We found that 57 % of the 255 admissions to the medical department had been made by primary doctors, distributed approximately equally between GPs and casualty clinics. Previous studies have shown a significantly smaller proportion of admissions from GPs. In a study conducted at St Olavs Hospital in week 23, summer 2003, Eikeland and collaborators found that only 12 % of the emergency admissions to the medical, surgical, orthopaedic and gynaecological departments had been made by a GP (5).
Our study was conducted during a regular working week and only included admissions to the medical department. As far as we are aware, no studies have yet investigated whether the proportion of admissions made by GPs has increased since the introduction of the Regular GP scheme in 2001.
Altogether 82 % of the patients were admitted to an inpatient ward; the others were assessed and treated in the reception and discharged the same day. In general, a complete assessment and treatment in the reception, with no use of an inpatient ward, may be an appropriate and effective treatment for many patients. Often, the patient is admitted because the primary doctor is concerned that an acute exacerbation may occur. Frequently these patients only need a clear diagnosis, a so-called diagnostic loop, and can be further followed up as outpatients or by the primary health service. However, establishing a full diagnosis in the emergency reception requires a high rate of staffing with skilled personnel, both in order to prioritise the patients correctly and to avoid excessive treatment time and general congestion in the emergency reception.
There is currently a debate unfolding over the organisation of the emergency receptions, widely recognising that increased resources may help achieve safer as well as more effective patient pathways (18–20). In our study, for example, we observed that some patients with deep-vein thrombosis completed their treatment in the reception according to an algorithm that has later become better defined. This has gradually become better known among the primary doctors in the hospital's catchment area. It is assumed that similar algorithms, for example pertaining to chest pain, also can help achieve more effective and safe patient pathways in the emergency reception.
Average hospitalisation time was 3.7 days. The corresponding figure for the entire year was 4.0 days (Einar Husebye, personal communication). Hospitalisation times tend to be shorter in Norway than in other European countries; according to an OECD report from 2010, they amounted to 4.5 days for all hospitalisations in Norway and 6.9 days in Europe as a whole (21). In general, hospitalisations are tending to decrease in length, in Norway as well as in virtually all other countries. This is primarily due to medical developments and pressure on beds and available resources as the number of admissions increases. Furthermore, primary health services are now required to receive patients who are ready for discharge at an earlier stage.
The diagnoses at admission corresponded relatively well with those at discharge in our study; this concurs with two large registrations previously undertaken in Norway (22, 23).
A little more than 20 % of the patients were admitted with a symptomatic diagnosis, while this figure had been approximately halved upon discharge.
In our review of the referral notes we found possible alternatives to hospitalisation for a little more than 13 % of the patients. In a retrospective review, an alternative to hospitalisation was found to be appropriate for only one-half of those for whom this was relevant upon admission. If 7 % is a realistic estimate of alternatives to admission, a correct identification of these patients would nevertheless help relieve the department. The difference between the assessments made upon admission and those made retrospectively after the hospitalisation period shows the degree of difficulty involved in identifying these patients in advance. This has also been the conclusion in other studies that have investigated opportunities for reducing the number of unnecessary admissions (5, 9).
In many studies, the necessity of admission has been assessed post hoc, in light of the results of the examination and treatment provided in the hospital (5, 9, 10). In our study, we have analysed the information available in advance as well as post hoc. We believe that in order to develop rational criteria for admission, it is essential to use the medical assessment made prospectively as a basis, and then retrospectively evaluate whether the sorting of the admissions is safe and medically sound.
We found little correspondence between the assessment of degrees of seriousness made in the reception immediately after admission and possible alternatives to it. This shows that triage tools are unsuitable for this type of sorting. Traditional triage tools are primarily intended to assess the degree of urgency at the pre-hospital stage and in emergency receptions, yet it is still surprising to see such a weak correlation between the triage and alternatives to admission.
It seems evident that primary doctors who need to consider referral cannot use tools that have low sensitivity. For example, if a primary doctor each year assesses 300 patients with acute conditions and 50 of these have a real need of hospitalisation, even a sensitivity of 90 % would mean that five patients would not be granted the admission they need. To keep this number as low as possible, the primary doctors will to some extent need to refer a number of patients for whom the further progression cannot be foreseen with any certainty. It is reasonable to assume that this is the reason why many hospital doctors feel that the primary doctors refer patients unnecessarily. If we use the same figures and assume that the primary doctors' decisions have a specificity of 90 %, the corresponding figure will be 70 admissions, whereof 25 will be unnecessary.
On the other hand, liberal referring practices on the part of the primary doctors will necessarily increase the burden on hospitals. It is a fact that an accumulation of patients, for example in the emergency reception, may give rise to an increased risk of medical error and increased mortality (24, 25).
We conclude that a minor proportion of the patients who are currently hospitalised (6–7 %) would have benefitted equally well from an alternative. However, these patients are difficult to identify at the time of admission. Since the number of admissions is steadily increasing, it seems decisively important to establish better routines and methods to identify those patients who ought to be hospitalised and those who could draw equal benefit from alternatives.