Due to the increase in musculoskeletal disorders of an aging population the World Health Organization (WHO) and the United Nations (UN) have declared the years 2000 to 2010 as to be the “Bone and Joint Decade”. In the beginning of the next decade, the booming supply of endoprostheses will still be a major issue. The large number  as well as the combination varieties of new prostheses modules urgently requires a quality ensuring joint replacement register.
Comparing accounting data and statutory health insurance data with registry data has limitations. The patient relevant outcome (time between primary implantation and revision surgery), however, can be extracted from both data sets and it is possible to compare these values. Comparing the present findings with data from the Swedish Hip Arthroplasty Register , one finds significantly smaller early revision rates than in the population of TK insurants. In this context the average early revision rate between 2005 and 2008 is 1.6% with a county council spread of 0.6% to 2.3%. Coincidently, a national comparison with the data of AOK (Local Health Care Fund in Germany) shows similar complication rates within an observation period of one year after hip endoprosthesis with coxarthrosis (2.8%), hip endoprosthesis with hip fracture (3.8%), and after knee endoprosthesis (1.7%) .
Most prostheses loosenings are due to aseptic loosenings according to the results published by BQS . This can be attributed to local inflammatory reactions of tissue to micro wear particles of the prosthesis material or bony transformation as reaction to a changed mechanical situation. This result has also been internationally reported (Sweden , Australia , England/Wales ).
Regarding possible differences in age distribution of patients with and without short-term reoperation, it cannot be concluded that the reason is increased exposure of younger patients’ endoprostheses or rather the lack of a possible surgery due to medical reasons or the less frequent wish of older patients to be reoperated.
Considering the top three index surgeries most often leading to revision, knee prosthesis designed for higher flexion (high-flex knees, OPS 5-822.b) is of special interest: The findings presented support the findings of Hamilton et al.  that high flex knee prosthesis shows a higher risk of early loosening. Endres  has shown that patients with high-flex knee prosthesis achieve no better range of motion than patients with usual knee prosthesis. It can be concluded that the patient related benefit of high-flex knee prosthesis is at least unclear, presumably not existent, but exposes the patient to a higher risk of earlier revision.
The Institute for Quality and Efficiency in Statutory Health Insurance, furthermore referred to as IQWiG (“Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen”) has carried out an investigation to level calculation and identified a U-shaped course between the number of cases and the risk for complications considering knee endoprostheses . Instead of the frequency of short-term reoperations the risk of a restricted postoperative joint mobility and postoperative wound infection has been analysed. These IQWiG results do not contradict the results presented in this publication. However, it is not the aim of the authors to foster the challenging discussion on minimum quantities [22, 23]. Instead, the authors would like to put emphasis on quality indicators in order to evaluate the success of treatment. The former are currently gaining importance in the context of selective contracts for integrated care, gatekeeper centred health care and disease managements programmes .
Limitations and strengths
The Techniker Krankenkasse has more than 8 million insurants representing approximately 10 percent of the inhabitants of the federal Republic of Germany. At TK, the sample analysed is a full sample and therefore representative for TK insurants. However, characteristics of the specific characteristics of TK insurants may differ from the entire German population e.g. regarding the socio-economic status and thereby the BMI that on his part may influence the rate of short-term reoperations. Unfortunately, with exception of somewhat additional documentation like for participants in disease management programs, there is no possibility to control confounders. Therefore it will be not allowed to extrapolate the results to the entire population.
The authors would have preferred to apply other quality indicators such as the “ten-year prosthesis survival” used by SALAR and SOS  but the localisation characters “R” (right sided), “L” (left sided), or “B” (both sides) relevant for the implant classification has only been documented since 2005-01-01.
The data on both hip and knee endoprostheses comprised more than 34,000 total joint replacements including follow-up and was thus comprehensible enough. Between 2005 and 2007 endoprostheses for shoulder, ankle, or elbow were so seldom implanted that reliable data to the joints of upper and lower limbs do not yet exist.
The assumption is that the de facto percentage of short-term complications will be higher than reported, since inoperable patients, the deceased and those who change their statutory health insurance fund have not been considered. Please note that the reported figure for primary implantations cannot be used for epidemiological information: On the one hand, the TK insurants differ from the total population of insurants in German statutory health insurance funds. On the other hand the complex primary implantations which are separately to encode (OPS 5-829.a) were just not considered as such as the use of tumour endoprostheses (OPS 5-829.c) or hypoallergenic prostheses (OPS 5-829.e). The OPS-catalogue has been adjusted several times during the observation period of primary implantations. The OPS 5-820.6 (femoral head) has been abandoned with the OPS-catalogue 2007 introducing the OPS 5-820.8 (hip resurfacing). The OPS 5-820.9 (neck-conserving femoral head prosthesis) used in the current OPS-catalogue 2010 was not evaluated since it has only been introduced with the OPS-catalogue 2008.
In addition to reoperations it is without doubt necessary to investigate re-reoperations, thus following the Swedish example . Moreover it would be enriching not only to focus on complications, but to consider all others medical services demanded in the context of the treatment. The data of statutory health insurance funds for example offers valuable data on post-hospital rehabilitation  or ambulatory care, physiotherapy and further medication. It also includes the ICD-codes not only of the primary diseases (e.g. coxarthrosis or rheumatoid arthritis) but also of the relevant comorbidities. In contrast to the options of the aggregated short-term information of hospitals transferred to the BQS the data of the statutory health insurance funds allows morbidity-orientated matching and benchmarking. Furthermore reoperation data can also be referred to the ICD-codes for injuries or periprosthetic fractures (ICD S72, S82). An overall analysis of joint replacement and all other related medical services would generate findings of the complete picture. Evaluation of statutory health insurance data will always have limitations (e.g. missing data on implants and operation techniques, the subjectiveness of patient evaluation or deficits in documentation). Nevertheless this data will deliver valuable, new findings on joint replacements and further research questions, especially when connected to register data.
Hardly anybody doubts the medical necessity of a high quality arthroplasty register [26, 27]. It is the basis for callbacks as known from the car industry where it is possible to unambiguously allocate vehicles to owners. An arthroplasty register will lead to valid revision rates which are four to fourteen times underestimated in clinical literature . And last but not least an arthroplasty register is highly interesting regarding economic aspects: The German Society for Orthopaedics and Orthopaedic Surgery, (DGOOC, “Deutsche Gesellschaft für Orthopädie und Orthopädische Chirurgie”) reckons savings of 45 million Euro in the third year of the register. In contrast, additional costs of the register will be distinctly less . A joint replacement register must necessarily have an implant to patient allocation and could usefully be supplemented by selected statutory health insurance routine data.