All patients hospitalised in the Departments of Internal Medicine, Pulmonary Medicine, and Cardiology (216 beds in all) were asked to do a symptom assessment by use of a slightly modified ESAS (Edmonton Symptom Assessment System) questionnaire [4], filled in by the patients themselves or by interview. All patients received written and oral information about the study and gave informed consent.

The ESAS questionnaire evaluates nine common symptoms in palliative care patients as well as general wellbeing [4-6]. Numerical rating scales (NRS) from 0 to 10 are used to describe symptom severity, 0 indicating no symptom and 10 indicating worst possible distress. The system has been widely used in palliative care settings [7]. In the present study, scores < 3 on the scales were defined as normal or no symptom present. Scores ≥ 3 were defined as symptom present, and scores ≥ 5 as symptom present and severe. These cut points have been used in similar studies [8-11] and are also in accordance with official Norwegian recommendations [4].

Ward staff recorded the following background information on the forms: age group, gender, main diagnoses (maximum three), and number of days in hospital.

Patients were excluded from the study if they were cognitively impaired or otherwise unfit to participate, if they had had surgery the preceding day, or if they were unwilling to participate.

Patients were defined as "palliative" if they had an advanced, serious, chronic disease with limited life expectancy and symptom relief as the main goal of treatment. This categorization was denoted by attending staff on the wards, nurses and/or physicians. In case of doubt, the staff was told to ask themselves the following question, "Would you be surprised if this patient died in the course of the next nine months?" If the answer was no, the patient could be defined as palliative.

In many cases it proved difficult for the staff to decide whether the patient should be regarded as "palliative" or not. Generally, the nurses and doctors were restrictive in their use of the term "palliative". Several patients were added to the palliative group when we asked our question "Would you be surprised if this patient died in the course of the next nine months?" When looking through the completed forms afterwards, we noted several patients whom we clearly would have defined as palliative, but who were not marked as such. However, in order not to introduce any bias, we did not alter anything. We therefore think that the number of palliative patients reported here is underestimated.

A large proportion of patients in the palliative group were excluded from the study due to cognitive impairment or because their general condition made them unfit to participate. This led to a reduced inclusion rate in the palliative subgroup. However, the high age and poor general condition of these patients add additional evidence that this group constituted a palliative care population.

In a recent review of symptom prevalence in advanced, life-threatening disease, the three symptoms pain, breathlessness, and fatigue were described among more than 50% of patients with advanced cancer, AIDS, heart disease, COPD, and renal disease [12]. This is consistent with our findings for dyspnea (70%) and fatigue (56%), which were the dominating symptoms in the present study. Solano and coworkers reported a pain prevalence of between 34% and 77% in heart disease and COPD in different studies [12]. Pain on movement was reported by 43% in our palliative group. The prevalence of depression in our study is also consistent with other reports [12].

According to Norwegian recommendations we used the value 3 on the numerical rating scales as cut-off point for symptom presence [4]. This is more restrictive than in several other studies and may influence the prevalence numbers [6,9,11].

This study demonstrates that patients in a Department of Internal Medicine have a high prevalence of distressing symptoms and that there consequently is a need for skills in symptom control on the ward. A thorough assessment is a prerequisite for successful symptom relief. We used a modified ESAS questionnaire in our study [4,5]. The ESAS is an example of a quick and clinically relevant symptom assessment tool. Although developed for palliative care cancer patients, the tool covers a range of symptoms which also commonly is present in advanced non-malignant disease [12]. Routine use of ESAS is a help to reveal new symptoms, prioritize between symptoms, and repeatedly evaluate the degree of symptom control obtained. However, symptom assessment is particularly challenging in some subgroups. These include cognitively impaired patients, patients whose language or culture differs from that of the health care professionals, and the imminently dying. The present study confirms that many patients in a palliative care population are too weak to use self-rating assessment tools, and that other approaches must be used. Family members and staff may be useful proxies for clinical and research purposes, but their data must be interpreted cautiously [11,13,14]. Numerous studies have demonstrated that observer and patient assessments are not highly correlated [11,13-15]. For non-verbal patients with severe cognitive impairment, pain assessment must rely on behavioural scales [16]. Although some observational tools exist, additional research is needed regarding instruments for symptom assessment in patients who cannot give subjective ratings.

Although the situation varies between the countries, the majority of palliative care services in Europe have cancer patients as their main target population [17]. In Norway, more than 90% of the patients in specialist palliative care are cancer patients [4]. However, our study reveals considerable palliative care needs in an inpatient population of a general medical department, representing an unselected, heterogeneous patient population. These findings clearly have implications for the planning and development of the palliative care services in our hospital. First of all, the attending staff on the wards must possess skills in palliative care to make a valid symptom assessment and provide basic symptom relief. In addition, there is a need for specialist palliative care providers with experience in non-cancer care to work in close cooperation with the ward staff. We also think that the present findings are representative of Internal Medicine Departments in many hospitals, both in Norway and abroad, and may even apply to Surgical services and Neurology Departments. In many hospitals, all of these departments mainly care for patients in the older age group, with advanced disease and significant comorbidities.

The present study only focused on symptom prevalence as measured by the ESAS, which is an obvious limitation with respect to needs assessment. Understanding needs outside of symptoms, like spiritual challenges, family and carer needs, and coordination of services, forms a natural part of palliative care and should be included in a complete survey. However, even if the evidence is limited, previous studies have shown that e.g. family anxiety and lack of psychological support are prevalent also in non-cancer [18]. Given the symptom burden, we do not anticipate that spiritual, social, or psychological needs are very different in a non-cancer population compared to a cancer population in the last year of life.