Diagnostic scope in out-of-hours primary care services in eight European countries: an observational study

We conducted an observational study based on retrospective data-collection from (computerised) medical records. The study was conducted by EurOOHnet, a European research network for out-of-hours primary health care http://​www.​euroohnet.​eu. Participants from the following countries participated in this study: Belgium, Denmark, Germany, the Netherlands, Norway, Slovenia, Spain, and Switzerland.

In each country, one to three regions with at least one out-of-hours primary care and emergency service were selected. Most settings were large-scale out-of-hours primary care settings (Additional file 1, Table S1). The exception was Switzerland, where no large-scale organisations for out-of-hours primary care existed.

In most regions, primary care providers (physicians and nurses) as well as hospital providers (physicians and nurses at A&E departments) were involved in providing out-of-hours care. In Germany the GPs were assisted by other physicians. In a number of regions the out-of-hours primary care setting delivered home visits. Different health care professionals performed telephone triage and advice such as GPs, nurses, and other health care workers. The opening hours were largely similar: all settings were open during weekends and on public holidays, although the exact opening hours varied slightly. The primary care settings were freely accessible to the general public.

All patients who contacted an out-of-hours primary care setting from the participating regions were included in the study population. We aimed to include patient contacts in the spring of 2009. First contacts of a disease episode were included, as were repeated contacts of different episodes of the same patients. Excluded were telephone stalkers and contacts with an administrative reason for encounter (e.g. prescription).

Data collection was the responsibility of the national researchers. We aimed to include 1000 patient contacts for each out-of-hours primary care setting. If the number of contacts exceeded 1000, a random selection procedure was performed. Measures included: gender and age of the patient (in predefined categories), patient symptoms and/or diagnosis by a professional. These symptoms and diagnoses were coded with the International Classification of Primary Care (ICPC)[16]. The ICPC allowed classification of symptom and disease diagnoses, using symptom codes and diagnosis codes. Accordingly, the term diagnosis in this paper referred to symptoms diagnoses as well as to disease diagnoses, as evaluated at the end of a contact by a professional.

ICPC coding was not routinely performed in all countries, and national researchers manually coded contacts afterwards with a symptom or disease diagnosis if necessary. Only ICD10 codes were available for the contacts of Slovenia and Spain. We recoded these ICD10 codes into ICPC codes, using a converter from Denmark http://​www.​dak-e.​dk/​icpc/​ and we checked this procedure using a converter from the WHO http://​apps.​who.​int/​classifications/​apps/​icd/​icd10online. Finally, possibly ambiguous contacts from Slovenia and Spain were discussed with national researchers. For detailed information on the final data set we refer to Table 1.

Patient contacts were the unit of analysis. As ICPC codes of professional diagnosis were present in all data sets and codes of patients' symptoms were frequently missing, we further analysed the codes of the diagnosis. Some countries had more than one ICPC code per contact. We added up all codes present in these data sets and calculated percentages using the total number of codes used (not the total number of contacts). For the Netherlands, Germany and Denmark, we observed respectively 20.6%, 27.6% and 8.5% missings in ICPC diagnosis. We therefore substituted these with ICPC symptom codes where possible. We stratified the main ICPC chapters for different age groups, because we expected diagnosis codes to be related to the age of patients.

In all out-of-hours primary care settings less than half of the contacts concerned male patients, with percentages varying from 33.7% in Germany to 48.3% in Denmark (Table 2). The mean age differed, from 34.1 years in Denmark up to 58.9 years in Switzerland.

The age distribution of patients per country showed a high consistency (Table 2). In general, there was a peak in the young age categories, followed by a decrease until the age group 18-24 years. In the next age groups, there was an increase again that slowly declined, except for the eldest age category, which showed another peak. The age distributions in Belgium, Denmark, the Netherlands and Norway were quite comparable. Evidently, differences were observed as well. The German and Swiss samples had fewer young patients and more elderly (over 74 years of age). In Slovenia, there were few children and for Spain the peak was between 18 and 44 years.

