Methodology of the Hospital Comparison Switzerland

Patients and their relatives have a right to information about the quality of medical services. The association Hospital Comparison Switzerland is committed to presenting relevant data correctly and in a generally understandable way. Our comparisons help those concerned to be able to choose hospitals and clinics with the best quality. In order to counter the risk of false conclusions, we treat all quality data with the greatest possible care.

Right to information, preventing false conclusions

Patients are legitimately interested in quality comparisons because they are personally affected. They are interested in finding the medical offer that promises them the best help in their situation. The Swiss Health Insurance Act takes up this concern by providing for free choice of hospital and obliging hospitals to provide medical quality indicators (KVG Art. 22a). These are published by the Federal Office of Public Health (FOPH) or by third parties on its behalf.

Quality comparisons in the hospital sector are of great importance for patients and for hospitals and must therefore be compiled and processed with the necessary care. For this reason, we do a great deal to eliminate the risk of erroneous conclusions. In doing so, we are guided by the recommendations of the Swiss Academy of Medical Sciences (SAMS) on the "Collection, analysis and publication of data on medical treatment quality".

Relevant, correct, understandable

We follow the principle of providing patients and visitors to our portal with relevant, correct and comprehensible quality information and comparisons.

Relevant for us are indicators that are an actual reflection of significant medical treatment quality. Therefore, we prefer to use data on the quality of outcomes, as these measure what has the highest priority for the patient: Cure or at least the greatest possible improvement. For the patient, it is ultimately the outcome that matters and less the way in which this outcome is achieved. Nevertheless, we also present data on structural and process quality in isolated cases, but point out their limited informative value.

We ensure correctness by using only data from recognised and official sources, which have been collected in a methodologically sound manner and according to recognised scientific criteria. We present important factors such as sample size, response rate, instruments, time of measurement, data collection methodology, risk adjustment, etc. ourselves or refer to background information from the original publications.

We pay great attention to the understandability of the data. Medical issues are often complicated. We summarise the interrelationships in the simplest possible terms. If background knowledge is necessary, we provide it ourselves or refer to generally understandable sources.

Stars and rating scores

To help patients and interested persons make better use of the existing quality indicators, we transfer the various data into one uniform scale. This way the data become more comprehensible and can be combined. Mathematically this is done via normal transformation . The scale being used throughout the website ranges from 0-5 (stars) , with higher values meaning better results. The average is 2.5.

Examples: A clinic with three stars (i.e. the value 3.0) ranges better than the Swiss average (precisely: better than 60% of all listed clinics). A clinic with four stars (i.e. the value 4.0) is clearly better than the average (precisely: better than 80% of all listed clinics). The stars / values always refer to the chosen quality indicator. Did the user choose more than one indicator, the stars / values refer to the combination of the chosen indicators.

Please note that small differences between the quality indicators are not of importance as they may be due to measuring inaccuracy.

Combined searches

Our hospital comparison allows you to combine quality indicators while weighing them individually. The website then adds the values accordingly. An indicators marked as "important" is counted once in the total ranking, "very important" is counted twice. Ticking the field "not important" results in the indicator not being considered at all.

We pay great attention to correct statistics. It is particularly important to have a sufficiently large number of patients, as this is the only way to exclude the influence of hazard. Especially for data on mortality and infection rates, this may lead to small and mid-size hospitals fall out of the ranking due to not sufficiently large patient numbers (especially for low-risk interventions). For this reason, depending on the choice of the quality indicator, the website may only display a limited number of hospitals.

Quality indicators

We list the links to the detailed original documentation on the indicators on our indicator detail pages. Please also note our presentation of the strengths and limitations of the quality indicators.

Patient satisfaction (acute care, rehab, psychiatry, children's hospitals)

The ANQ carries out a risk adjustment. To ensure that our comparisons are statistically sound, we only include hospitals and clinics with more than 20 patient responses.

Infection rates

Swissnoso is tracking infections occurring one month within a performed operation (or one year after device inplantation). The numbers were risk-adjusted by Swissnoso. To allow meaningful calculation of infection rates, we only included hospitals with sufficient patient numbers (i.e. more than three expected or observed infections).

Re-operations

The rate of re-operations within two years for hip and knee prostheses is taken from the Swiss implant register SIRIS. It was set up by the Foundation for Quality Assurance in Implant Medicine. The data are risk-adjusted by the ANQ. In favour of statistically reliable statements, we only include hospitals and clinics with a sufficiently high number of patients in our hospital comparisons. This means that we only list hospitals and clinics that have more than six expected or observed two-year reoperations. This corresponds to a minimum of about 240 hip and about 180 knee implantations per year.

Relapses/re-hospitalisations

DThe rate of potentially avoidable rehospitalisations is calculated by the ANQ on the basis of the medical statistics of the Federal Statistical Office. The decisive factor is the number of patients who have to be readmitted to hospital within 30 days of leaving hospital for the same illness. Disease patterns or constellations for which such re-entries are to be expected are excluded by the ANQ. In addition, the ANQ carries out a risk adjustment. To ensure that our comparisons are statistically reliable, we only include hospitals and clinics with a sufficiently high number of patients. In concrete terms, this means that we only include hospitals and clinics that have more than six expected or observed potentially avoidable readmissions. This corresponds to a minimum of about 130 patients per year.

Mortality

The numbers were risk-adjusted by the Federal Office of Public Health (BAG). To allow meaningful calculation of mortality rates, we only included hospitals with sufficient patient numbers (i.e. more than five expected or observed deaths).

Patient numbers for specific procedures or diagnoses

Hospitals / clinics with no cases were filtered out.

Total patient number

We have calculated hospitals, rehab clinics, psychiatric clinics and birthplaces separately.

Number of physicians

Data from the Federal Office of Public Health (BAG) include non-plausible extreme values in the upper range. Possible reasons include: differences in methods of counting the positions, duration of inpatient stay, number of patients and scope of service. To correct these extreme values (outliers), we normalized the data using a rank transformation.

Number of nurses

Data from the Federal Office of Public Health (BAG) include non-plausible extreme values in the upper range. Possible reasons include: differences in the methods of counting the positions, duration of inpatient stay, number of patients and scope of service. To correct these extreme values (outliers), we normalized the data using a rank transformation. Birthplaces were calculated separately.


This page was last updated on May 24, 2022.