I always read "the higher the score the better the health", however, it seems to me that it is not true by the way the two summary measures are constructed. Or maybe - highly plausible - I am missing something.

Following "Ware, J. E., Keller, S. D., & Kosinski, M. (1995). SF-12: How to score the SF-12 physical and mental health summary scales. Health Institute, New England Medical Center" and the Appendix in "González-Chica, D. A., Dal Grande, E., Bowden, J., Musker, M., Hay, P., & Stocks, N. (2017). Are we reducing the risk of cardiovascular disease and improving the quality of life through preventive health care? Results of a population-based study in South Australia. Preventive medicine, 99, 164-170." consequently, I can assume that the "healthiest" individual will get PCS= 56.57706 and MCS= 60.75781 since, for each item, indicator is not created for the response choice indicating the better health, thus s/he will have all the weights multiplied by zero and then the summary measures will coincide with the constant .

A second individual "less healthy", may have, however a higher PCS or MCS if, for instance, has for all the items the healthier answer except for one item (so all the weights will be multiplied by zero but one, assume MH3_1). This individual will have PCS = 3.46638 + 56.57706 .. Higher than the former individual the "healthiest" one .