Overutilization of ambulatory medical care in the elderly German population? – An empirical study based on national insurance claims data and a review of foreign studies

According to the criteria used in this study, it is estimated that, depending on the study population (see strengths and weaknesses section) some 20 % of the ≥ 65 year-old, statutorily insured German population are to be considered high utilizers of outpatient care. This figure corresponds to approximately 3.5 million persons covered by statutory health insurance. The frequent attenders in this study had on average 62 contacts with an ambulatory care practice (1.2 contacts per week) spread over 9 different physicians, usually 1 GP and 8 specialists, per year. Two main HU-types were found. The first type (HU-type A) is characterized by the large average number of practice contacts (71/year = 1.4 contacts/week), found predominantly among the oldest age group (42 % ≥ 75 years), in patients with a high percentage of multimorbidity (89 %), and relatively often in need for nursing care (15 % as compared to 11 % in the general population in this age group). This type corresponds to the traditional public conception of a high utilizer as a very old person suffering from multiple burdening diseases that require continuous medical and nursing services. The other HU-types - a combination of HU-types B and C - were comparably prevalent but consisted of a significantly younger age group (30 % ≥ 75 years) that was somewhat less frequently multimorbid (83 %), and which particularly less frequently received nursing care (5 %). These HU-types consulted many different physicians (11 - 12 per year), often from the same specialization (36 %). These types of high utilization are frequently called “doctor (s)hopping” in the literature [26, 27]. Remarkably, gender differences were not found.

The criteria for this typology were set deductively on the basis of the empirical distribution of the data. Therefore it should be kept in mind that using stricter or more tolerant criteria for frequent attendance would have lead to different results. For example, using ≥ 8 instead of ≥ 10 different practices as a cutoff for HU-type B would have raised the percentage of B from 9 to 12 % of the elderly population; inversely, a cutoff of ≥ 2 instead ≥ 3 practices of the same discipline would have raised the percentage of C-type high utilizers from 5 to 24 %. In other words, a quarter of the elderly contacts ≥ 2 specialist practices of the same discipline per year.

High utilization of statutory ambulatory care is related to specific diagnoses. All 46 ICD-10 diagnoses for chronic conditions investigated in this study showed a minimum odds ratio of 1.3 for belonging to the high utilizer sample, and 27 of 46 had a respective odds ratio ≥ 2. Also, complex associations were found between diagnoses and HU-type. On the one hand, several chronic conditions were found in the list of top 10 relative risks for all three HU-types A, B, and C, albeit at different positions in the list for each type. This is the case for urinary incontinence, cancer, anemia, neuropathy, chronic polyarthritis, and anxiety (see Table 3). In diagnoses like cancer, the most prevalent disease of the ABC-subtype, the combination of serious somatic disease, psychological stress, and the search for further treatment options could be the reason for often contacting several practices, including those of the same specialization. Neuropathy also, the top ranking disease in type B, might fall under the category of symptoms that are difficult to treat, resulting in consultations with multiple physicians. On the other hand, certain groups of chronic conditions are specific for single types of high use. Dementia and Parkinson’s disease, for example, only appear among the diseases associated with many contacts (type A) but not with consulting many specialists; in the case of dementia, contacts are centered mainly around the GP [28]. Furthermore, several syndromes with a possibly psychiatric background are found especially in HU-types B and C, such as somatoform disorders, sexual problems, anxiety, and migraine/headache. Very probably, these patients migrate through the care system and contact many different specialists because of a lack of satisfaction with the explanations or treatments offered to them. The relevance of psychiatric and psychosomatic factors for “doctor (s)hopping” has been confirmed in many international studies [13, 16, 27], as well as in German studies. In a representative study of the general population in Germany, Mewes et al. found a higher level of somatization and/or post-traumatic symptoms to predict high utilization [29]. Schneider et al. found a higher utilization rate of GP practices by patients with recorded psychiatric diagnoses compared to those with somatic diagnoses [30]. Apart from morbidity, for a patient to become a high utilizer of a particular type is probably dependent on additional factors, such as the burden of comorbidity [8], and vulnerability [31]. Furthermore, personality variables (e.g., subjective disease concepts [32]), personality traits [33], situational factors (e.g., regional service supply differences or previous experiences with the healthcare system) might play a role. Further research is needed in order to identify such factors.