Table 3 presents the distribution of diagnoses by ICPC chapter. In general, in Denmark, Norway, and to some extent the Netherlands, we found very similar health problems presented during out-of-hours. Also, Germany and Switzerland were quite similar. The ICPC codes that were used most often were from the chapters 'general and unspecified'(A), 'respiratory'(R), and 'musculoskeletal' (L). Furthermore, 'skin' (S), and 'digestive' (D) chapters were frequently used. The chapter S contained codes for wounds and bruises. In Germany and Switzerland cardiovascular diseases were similarly coded (respectively 10.8% and 9.2%).

We also looked at the top five of ICPC chapters for the four main age categories (Additional file 1, Table S2). Again, the results were quite similar. Chapters 'general and unspecified' (A) and 'respiratory' (R) were in the top 5 for most countries and age categories, as were 'digestive' (D) and 'skin' (S). In the age group 0-17 years 'respiratory' (R) and 'general and unspecified' (A) were frequently present, as well as diagnosis related to ear problems (chapter H), although less frequently. In the middle age group (18-44 years) codes for 'musculoskeletal' (L), 'respiratory' (R), and 'skin' (S) problems were present in higher numbers. In the age group 65 years and older 'cardiovascular' diagnosis (K) entered the top five of ICPC chapters.

Additional file 1, table S3 shows the top 10 of ICPC codes used in each country. Again, we found consistency across countries. ICPC codes present in three or more countries included: acute upper respiratory infection (R74), cystitis/other urinary infection (U71), laceration/cut (S18), tonsillitis acuta (R76), acute bronchitis/bronchiolitis (R78), fever (A03), abdominal pain/cramps general (D01), infectious conjunctivitis (F70), gastroenteritis (D73), musculoskeletal injury (L81), and uncomplicated hypertension (K86). Belgium, Denmark, Norway, and the Netherlands had a largely comparable top ten. Finally, we focused on a set of ICPC codes related to emergency cases, occurring in out-of-hours primary care services. Life-threatening health problems such as acute myocardial infarction (K75) and cerebrovascular accident (K90) had low incidence figures (respectively 0 to 0.8% and 0 to 1.0%). The incidence ranges for transient cerebral ischemia (K89), acute alterations of conscience (A07, N07, N88) and severe infections like pneumonia (R81), appendicitis (D88), and pyelonefritis (U70) varied from 0% up to 2.2% for pneumonia. Overall, about one in 20 patient contacts dealt with potentially life-threatening problems.

Our study found a highly similar diagnostic scope in out-of-hours primary care services across different regions in eight European countries. Particularly regions of Denmark, the Netherlands, and Norway showed a high consistency. Acute life-threatening health problems had a low incidence in all regions. We found relatively high numbers of patients in out-of-hours primary care present with infectious diseases, such as respiratory and viral infections, or with acute pain related syndromes. Corresponding diagnosis codes were mainly from respiratory, musculoskeletal, skin, and digestive chapters. We found some differences between the regions, for instance related to the distribution of patient age and frequency of coding from chapter A ('general and unspecified').

The consistency of the diagnostic scope across regions was also found in primary care within office hours[12]. Interpretation of the findings was challenging, due to possible effects of differences in the coding process, characteristics of patient population, the health care system, and the study method used. Nevertheless, some general trends could be observed.

Regarding patient characteristics, our results were consistent with previous studies. More women had contacted out-of-hours primary care than men, who tended to visit A&E departments[1, 17, 18]. A high proportion of children attended the out-of-hours settings in most countries, a finding that was also observed in earlier studies[1, 18]. This could be the reason for the large number of contacts for infectious problems, which are highly prevalent in children. The relatively identical age distribution and ICPC codes of patients from Denmark, the Netherlands, and Norway suggests the similarity of the out-of-hours health care organisation as well as the role of the GP as a gate keeper[11, 9, 19]. Likewise the differences found between some regions of countries may be explained by variations in the health care organisation across Europe. For example, in Slovenia out-of-hours care for children in one of the observed settings is performed by primary care pediatricians,[20] whereas in Denmark, Norway, and the Netherlands these patients frequently visit GP cooperatives. Organisation of out-of-hours (primary) care and the role of primary care in general can be linked with patients' reasons for encounter, and subsequently diagnosis. If other organisational settings exist and are accessible out-of-hours (e.g. A&E department, specialists), this may influence the flow of patients. Stratifying the ranking of ICPC chapters for age showed that differences between regions could at least partly be explained by this variation in age distribution.

