The Swedish Agency for Public Management

Review of the healthcare model in the cost equalisation for county councils (2013:9)

Statskontoret (The Swedish Agency for Public Management) has conducted a review of the healthcare model in the cost equalisation for county councils. This has been carried out within the framework of the Agency's instructed assignment to monitor the system of equalisation in municipal finances.

In this review we have worked on the healthcare model with the design proposed by the Parliamentary Committee.08 (Utjämningskommittén) in the report Equivalent conditions – Review of the municipal equalisation (SOU 2011:39).

Purpose of the review

The study focused on answering three questions.

  1. How should the model be able to ensure access to necessary information about the costs of healthcare?
  2. How can and should a future model take into account any regional differences in healthcare needs due to differences in socio-economic background factors?
  3. How can the model take into account cross-border commuting, i.e., that the Swedish registers do not show cross-border commuters' real income or employment status?

Statskontoret's proposals

In light of what is presented in the report, Statskontoret proposes the following:

  • That an alternative method for the collection of cost data is used during the next update of the cost matrix in the healthcare model. That the method is based on the National Board of Health and Welfare's patient register in combination with certain other systems that make it possible to put a cost on the different healthcare interventions.
  • That the current variables in the cost matrix are retained.
  • That a new class division of the age variable is made so that, following the next update, the age classes are as follows: 0 years old, 1–4 years old, 5–19 years old, 20–49 years old, 50–59 years old, 60–69 years old, 70–79 years old and 80 years old and above.

Design of the healthcare model

In the healthcare model proposed by the Parliamentary Committee.08, costs have been estimated through the whole population being divided into a matrix consisting of six demographic or socioeconomic variables that are costed. The variables used in the model are: income, age, gender, civil status, employment and housing.

The cost matrix is based on data for all residents in Skåne County Council who have been the subject of healthcare interventions in a given year.

The cost for persons with HIV is calculated using a special procedure in the model. In addition, a supplement is calculated for those county councils that have a sparse settlement structure, which is funded by a deduction for those county councils that have a more favourable structure.

The standard cost for the model as a whole is calculated as: cost for the entire population + cost for HIV + supplement/deduction for sparse settlement structure.

Justification for changing data collection method

Statskontoret notes in the study that there are a number of both conceptual and practical problems with data collection in the current model. Statskontoret considers these combined deficiencies to be so significant that, in the future, an alternative collection method should be used when developing the cost matrix for the healthcare model.

A basic premise of the system of cost equalisation is that equalisation should only be performed for those costs that cannot be influenced by local government. Equalisation should therefore not be performed for cost differences due to differences in selected service levels, quality, tariffs and efficiency.

This creates problems when, as in the healthcare model, redistribution is only based on cost data from one county council. The reason is that it is not possible to assess the extent to which the costs for various healthcare interventions and patient groups within this county council have been affected by the ambition and efficiency levels established within the various operations.

Another question is whether the cost differences found in Skåne County Council between different variable options (e.g., between residents in small houses or in rented apartment) would have the same weight and thus the same redistribution effect if they were based on cost data from the whole country.

It is questionable whether cost data based only on observed differences within a county council can be directly used as a yardstick for a redistribution aiming to compensate for structural cost differences between county councils.

Furthermore, the current collection method presupposes access to personal identity number-based cost data for all residents in Skåne County Council who have been the subject of healthcare interventions in a given year. The county council is currently not obligated to submit this information to government representatives.

People who live in one country and work in another are, in the majority of cases, not included in income and employment statistics in their country of residence. This means that the official statistics provide an incomplete picture of the number of individuals with earned income and the level of employment in municipalities and county councils with extensive commuting across country borders.

Since the cost matrix is based on Skåne, which has a relatively high number of cross-border commuters, the healthcare costs will be underestimated for those individuals who have a low level of employment and are lacking registered earnings.

The alternative data collection method

In this study, Statskontoret has used an alternative method for collecting cost data. The method is based on taking data on healthcare interventions from the patient register (PAR) and then combining it with the cost data generated by systems for Diagnosis-Related Groups (DRG) and Cost Per Patient (CPP). With this method, you can collect supporting data for the cost matrix which is built on information from all healthcare authorities. However, the patient register does not include information on healthcare interventions within primary care or healthcare interventions within outpatient specialised care performed by healthcare professionals other than doctors.

The data collected relates to healthcare interventions and costs in 2008. We have chosen this year because in doing so we have the opportunity to compare the results of the alternative method with the results reported by the Parliamentary Committee.08.

The presented data includes information from the patient register on more than 3.7 million persons. The total cost of healthcare interventions for these persons was calculated with the alternative method at SEK 75.3 billion. SEK 53.2 billion of this sum relates to costs for inpatient care and SEK 22.1 billion relates to costs for outpatient specialised care. These costs have been adjusted so that they comply with healthcare authorities' own reported costs.

The table below shows the results for the various county councils when using a cost matrix based on the alternative collection method (PAR08). A comparison is also made with the results according to the proposal by the Parliamentary Committee.08 (RS08). For the comparison to be fair, the costs according to the alternative method are adjusted upwards to SEK 128.5 billion, which was the cost level used by the Committee.

The second column of the table shows the results in SEK per resident and the third shows how large the redistribution (in SEK per res.) becomes in relation to the average cost in the country (according to PAR08). The fourth and fifth columns show the corresponding figures when the results are calculated according to RS08.

Table: Results and redistribution according to PAR08 and RS08 respectively

Table: Results and redistribution according to PAR08 and RS08 respectively

The table shows that all county councils that are net contributors with RS08 will also become so when the alternative method of calculation used. Consequently, all beneficiaries under RS08 also become beneficiaries with the alternative calculation method.

We can however conclude that there are some changes to the results. These changes depend on certain variables and variable divisions being given more weight when calculated using the alternative method. The variables income, civil status and employment all redistribute more with the alternative calculation method. For many of the classes in the variable age, redistribution is roughly the same regardless of calculation method, but the oldest age group has a higher average cost when using the alternative method. The housing variable redistributes more with the alternative method and the variable gender redistributes to a lesser extent.

In total, PAR08 redistributes more than RS08 – SEK 2.1 billion compared with SEK 1.6 billion.

Examination of the variables in the model

In Statskontoret's analysis, we have found that the current variables reflect in a consistent way the cost differences for healthcare interventions in the population; both in the country as a whole and in individual county councils. We have also found that none of the variables are irrelevant to the end result of redistribution.

Another question that is answered in this study is whether the current variables also manage to capture the regional cost differences for the most care-intensive group (in practice the group with the highest healthcare costs). Our conclusion from the analysis we have conducted is that the matrix fulfils this requirement.

We have also used a statistical method (t-test) to test these conclusions. The results show that healthcare in all county councils is significantly more costly for the most expensive group for each of the variables (at a threshold value of 0.05).

The need for further development

A shortcoming of the alternative method – at least for now – is that the patient register does not contain information on interventions within primary care and interventions within outpatient specialised care performed by healthcare professionals other than doctors. However, Statskontoret's assessment is that this ratio only affects the results to a limited extent. There is also reason to believe that the availability and reliability of data will eventually improve.