The SHI system and the Association of SHI Physicians in Germany
Germany's healthcare system is premium-funded through statutory health insurances (SHI) for 90% of the population. There are about 115,000 physicians providing outpatient services within SHI in Germany. SHI pay for services of outpatient physicians through regional associations of SHI physicians which handle all billing information on an individual basis. To bill patients, outpatient physicians are obliged to be members of a regional branch of the 'Kassenärztliche Vereinigung' (KV). For payment, physicians submit electronic invoices based on medical codes for each visit to the KV. Any physicians' contacts and preventive measures, such as vaccinations, can be identified on the basis of a specific medical code. Each of the 16 states in Germany has at least one regional KV. Data of the Bavarian KV (KVB) allow identifying the different vaccine antigens administered at vaccination visits since July 2001.
Patients and setting
Data of the KVB were used. Bavaria is the largest (70,552 km2) and the second most populated state of Germany with 12 million inhabitants . We analysed physicians' consultations and vaccination histories for children born between 2001–10–01 and 2002–12–31 with a follow-up until 2004–12–31, allowing for analysis of any billing for these children up to an age of at least 24 months.
Nature and structure of data
The KVB supplied data of all patient consultations that were billed by outpatient physicians on a quarterly basis. Each patient has a unique identification code within his SHI. This may change in case the patient changes the SHI, which does occur because of competition between SHI. The child's vaccination history therefore is completely mapped by the billing data unless the child switches to another SHI or to a physician outside of Bavaria during the observation period. The following data were available for the statistical analysis of the KVB data set: an identification code, member of the major health insurance company of Bavaria (AOK Bavaria; yes/no), birth date of the child, date of consultancy and – if vaccinated – 4- to 5-digit codes identifying the vaccines administered, or consultancy for any other reason per quartile year.
To calculate vaccination coverage, the number of vaccinated children (numerator) and the total number of children eligible for vaccination (denominator) has to be known. While numerator data are directly provided in the KVB dataset, an appropriate denominator has to be derived from KVB data. A child not consulting a Bavarian outpatient physician or not insured in SHI will not appear in the KVB data. A simple approach to assess the number of children eligible for vaccination within the SHI is to consider all children with any physician billing up to 2004–12–31 (original KVB dataset). This approach does not take the following censoring events into account: (a) children migrating from or to Bavaria; (b) children treated by a Bavarian physician during their holidays and therefore leaving or entering responsibility of the KVB; (c) children changing between different SHI or between private health insurances and SHI. The magnitude of these problems and its influence on vaccination coverage estimates was not known. We therefore compared two datasets: (1) the original KVB dataset, consisting of all children with at least one physicians' visit during the defined follow-up period; and (2) a reduced KVB dataset, applying a simple algorithm which restricts the original dataset to a cohort of children with at least two physicians' visits, independent of its medical character.
The number of children in the two KVB datasets for a Bavarian birth cohort born 2001–10–01 to 2002–12–31 is compared to the official numbers provided by the Bavarian Bureau of Statistics  and differences will be analysed.
Vaccination coverage is assessed both for the original and the reduced KVB dataset. Furthermore, to check if the reduction strategy of the KVB dataset was valid, additional information from one health insurance company of Bavaria (AOK) was obtained. The aim was to define a population that was continuously insured with one SHI during the observation period. The AOK Bavaria supplied data on all children born between 2001–10–01 and 2002–12–31 who were continuously insured in the AOK Bavaria from birth to 2004–12–31. The following data were available: an identification code, birth date of the child, date of entrance to and exit from the insurance. The vaccination histories of the children in the AOK dataset could be obtained by linking the AOK identification number to the identification number for AOK insured children in the KVB dataset. Vaccination coverage estimates from children continuously insured in the AOK Bavaria is compared to those from a reduced KVB dataset for AOK insured children only.
Definitions for vaccination status
According to the German vaccination schedule  we defined a child as 'fully primed' if he or she received at least three doses of a vaccine containing acellular pertussis components or two doses of a vaccine not containing acellular pertussis components. A child was defined as 'fully immunized' if she or he received a booster dose at the age of 11 months or later following full priming. In accordance to the above definitions, we estimated the coverage of the following combined series: first dose, full priming and full immunisation against diphtheria, tetanus, pertussis, polio, Haemophilus influenzae type b and hepatitis B; first dose against measles, mumps and rubella (MMR) . Uptake and timing of immunisation by age in months was calculated according to the Kaplan-Meier method which is described in detail by Laubereau et al. .
All data provided by the KVB and analysed in the Institute for Social Pediatrics and Adolescent Medicine, Munich, were completely anonymous in accordance with German data protection laws.