Corrigendum to “Trade-offs in long-term care for older people in an ageing society

Abstract

The authors regret an unfortunate coding error in the optimal portfolio analysis presented in Table 3 of the published paper. This was discovered while performing new analyses on the data used in this study for another study. In the optimal portfolio analysis, the estimates of the multiple discrete-continuous extreme value (MDCEV) choice model were used to calculate the expected utility of each feasible portfolio (i.e., every possible combination of policy alternatives given the budget constraint, with attribute levels averaged over the sample). The expected utility was calculated correctly, but when preparing Table 3 for the paper, the portfolios were ranked incorrectly (i.e., by portfolio costs instead of by their expected utility). As a result, the top ten portfolios presented in Table 3 in the paper were not in fact the top ten portfolios in terms of expected utility, but in terms of implied additional expenditure on long-term care. In this corrigendum, we present the correct Table 3 and discuss these findings. The authors wish to emphasize that all the other analyses and results, i.e., the descriptive results (Figures 2, 3, S1 – S3, S8 – S14), the MDCEV estimates (Tables 2, S3, and S5), and the Latent Class Cluster Analysis (Tables 4, 5, and S2 and Figures S4 – S7) stay the same. Moreover, the abstract and highlights, and thereby the main message of the paper, remain unchanged. After ranking the portfolios by their expected utility, as originally intended, Table 3 reads as follows: [Table presented] The correct top ten portfolios indicate a stronger preference for the institutional and home-based formal care policy alternatives (i.e., increasing capacity of nursing homes, nursing care and social care at home, and introducing care homes), relative to those previously presented. At the same time, they indicate a lower preference for providing respite care to informal caregivers and introducing compulsory social service for young adults. Also, increasing the use of supportive care technologies is less preferred, but remains included in most of the top-ranked portfolios. This implies that portfolios involving a moderate increase in formal care options were generally preferred. The results reinforce the paper's conclusion that respondents seemed to prefer distributing public resources towards multiple policy alternatives over investing substantially in one or two particular policy alternatives. Finally, while the ten highest-ranked portfolios do not exhaust the budget constraint entirely, like in Table 3 of the paper, most portfolios still imply an expenditure increase close to exhausting the budget constraint. Based on the above, the subsection ‘Optimal portfolio composition’ of the Results section in the paper should be as follows: “Table 3 shows the ten portfolios with the highest expected utility. For example, portfolio 10 includes an increase in the capacity of nursing homes and nursing care at home by 10,000 places each, the introduction of care homes with 20,000 places (i.e., two times 10,000 places), an increase in the capacity of social care at home by 10,000 places, and the provision of respite care to informal caregivers for a maximum of three months, while increase in use of supportive care technologies and compulsory social service for young adults are not selected. Several patterns can be observed from the top ten portfolios. For example, each of these portfolios included at least one of the policy alternatives regarding nursing care and at least one regarding social care. Besides, all portfolios contained five or six of the seven policy alternatives. Additionally, both institutional care alternatives and both home-based care alternatives were included in all of the ten highest ranked portfolios, with the increased use of supportive care technologies also included at least once in seven out of the ten highest-ranked portfolios. All ten highest-ranked portfolios resulted in substantial expenditure increases, with eight of the ten portfolios nearly exhausting the resource constraint (i.e., > €90 euros per adult per month).” The sentences related to this part of the analysis in the Conclusion and Discussion section of the paper should be as follows: “In the optimal portfolios, increased capacity of institutional and home-based care and use of supportive care technologies were often included.” “Most of the highest-ranked portfolios nearly exhausted the budget constraint.” “The policy alternatives regarding the increased capacity of institutional and home-based care and use of supportive care technologies are particularly encouraged, conditional on the policies’ effectiveness and efficiency in practice. While various forms of institutional and home-based care are more commonly adopted and arguably less challenging to implement, the use of supportive care technologies in long-term care remains relatively limited.” Finally, the optimal portfolio composition in the sensitivity analysis also changes. It is similar to what is presented in the corrected Table 3 above, but with portfolios 6 and 7 and portfolios 8 and 9, respectively, reversed. The authors would like to apologise for any inconvenience caused.