The increase in the number of elderly people in industrialized countries is generally considered to result in higher health services usage and costs. In Germany, the absolute number of inhabitants 65 years old and more will increase by 41% between 2010 and 2040, and the ratio of elderly people to every 100 persons in the age group 20 to 64 years ("old-age dependency ratio") is even likely to increase by 82% in the same period due to a decreasing birth-death ratio. The number of people 80 years old and older will increase from current 4 million to 10 million by 2050 [1].

It is customary to consider age and concurrent multimorbidity as driving forces in the growing utilization of health services [2–6]. At first glance, the interrelation of old age, morbidity, demand and utilization of health services seems clear. Scientific studies, however, have come to less concordant results. In a review on predictors of health care utilization in the chronically ill, De Boer et al. found ambiguous results with regard to age and gender as predictors of the frequency of physician visits and the degree of hospital utilization [7]. Results presented in the German literature have also been non-conclusive. Bergmann et al. found a curvilinear relationship between age and frequency of physician contact [4], whereas Hessel et al. found no effects of age and gender on the frequency of physician utilization [8]. Laux et al. described an impact on utilization for age but less for gender [9]. Winter et al. saw age as more important than morbidity in explaining ambulatory medical care utilization costs [3], whereas Siegrist declared morbidity patterns to be the major factor responsible for utilization differences between sexes and age-groups [10]. Also, Wiesner et al. described a larger effect of the number of chronic conditions than of age on utilization [11]. As for gender itself, most studies have found a higher utilization pattern in women [12–14], whereas others found no relationship [8]. Coenen et al. also found gender differences in healthcare utilization but stressed the point that such differences were (mostly) related to study design [13].

The manifold inconsistencies in results may be due to differences in the populations studied, study designs, data collection techniques and data analysis methods. For example, many studies on utilization related to age and gender are based on survey data with a recall-biased a posteriori investigation. Also, most studies compare utilization patterns in the study population over a long range of age, finding a higher utilization in the elderly as a group but paying less attention to developments within the older age group [3, 6]. Finally, the effects of multimorbidity are often not taken into account.

In order to interpret the data on utilization in Germany and the conclusions of this study, a brief introduction to the medical services structure in Germany is provided. Today, health insurance is mandatory for the entire population in Germany, either within a statutory health insurance scheme of the "Bismarckian" type or - for about 10% of the population - a private one. The number of health care professionals - physicians, nurses, physiotherapists etc. - per person is among the highest in Europe [15]. Ambulatory and hospital care are separate sectors within the health care system. As a result, hospital physicians are usually not allowed to work simultaneously in ambulatory care and vice versa. Parallel to the hospital sector, a dense network of ambulatory primary and specialist care exists, mainly provided by physicians in solo or small group offices. These physicians also care for patients in protected living institutions and nursing homes. The group of primary care physicians (PCPs) consists of general practitioners (GPs), all pediatricians and those internists opting to work as a PCP at the moment of opening their practice. The latter account for 60% of all internists. The average density of physicians in 2004 was 1:1,433 population for PCPs and 1:1,218 for specialists [16]. In metropolitan areas the density of specialists is nowadays already much higher than for PCPs.

All physicians function as entre preneurs in a market characterized by free access to all medical disciplines without compulsory gatekeeping. As a result, there are neither rules for the referral to the specialist nor for the re-referral to the GP. In daily life, however, the elderly heavily rely on their PCP, also for referrals. Since 2004, going to see a specialist as the fist physician in a quarter without a referral from the PCP is "punished" by an additional co-payment of 10€.

In order to estimate the future demands for health services, the analysis of current utilization patterns of the elderly and their determinants is crucial. The aim of this study is to analyze ambulatory medical care utilization by patients aged 65 and over in relation to age, gender, number of chronic conditions, individual multimorbidity patterns and nursing dependency in order to determine the influence of these supposedly influential factors.

The study is based on an unselected primary care population consisting of all members aged 65 and over (n = 123,224) of a statutory health insurance company operating nationwide in Germany, the Gmünder ErsatzKasse (GEK) in 2004. The GEK insured 1.7 million persons, a figure corresponding to approximately 2.4% of the statutorily insured population. Originally, the GEK primarily insured craftsmen, and the proportion of insured men therefore still exceeds that of women even today. Previous studies have shown that results from the GEK database can be transferred to the German population as a whole if age and gender adjusted [17]. The claims data stem from physicians in the ambulatory medical care sector only. They were provided by GEK in a pseudonymous form. The claims are forwarded on a quarterly basis from the individual physician to the regional physicians' panel association, checked there for comprehensiveness and plausibility and then transferred to the insurances. The observation period of this study was 12 months between January 1 and December 31, 2004.

