4.7.4. Issues specific to Switzerland

The cost data was obtained from the following settings: inpatient acute care; inpatient rehabilitation care; inpatient psychiatric care; nursing homes; outpatient physician care; outpatient drugs; domestic care.

For the inpatient settings, the latest available version of the hospital inpatient registry (Medizinische Statistik der Krankenhäuser, published by the Federal Office of Statistics FSO) from 2016 for the estimation of inpatient spending (acute, rehabilitation, and psychiatry) was used. The registry contains complete and detailed information about each inpatient stay at a Swiss hospital/clinic, including diagnostic data coded according to ICD-10-GM. For the outpatient setting and nursing homes, the study “Alcohol-related spending in Switzerland” (Alkoholbedingte Kosten in der Schweiz) by [Fischer, Telser, Widmar and Leukert, 2014 [18]] from 2014 was used as a source instead, since the Swiss research team had no access to individual-level data for the outpatient sector as well as for nursing homes.

As much as possible, the same disease definitions were used in the Swiss study as in the other anchor country studies, although such information was not always available. The data was stratified by gender and by the same age groups as in the studies for the other anchor countries.

Like in the other anchor countries where the healthcare data was available, a bottom-up estimation approach was used for the estimation of medical costs in Switzerland. However, unlike in the other countries, data was not available about the spending of people without any of the specified diseases (i.e. for the so-called residual costs). Therefore, it was impossible to estimate incremental spending for each disease in using a regression-based approach. Instead, average spending per capita in each age and sex group according to National Health Accounts was estimated.

For the inpatient sector, total spending per disease category (and for a relevant age/gender group) was obtained using a bottom-up estimation approach with individual-level inpatient registry data, covering acute, rehabilitation, and psychiatric inpatient care. Diseases were identified according to the ICD-10 code of the primary diagnosis of each hospital stay. While the acute sector is reimbursed based on a DRG system, the rehabilitation and psychiatric sector are reimbursed based on a per diem tariff. Thus, spending was estimated differently for the acute sector and for the rehabilitation and psychiatric sector.

For the outpatient sector, due to the absence of data, existing spending estimates from the literature were adapted to the remaining health care settings (including outpatient care by physicians, inpatient care at nursing homes, drugs, and domestic care) based on study by [Fischer, Telser, Widmar and Leukert, 2014 [18]], which relied on the top down approach to calculation of costs. The adaptation mainly consisted of a matching of the different disease and age group definitions between those used in the OECD study and the study by [Fischer, Telser, Widmar and Leukert, 2014 [18]]. as well as adjusting spending for price inflation in order to obtain estimates for 2016.

To estimate average incremental costs per patient, the total estimated costs were linked with the prevalence data from the Global Burden of Disease study, which allowed to calculate average spending per prevalent case for each disease and age/sex group.

To convert this average spending data into estimates specific for people with and without any comorbidities (which is required by the model), a comorbidity adjustment factors were estimated. First, average disease costs were translated into baseline disease costs for people without comorbidity, using the information on the prevalence of modelled diseases with and without comorbidities in France . Second, using the available data from the Dutch study described in this paper, the ratio of disease costs for people with to costs for people without comorbidities was estimated. Finally, baseline disease costs for people without comorbidities were multiplied by this ratio to estimate the disease costs for people with comorbidities.

The costs for the following diseases were estimated with this methodology using the Swiss data: AUD; depression; IHD MI; dementia; haemorrhagic stroke; ischemic stroke. The data on several other diseases were provided, but the disease definitions were not compatible with those used for the other anchor countries. In addition, it was not possible to estimate the residual costs and the extra costs of comorbidity using this methodology, and therefore, they were extrapolated from other countries using the general extrapolation method described in this document.