We studied 101 referrals from the Tromsø Casualty Clinic to the psychiatric wards at the University Hospital of Northern Norway and found that the hospital specialists accepted all referrals except one and mostly agreed with the casualty clinicians on diagnosis and upon the application of the law. The police was more often involved in the involuntary admissions than intended in the law. The proportion of patients with substance abuse was significant.
Data were collected retrospectively from written sources. As a naturalistic study, information and assessments made by the doctors may be missing. Information on substance abuse, the role of the police and if the patients had stopped taking their psychopharmacological medication could be underestimated. Information on social background factors was insufficient. One the strength of our naturalistic study is the simplicity of the health care system we studied; there was only one casualty clinic and one psychiatric hospital, and no private sector.
Agreement between referring doctors and hospital specialists
The processes leading to specialist care and how the GPs act as gatekeepers, vary between countries. A Dutch study documented that the number of self- referrals is much lower in areas with fully developed gatekeeper function . Two studies on pathways to psychiatric care, focusing on acute psychiatric referrals, both show that GPs' referrals are accepted more often than self- referrals [8, 17]. In Norway referrals to specialist care are obligatory. The personal list doctors (GP's) and the GP-based out-of-hours casualty clinics' doctors act as alternate gatekeepers for access to specialist care.
The doctors at the casualty clinic rarely were the patients' personal doctors, and the working conditions on the casualty clinic are not optimal for a psychiatric observation. Hence, some referrals from the casualty clinic could potentially be assessed as inappropriate at the in-hospital psychiatric examination. Interestingly, we found that all referrals but one from the GP-based Tromsø Casualty Clinic were admitted. We further expected that the referring doctors and the admitting specialists assessed the patients' condition differently, both regarding diagnosing and application of the law. Because the Norwegian Mental Health Act  allows a maximum time lag of 24 hours from admittance of the patient to verification of the legal basis, we anticipated that the possible time lag would cause a shift of the legal basis, because the patient would calm down after being hospitalized. However, we found that the legal basis was changed in only 15 % of the compulsory admissions and we consider the diagnostic agreement between the referring doctors and the specialist to be fairly good, especially for the psychotic patients.
A good gatekeeper should see to that all patients receive care at the right care level. We think the good agreement among referring doctors and specialists illustrates that the referrals were appropriate, and this may indicate that the casualty clinic doctors are good gatekeepers for acute psychiatric hospitalization. As almost all admissions were accepted we might further speculate that the casualty clinicians are too good gatekeepers and that too few patients are referred.
Nevertheless, to explore the appropriate threshold for referrals, and hence more thoroughly define the gatekeeper function, we need more information about patients not being referred. We also need information on other pathways to acute psychiatric admissions, especially referrals from personal doctors, to understand how the personal doctors handle acute psychiatric patients.
The proportion of referrals from the personal doctors was lower than those from the casualty clinic. The personal doctors can only be accessed one fourth of the 168 hours in a week. The personal doctors also have more available treatment and care strategies during daytime compared with the casualty clinicians, working evening and night shifts. Furthermore, factors leading to psychiatric emergencies such as substance abuse tend to escalate out-of-hours.
The role of the police
In our study the police was the legal representative in 52 out of 59 of the civil commitments. Seventy-five percent of the admissions took place when the municipality chief physician was inaccessible. As stated in the new Norwegian Mental Health Act  the police, as the social services and the prison administration, can request involuntarily hospitalization only if they are directly involved in the specific case. We find it reasonable that the police acted as legal representative when the police was directly involved in the referring and admitting process. In the rest of the involuntary admissions, the public health officer or her substitute would have been the natural choice of legal representative. Civil commitments procedures as they were practiced, raises questions about both legal safeguards and the quality of the professional assessment, since the police as legal representatives in our study had a passive role in most involuntary admissions. We expect that health professionals have a better basis than the police for decision-making in compulsory admission procedures. This would require a 24-hours on-call duty for public health officers, at least in cases where the police are not directly involved.
We found information on substance abuse in more than 40 percent of the admissions. The estimate is nevertheless probably too low. We expect that both the hospital's and the personal doctors' patient records would have given additional information.
Other studies indicate that substance abuse among psychiatric patients increases. A study from London demonstrated a two-fold increase of patients with drug problems from 1988 to1998, showing that 49 % of the patients had a history of substance misuse in 1998 . A British study found that 44 % of Common Mental Health Teams patients reported past-year problem drug use and/or harmful alcohol use .
In a Norwegian study , substance abuse was relevant for the psychiatric condition for 54 % of the patients. Information of a substance abuse history might influence the choice of treatment and the interpretation of the current symptoms of the patients, at least for persons with earlier psychotic episodes . Since comorbidity seems to be the norm rather than the exception, it is important to pay more attention to the patients' history of substance abuse. By performing a standardised interview most patients are found to give correct information on their history of substance abuse . This is supported by the British study, where hair and urine analysis did not add any significant information to self-reported data.
In our study, the proportion of admissions with a short length of stay was rather high and more admissions were first-time admissions than shown in another study from Northern Norway . As Hatfield et al, we found that most patients admitted were unmarried and lived alone without work . This proportion was higher than reported in another Norwegian study . Altogether this may indicate that more patients admitted from Tromsø Casualty Clinic than from other referring agents were a socially selected group with scarce social network.
Implications of the study
Characterizing acute admissions to a psychiatric hospital might help to develop appropriate service delivery models. Our study focuses on the most important pathway to acute psychiatric care in Norway and concludes that most referrals from the casualty clinic were accepted indicating that the casualty clinicians are efficient gatekeepers for the acute psychiatric departments. On the other hand we do not know whether some patients in need of acute psychiatric hospitalization are denied access to the right care level. Our study implicates that civil commitments procedures as they were practiced, raises questions about both legal safeguards and the quality of the professional assessment, and call for better procedures. We underline the importance of increased focus on the substance abuse of the acute psychiatric patients. In our opinion, better alternatives to admissions in an overcrowded psychiatric acute department should be developed. Observation and detoxification beds, better psychiatric out-patient care, including extended opening hours in weekends for the ambulant acute-team, may improve quality of care for some of the patients.