We selected the most frequent chronic medical conditions in GP surgeries as described in a panel survey ("ADT-Panel") of the Central Research Institute of Statutory Ambulatory Health Care in Germany [18]. Chronicity of conditions was assessed using the "Expert Report for the Selection of 50 to 80 Diseases to be Included in the Morbidity Based Risk Adjustment Scheme" in the German Statutory Health Insurance [19]. In order to capture a comprehensive picture of the disease patterns in individual patients we amended this list for all chronic conditions with a prevalence ≥ 1% in the age group ≥ 65 years in the data set of the Gmünder ErsatzKasse in 2006. The ICD10 codes related to the individual chronic conditions were grouped by an expert panel of family physicians of the Hamburg Institute of Primary Medical Care in order to account for coding variance for the same syndrome among physicians. For example, F00-F03, F05.1, G30, G31 and R54 were grouped under the heading "dementia". The result of this procedure was a list of 46 single codes and code groups further referred to as chronic conditions in this paper. This list includes all frequent somatic and psychological chronic disorders (see additional file 1).

A person was defined as chronically ill if she/he had at least one of the 46 chronic conditions in at least three quarters within the one-year observation period 2004. The three-quarters criterion was chosen in order to avoid transitory or erroneous diagnoses, a usual procedure in using health insurance claims data for research in Germany. Also, acute or sub-acute forms of certain conditions were to be largely excluded by using this criterion. Among the chronically ill, a person was considered as multimorbid when he/she had 3 or more chronic conditions from the list. The criterion of three chronic conditions minimum was considered to be a more valid cut-off score for multimorbidity in elderly patients treated in the ambulatory care setting instead of the usual two chronic conditions criterion, frequently leading to very high rates of multimorbidity in the age group over 65 years [20], as in our data set as well. Recent research supports using the criterion of ≥ 3 conditions for investigations of multimorbidity in the ambulatory care setting, especially when the aim is to compare the multimorbid sample and the non-multimorbid sample (further abbreviated as mm-sample and nmm-sample) in the study population [21, 22]. On the basis of this criterion, the total study population of 123,24 persons was divided in a multimorbid sample (abbreviated further as mm-sample) of 76,540 (62%) and a non-multimorbid sample (abbreviated further as nmm-sample) of 46,684 persons (38%). Details on the prevalence of chronic conditions in the study population and the two samples are given in a previous publication [23]. Within the mm-sample, the prevalence of chronic conditions ranged mainly between 3 and 15. We analyzed the effects of the presence of every single chronic condition and of their most frequent triadic combinations ("multimorbidity patterns" [23]). We analyzed the influence of age (available: year of birth), gender, number of chronic conditions (based on the list of 46 described above) and statutory nursing dependency a) on the number of contacts with ambulatory care physician practices per year and b) on the number of different physicians contacted within the year. Statutory nursing dependency is given when a patient receives services from a statutory nursing insurance fund, a parallel agency to the statutory health services insurance scheme. Receiving services from the statutory nursing insurance is used as a proxy for disability in this study, but it should be understood that these figures underestimate the prevalence of disability as disabilities with no or little impact on Activities of Daily Life ("ADL") usually do not lead to receiving benefits from the Insurance.

As for the term "contact", it should be understood that a contact does not necessarily imply an intensive consultation with the physician. Contacts are also needed for prescription renewals or other administrative acts, and in such cases the real contact with the physician may be very short or even absent. This is why we preferred the term contact over terms like visit or consultation.

For every single contact with a physician the database allows to discern if it took place on the basis of a referral document issued by another physician (the primary care physician in the first place) or not.

The main statistical analyses consist of multidimensional frequency tables with the corresponding percentages. Arithmetic means, standard deviations, and confidence intervals were conducted on all continuous outcomes. A linear regression model was performed for the study population (n = 123,224) in order to determine the independent effect of multimorbidity and other factors on physician utilization. We included age, gender, nursing dependency (yes/no), and the number of chronic conditions. Due to the non-linearity of the connection between the frequency of physician contact and the number of chronic conditions, we found an adequate adjustment by adding a non-linear term (logarithm of the number of chronic conditions) to the model.