A consistent finding across countries was that the large majority of patients presented at primary out-of-hours care settings with non-acute, non-life-threatening health problems. Data on urgency assessment, which were available for six countries, supported this impression. Research on self-referring patients at A&E departments and GP cooperatives has shown similar results[1]. While previous research has found that some urgent health problems are overlooked in out-of-hours care[21]. We suggest that the large majority of patients' symptoms and diagnoses are not life-threatening and not urgent from a medical perspective. On the other hand, patients may perceive the presented health problem as urgent, potentially urgent (e.g. they feel incompetent to assess this), or urgent because of non-medical reasons (e.g. lack of time during office-hours)[17]. Despite the low incidence of life-threatening health problems, professionals in out-of-hours primary care should remain alert.

The high frequency of chapter A codes ('general and unspecified') might partly be related to a lack of specific coding by health care professionals. Furthermore, this could reflect an early stage of presented acute symptoms, such as A03 ('fever'), A77 ('viral disease other'), and non-diseases as death (A96). Also, it might be inherent to primary care, which has a higher probability of nonspecific complaints and diagnosis. Other chapters used frequently were 'musculoskeletal' and 'skin', which both contain injury related codes, such as wounds and bruises. These are one of the main reasons for seeking out-of-hours health care[14].

Some limitations of the study should be mentioned. Our aim was to include data of similar periods for all regions, in order to avoid seasonal effects. The contacts of Belgium and Switzerland occurred during the winter period, a fact that might have influenced the frequency of health problems presented, such as respiratory infection and fever. Furthermore, we included one to three regions per country. In some countries a regional variation in out-of-hours primary care organisation and population characteristics may be observed. Therefore, the selected region(s) might not be representative for the whole country. So, our comparison partly is of regions of eight different countries.

Some differences in the coding process had to be accepted, such as numbers of codes used per contact and per setting, individual coding decision in a particular case (such as choice for diagnosis codes instead of symptoms codes) and relation of coding with practice income. This might have influenced the content of the tables to some extent, but it is difficult to predict in what direction. In some regions ICPC were deduced from ICD10 codes, which may have induced information bias. In case of retrospective coding, the quality of the coding depended on the quality of the medical record of the out-of-hours service. The risk of information bias due to coding differences was reduced by clustering our main results in more general categories of the ICPC chapters. Data from the Netherlands showed similar patterns as in earlier research,[1] which suggests that the methods were valid. We primarily focused on professional diagnosis codes, but for three regions we substituted missings on professional diagnosis codes with patients' symptoms codes. This allowed us to keep as close as possible to the original data without exclusion, although this might have introduced some information bias. Often, the diagnosis as reported by the professional is from the same ICPC chapter as the patients' symptom. Also, coding of the professional diagnosis can be with a symptom code or a disease diagnosis code.

The similarity of diagnostic scope at out-of-hours primary care is important for comparisons of out-of-hours care across countries. Moreover, it stimulates international collaboration in clinical studies in this setting. For instance, studies on the use of antibiotics are warranted given the increasing numbers of resistant bacteria and the relation to antibiotics use[22]. In our study, the ICPC code U71 ('cystitis/urinary infection other') was frequently used and this subgroup could be analysed in epidemiologic cross national research focusing on actual clinical behaviour and the prescription of antibiotics[23, 24].

The high proportion of non-life-threatening health problems presented poses serious questions for policy makers, particularly in a time of economic challenges, an ageing population, and expected shortages of health care professionals. The trend towards larger organisations for out-of-hours care is unlikely to be reversed, but managing the increasing patient demand is a crucial challenge. A previous cross national survey showed that a large diversity of organisational models for out-of-hours care exists[11]. International studies can provide relevant information for policy makers in the ongoing discussion and the reforming of the organisation of out-of-hours primary care.