With an average of 28 contacts per year distributed over 5 physicians in the total elderly population, 36 contacts per year spread over 6 physicians in the multimorbid elderly subsample, and 62 contacts per year with 9 different physicians in the 19 % HU-subsample, Germany seems to be at the top of the rank order of ambulatory utilization in OECD countries as shown in Table 4 which summarizes the results of a review of 27 empirical studies on the extent and frequency of utilization of ambulatory care in several OECD countries, with special reference to utilization by the (multimorbid) elderly and high utilization.

The top position is also found for the referral rate to secondary ambulatory care by a primary physician (93 % of the elderly at least once in a one-year period) and – inversely – for the length of contact with a physician in minutes.

Table 4 shows important differences between countries and studies. These differences are partly due to differences in definitions, sampling strategies, and data handling methods. For example, the mean of a range is not always adequate to account for the large variations between regions [34, 35] and practices [36]. Also, many studies do not describe in detail the nature of the contacts investigated (face-to-face, telephone etc.). Furthermore, several studies do not differentiate between contacts with physicians and those with practice nurses or practice assistants, whereas others do [37, 38]. Still, the following cautious conclusions can be drawn from Table 4:

Even with all of these differences between the studies taken into account, the German figures far surpass those of most other industrialized countries. In this context, it should be kept in mind that the German figures do not include oral health services, most of the complementary medical services not covered by the health insurance, and the visits to emergency departments (currently estimated at 125/1,000 of the total population per year and rapidly increasing). Also, according to the study by Koch et al., no substantial difference between the type of contacts in German GP-practices in comparison to other countries are given, as German GPs declare that 70 % of their contacts with patients are personal contacts, a figure corresponding to a middle ranking (min: 66 %, max: 87 %) among the 11 countries in that study [40].

Further research is needed to explain these exceptional features of the German ambulatory healthcare system. Several reasons may contribute to this phenomenon, probably involving mechanisms on the side of the providers, the patients and the system itself. Among the reasons on the patients’ side, possibly greater attention to complaints and diseases, greater health related fears [41], tendencies to delegate responsibility from one self to physicians, and greater beliefs in procedural approaches to diagnosis may play a role. German physicians in ambulatory care may practice contact increasing strategies, e.g., by recall habits, client binding strategies, and competition with specialists. Also, the impact of bureaucratic directives of the insurance system on utilization (e.g., quarterly prescription regulations, remuneration specifics and budget limits, certification requirements for short time sick leave etc.) should be analyzed. Like in the United States, specialists in Germany play an important role in caring for common conditions, independent of the level of comorbidity and of the medical need for highly specialized care [3, 42]. This may be due to the particular “double specialist track” in Germany, a term describing the doubling of hospital based specialists working solely in the hospital (with an actual density of 1 hospital specialist [including resident specialists] to 814 inhabitants) by a supply of specialists working solely in ambulatory care (actual density of 1 specialist to 1,192 inhabitants) [43]. Furthermore, the German ambulatory healthcare system is based solely on the figure of the physician, whereas nurse practitioners and qualified registered nurses hardly play a role in the provision of services in the practices, which leads to a complete lack of alternatives to the physician in health services.

The high number of contacts in Germany is associated with smaller time slots per consultation when compared with the UK and especially with the USA (Table 4: [40, 42, 44, 45]). Other authors forwarded that in gatekeeping countries (with fixed lists of patients) GP consultations are longer, more time is spent talking with patients, especially about psychosocial aspects of their conditions, and the workload of general practitioners is lower [46]. Table 4, however, shows that this is not generally the case.

The main problem of high patient numbers per day and short consultation lengths may be that those patients who need intensive visits may not get them because the waiting room is always full, such as is the case in Germany, where it is not uncommon for GPs to see 70 patients per day. Østbye et al., who calculated the number of hours needed to provide guideline-conform care for the 10 most common chronic diseases in the USA, “provided the disease is stable and in good control”, found a requirement of 3.5 h a day. When adding uncontrolled disease, the estimated time required increased by a factor of 3. Applying this factor to all 10 diseases, time demands would increase to 10.6 h a day. The authors concluded that “current practice guidelines for only 10 chronic illnesses require more time than primary care physicians have available for patient care overall” [47]. German GPs are aware that their patient load does not allow them to allocate sufficient time per patient, and they consider a reduction of their patient load by around 25 % to be necessary for sufficient care [48].