All analyses were made with the SAS statistical software (Version 9.2) and SPSS 16. Figures were created using R (version 2.12.0) and Windows Microsoft Excel 2003.

The study was approved by the Ethics Committee of the Medical Association of Hamburg (approval no. PV3057).

The aim of this study was to analyze ambulatory medical care utilization by patients aged 65 and over in relation to chronic diseases and multimorbidity in Germany. We found a high rate of utilization of ambulatory medical care services by the elderly. Being multimorbid corresponded to more than the double of contacts compared to the non-multimorbid sample. On average, multimorbid elderly persons had 36 contacts with physician practices per year. Some 20% of the multimorbid elderly had one contact with a physician practice every week. In general, both age differences among the elderly and sex had at best a small influence on the number of contacts with physicians and the number of contacted medical professionals in the mm-sample. The number of contacts per year varied largely according to individual chronic conditions and to their combinations in the multimorbidity patterns. Chronic conditions leading to nursing dependency was the most important factor related to high utilization rates.

It is customary to pretend an almost inevitable increase of utilization of medical services due to the growing number of elderly people in the population worldwide. In our study, however, the bivariate analysis showed only a very moderate increase in the number of contacts (15%) between the youngest and the oldest age group in the mm-sample, and this effect disappeared when controlled for other factors. The number of physicians contacted even decreased among the oldest old, especially in females. These results contradict the thesis on the association of growing old and increasing use of medical services. Multiple reasons may explain this non-association. It may be that people become older without a parallel increase of the number of chronic diseases and/or of disease burden and/or complications, e.g. due to earlier diagnosis, success of treatment and/or secondary prevention. This hypothesis is known as compression of morbidity, a thesis still under discussion [24]. There is still no scientific proof that an increasing longevity is associated with a shortened period of morbidity and/or disability, but many studies point in this direction [25]. In our study, the difference in the number of chronic conditions between the youngest and the oldest age group was only 1, although the difference in average age between the youngest and the oldest age group was 17 years [23]. The demand for services by the elderly also depends on the social setting (socially integrated or living alone) or the living conditions of the patients (in the community or in the nursing home). Also, explicit rationing and/or silent age discrimination by professionals might contribute to the cessation of the increase of utilization in the oldest old. On the other hand, the absence of an age-related increase in utilization may also be due to the one-year time span of this study, since frequent utilizers may die at an earlier age. Here, analyses of utilization covering several years are needed. As for research on utilization by patients and for care supply planning, the potential survival phenomenon described above is irrelevant as the object of health services research is the real population under actual care conditions.

In contrast to many studies [12, 13, 26, 27], we also did not find clear-cut signs of higher utilization patterns in the multimorbid female elderly compared to men. In the bivariate analysis, the difference in visit numbers was small for all age groups, whereas the number of different physicians contacted was higher for multimorbid women in the younger age groups but lower in the older groups. Our data suggest that the number of specialists seen by multimorbid patients decreases with age in female, but not in male patients, except for the oldest old. Of course, a study covering a 12 months time span cannot definitively prove the asserted non-associations, for which reason observations over several years will follow. The small effect of gender was confirmed in the regression analysis.

The positive association between the number of chronic conditions and the two utilization indicators in this study was not linear. Instead, both indicators showed the relatively largest increase from 0 to 1 chronic condition. In other words, the increase in utilization started already with the appearance of the first chronic condition and grows relatively consistently with every further number. The number of chronic conditions had a great influence on utilization in bivariate analyses, but its effect also became unspectacular when controlling for other factors as every additional chronic condition increased the number of contacts by 2.3 per year only. The only really important factor for high utilization was ADL-related disability, as expressed by the reception of services from the statutory nursing insurance system. Nursing dependency led to a small decrease in the number of physicians contacted but to 10 more visits to physicians per year. This association underscores the idea that primary and/or secondary prevention of disability might lead to a reduction of the utilization of ambulatory medical services.

The increase in utilization of ambulatory medical care due to chronic diseases and mutlimorbidity found in this study confirms the results of many other studies. In a comprehensive review of the literature by Gijsen et al., the authors concluded that "comorbidity was consistently related to health care utilization (costs, length of hospital stay, and number of physician visits)" [28]. Also, the German Robert-Koch-Institute study based on survey methods found a doubling of the number of physician contacts in ambulatory care within the multimorbid population under 75 years in Germany [11].

