Long-term medical costs of Alzheimer's disease: matched cohort analysis

TY - JOUR

T1 - Long-term medical costs of Alzheimer's disease

T2 - matched cohort analysis

AU - Sopina, Elizaveta

AU - Spackman, Eldon

AU - Martikainen, Janne

AU - Waldemar, Gunhild

AU - Sørensen, Jan

PY - 2019/4/1

Y1 - 2019/4/1

N2 - OBJECTIVES: Medical costs associated with Alzheimer's disease (AD) are characterised by uncertainty and are often presented in a format unsuitable for decision modelling. We set out to estimate long-term medical costs attributable to AD compared to the general population for use in decision modelling.METHODS: We used multiple logistic regressions to generate propensity scores to match 26,951 incident cases of AD with 26,951 people without AD, identified from Danish hospital and medication registries. Costs were available for up to 11 years for each individual, representing costs for 10 years before and 5 years after diagnosis. Generalised estimating equations were employed to investigate the effect of having AD on primary care, medication, hospital and total costs in the matched cohort. We also explored the impact of other socio-economic and demographic factors on healthcare costs.RESULTS: We report costs by year to diagnosis, from 10 years before to 5 after. AD was associated with significantly higher costs, driven by medication and hospital costs, especially around the time of diagnosis. Mean total medical cost was €4996 higher for AD than for the control group in year of diagnosis, after which primary and hospital costs decreased to pre-diagnostic levels. AD had higher attributable primary care costs in years preceding diagnosis.CONCLUSIONS: Reporting AD-attributable costs by year to diagnosis can be useful for use in decision modelling. Medical costs attributed to AD are driven by diagnostic procedures and medication, and the impact of AD on medical costs may not be as high or prolonged as previously suggested.

AB - OBJECTIVES: Medical costs associated with Alzheimer's disease (AD) are characterised by uncertainty and are often presented in a format unsuitable for decision modelling. We set out to estimate long-term medical costs attributable to AD compared to the general population for use in decision modelling.METHODS: We used multiple logistic regressions to generate propensity scores to match 26,951 incident cases of AD with 26,951 people without AD, identified from Danish hospital and medication registries. Costs were available for up to 11 years for each individual, representing costs for 10 years before and 5 years after diagnosis. Generalised estimating equations were employed to investigate the effect of having AD on primary care, medication, hospital and total costs in the matched cohort. We also explored the impact of other socio-economic and demographic factors on healthcare costs.RESULTS: We report costs by year to diagnosis, from 10 years before to 5 after. AD was associated with significantly higher costs, driven by medication and hospital costs, especially around the time of diagnosis. Mean total medical cost was €4996 higher for AD than for the control group in year of diagnosis, after which primary and hospital costs decreased to pre-diagnostic levels. AD had higher attributable primary care costs in years preceding diagnosis.CONCLUSIONS: Reporting AD-attributable costs by year to diagnosis can be useful for use in decision modelling. Medical costs attributed to AD are driven by diagnostic procedures and medication, and the impact of AD on medical costs may not be as high or prolonged as previously suggested.

U2 - 10.1007/s10198-018-1004-0

DO - 10.1007/s10198-018-1004-0

M3 - Journal article

C2 - 30171490

SN - 1618-7598

VL - 20

SP - 333

EP - 342

JO - European Journal of Health Economics

JF - European Journal of Health Economics

IS - 3

ER -