arrangements registered at the hospital [17]. More coordinated and frequent follow-up by the GP could probably reduce this gap in the chain of care. It is important that the hospital aims for a medical review to be sent as soon as possible after discharge, which has been an area in need of improvement [40].

In a study of self-poisoned patients, the waiting time for a follow-up appointment, for example at psychiatric outpatient units, was up to 3 weeks [17]. This is too long for patients in crisis and can be compensated for by consultations with the GP.

The average time from discharge to the first consultation in the intervention group in the current trial was 7 days. This is an advantage when accounting for the elevated risk of suicidal behaviour the first week after discharge [41].

There was a significant higher frequency of ICPC-2 diagnoses of psychiatric, social, or poisoning in the intervention group. The reason for this is not known. One possible explanation could be that these diagnoses were coupled to the planned consultations as part of the intervention. Another possible cause may be be, that the GPs in the intervention group were more aware and thoroughly in their assessments of psychosocial problems among their patients.

Future research

The need for clinical trials in suicide prevention is highlighted by the WHO’s Mental Health Action Plan 2013–20, where the Member States have committed to work towards reducing the rate of suicide in countries by 10 % by 2020 [5].

The need to target secondary prevention measures in the large group of hospitalised DSH patients should be emphasised and accompanied by financial support for performing and facilitating sufficiently powered multicentre and international cooperation clinical research whereby suicidal behaviour constitutes a primary outcome measure in line with the trial by Fleichmann et al. [42].

The need to provide sufficient health care services for this patient group frequently attending general practice is clear, but more research is neccessary to give firm recommendations. The multifaceted problems and high levels of psychiatric and physical morbidity, social deprivation and suicidal symptoms must be addressed, particularly to prevent the burden on the patients, their partners, children and other relatives. The reasons behind DSH are diverse, and include crisis reactions, severe depression or psychosis; tailored interventions for subgroups may make it easier to show the effect of intervention.