The number of contacts and of contacted physicians also varied with the presence of individual chronic conditions and of their (triadic) combinations in the multimorbidity patterns of the patients. The number of contacts varied between 35 and 54 per year and the number of contacted physicians between 5 and 7 depending on the presence of individual chronic conditions. Similar variances are also found for patterns which included the 50 most frequent triadic combinations (maximum 48 and minimum 36 contacts). Further research is needed to explain these variances. At first glance, the number of contacts is not related to the question if the diagnoses are related to one or several specialist disciplines ("morbidity mix"). Neither is there an obvious association between the number of contacts and the number of contacted physicians. In any case, the data suggest that the utilization figures are independent of the prevalence of the chronic conditions and their combinations.

With an average of 36/year, the number of contacts of multimorbid elderly with physicians in ambulatory care is likely to be far greater in Germany than in the health care systems of other countries. In the Netherlands, the number of contacts per year with physicians in ambulatory care is around 11 in patients with chronic disease(s), 5 of them with specialists [29]. Nie et al. reported 10 visits to ambulatory care physicians in Ontario in 2005/06 [6]. For the United States of America, Starfield reported 15.5 visits (6.6 visits to the PCP and 9.0 visits to specialists) in the sample with the highest multimorbidity burden in a Medicare sample aged 65 years and older in 1999 [30]. Also, the average number of contacted specialists (estimated 4,7 for Germany) and the proportion of the multimorbid population referred to specialists (93%) is obviously higher than in other countries. Forrest et al. found that 14% of the population in the United Kingdom and 30 to 37% in managed care plans in the United States of America were referred to specialists in 2002 [31]. The percentage of elderly with at least one chronic condition referred to at least one specialist was 80% in the Netherlands in 2004 [29]. Nie found 3 visits to specialists per year for the elderly in Ontario, Canada [6]. Regardless of the comparison problems of health systems in general and the definition of contact or visit in particular as described in the methods section, Germany seems to be leading country in the world with regard to the proportion of people referred to specialists, the number of different specialists visited and the number of contacts with both PCPs and specialists per year. Further research is needed to explain these exceptional features of the German ambulatory health care system. Apart from patient-based utilization habits, the contribution of morbidity-independent bureaucratic regulations in the insurance system (e.g. prescription regulations, budget limits, co-payment rules etc.) deserve more attention with regard to their effects on utilization. Like in the United States of America, specialists in Germany play an important role in caring for common conditions, not particularly when the level of comorbidity is high or the individual chronic condition warrants highly specialized care [30, 32]. As many consultations of specialists end with recommendations for further diagnostics and/or drug prescriptions, problems of guidance, documentation, coordination and cooperation between professionals arise, for which tools and routines are largely lacking.

Our study has weaknesses but also strengths. As this study is based on claims data, diagnoses were not clinically verified by professionals specially trained for this study. Erroneous and transitory diagnoses were minimized by monitoring the persons over a whole year, only acknowledging a condition as chronic if a ICD code from the 46er list was found in at least three quarters of the year. Privately insured patients (some 10% of the population) were not included in this study. The claims data do not allow the analysis of other important aspects of utilization, such as subjective utilization needs, disease severity or socio-economic characteristics of patients. On the other hand, insurance claims data do allow the analysis of large populations over long periods of time, including those living in protected institutions and in nursing homes and those of frail individuals as well as of the oldest of the elderly, all of whom are frequently excluded from field studies. The same applies to the lack of selection bias with regard to the service providers, an even greater problem in field research. Also, recall bias and social desirability problems in interviews concerning the utilization of services are excluded.

In this study we only examined the utilization of ambulatory physician services but we intend to expand this analysis to hospital utilization, physiotherapy services, and pharmaceutical use in order to obtain a complete picture on utilization and cost of medical care for the elderly in Germany.

In absolute terms, we found a very high rate of utilization of ambulatory medical care by the elderly in Germany, when multimorbidity and especially nursing dependency were present. This extent of utilization was related to the number of chronic conditions and to the presence of individual chronic conditions or their combinations in the multimorbidity pattern of the patients. On the other hand, the extent of utilization was not related to gender and was almost not higher in the older age groups among the elderly. These figures exceed those published by other countries by far. Further research is needed to explain these exceptional features of the German ambulatory health care system.