The other danger of high utilization is that most of the contacts produce diagnoses, suggestions for further diagnostic procedures, referrals, drug prescriptions, physiotherapy prescriptions, and lifestyle recommendations that need to be understood by the patients and integrated into their lives. It is unclear how elderly patients handle this in their daily lives. Barbara Starfield concluded from several studies that “an excessive specialist supply with inappropriate specialist use leads to greater frequency of tests, more false-positive results, and worse outcomes than appropriate specialist use. The more physicians patients see, the greater the likelihood of adverse effects” [42]. As people’s use of specialist services is rapidly increasing, at least in the United States [42] and in Germany, the concepts of personal responsibility and self-management by (elderly) patients appear equally as fanciful as those stressing the need for continuity and comprehensiveness in medical care in these countries.

In spite of the dissatisfaction on both sides – primary care physicians and patients – professional and/or political initiatives to reduce the utilization rates have not occurred, so far. Politicians might fear that the introduction of visible restrictions to utilization may result in criticism for a presumed cutting the welfare state. The size of high utilization described in this paper is generally considered to be a social achievement rather than a societal problem of medical care. The political and public disinterest in the issue of utilization frequency is also demonstrated by the fact that a reform of the reimbursement system in ambulatory care in 2008 went ahead with a discontinuation of the administrative documentation of utilization rates. Since that year, the responsible institutions – statutory health insurance companies, statutory physicians’ associations, and also the government – consciously renounce to document data on contact and referral rates for Germany.

The exceptional figures for Germany have not been noticed internationally because the OECD publishes mistakable data on the contact rates in German ambulatory care. Under the heading “number of contacts with physicians”, OECD does not report the number of contacts (actually estimated at 20 contacts per year [49]) but the number of contacted practices per patient and year (e.g., 8.2 in 2009). By means of this unusual – and misleading – definition of “contact”, explained only in a rather cryptic footnote, Germany occupies a prominent but still inconspicuous position in the international rank order of contact numbers [50].

Of course, high utilization rates do not mean unjustified and/or dangerous overuse per se. Also, apart from figures on the general population and for the elderly as described in this paper, specific figures for the German working population are lacking. Also, evidence for the motives behind high utilization cannot be found in claims data. As there is no objective criterion for high utilization and, thus, no valid reference for diagnosing overuse, the very high utilization figures in German ambulatory medical care do not allow to use the term overutilization. On the other hand, the far-above-neighbor OECD countries utilization figures do not correspond to German outcome figures [51]. Therefore, questions arise about the relationship between benefit and harm of such high contact rates. As the utilization of inpatient care in Germany is also at the top of the OECD ranking list [18], more research is needed on the reasons and motivations for utilization rates in both healthcare sectors as well as on the interaction between the two [52]. In contrast, older people generally need a comprehensive and coordinated treatment plan, which is difficult to achieve with a two-digit number of contacts with specialists, especially considering the well-known deficits of communication between these specialists [53].

This study is based on claims data and therefore shows strengths and weaknesses inevitably related to this database. Apart from age and gender, no further socio-demographic variables were gathered that could be used to explain the differences in prevalence and risk. By including 46 diagnoses, we assume that all common and serious chronic diseases have been investigated. These diagnoses are documented by physicians – GPs and specialists – working in statutory healthcare and thus not by specially trained study personnel. It is known that there are limits to the validity of coded, claims related diagnoses [54, 55]. Validity issues of this nature were also found in the primary data collected via interviews with general practitioners [56, 57]. On the other hand, claims data allow a larger study population, including patients that are hard to reach in field studies, such as nursing home residents, very elderly patients, and severely ill patients. Memory bias regarding utilization and a bias due to socially acceptable answers of patients are excluded. This is the case for provider data as well. Finally, this cross-sectional study can detect associations but not causes.

The claims data stem from one insurance company, which poses the question of representativity. As there was and still is a high number of insurances (some 150 around 2006) some operating at regional and others on a national level, there is no possibility to draw a representative sample of the insured population in the strict sense of the term. The few studies comparing data from several insurers show that both problems of internal and external validity are given. With regard to internal validity, the GEK-Data show a higher proportion of males for historical reasons. As analyzed in a previous study, these gender difference does not have a major impact on utilization figures in the elderly population insured by the GEK [3]. As for external validity, comparisons between insurances are limited to the epidemiology of certain diseases, and studies on utilization data are lacking [58, 59]. Exercising extreme caution we may postulate from other data sources on insurance characteristics that there is no reason to assume that data from GEK-sources are exceptional or extreme.