Paul, I believe your statement--that the likeliest explanation for this study's findings is that illness disturbs sleep rather than too little or too much sleep causes illness--is itself an unsupported "headline." This is a highly sophisticated set of analyses in a very large population, using Mendelian randomization, propensity matching, and non-linear structural modelling to explore that very question, is sleep duration or disease the causal agent. These are not lightweight academics looking for a quick pub and a flashy headline. If you read the article carefully, including the supplements, you will come to appreciate their conclusion, that potentially modifiable sleep duration probably explains most, but not all, of the association(s) between insomnia/hypersomnia and a wide range of biologic markers of accelerated aging. This is an important study.
Jonathan, thanks for reading so closely and for your comment. You make excellent points about this being a serious, sophisticated paper and that the authors worked deliberately on the question of reverse causation. I don't believe I made light of the study, and I respectfully disagree that my comment that long sleep is often a function of underlying illness, rather than the driver of accelerated aging, is unsupported by the research. In fact, the authors themselves spend considerable ink on the reverse causation question. They're careful to say their mediation analysis was "not for causal inference." To your point, they did say that their MR work "did not support a widespread causal effect of disease on sleep disturbances," but they also say plainly that reverse causality "cannot be fully excluded."
But they also acknowledge that long sleep "may not be a direct risk factor per se, but rather a marker of underlying physiological compensations or subclinical disease processes."
That said, your note made me rethink how I characterized my view as "the likeliest" explanation. I think I overstated that, in retrospect. I'm grateful for close readers like you.
Thanks for this. Sounds like we are in the same zone. I am sure the causality runs both ways. The article made me take more seriously that poor sleep can cause disease, and to empathize with the many folks whose family responsibilities, social positions, and economic imperatives make 7 good hours almost impossible.
Thanks also for all your work, which I only recently began to follow.
Paul, I believe your statement--that the likeliest explanation for this study's findings is that illness disturbs sleep rather than too little or too much sleep causes illness--is itself an unsupported "headline." This is a highly sophisticated set of analyses in a very large population, using Mendelian randomization, propensity matching, and non-linear structural modelling to explore that very question, is sleep duration or disease the causal agent. These are not lightweight academics looking for a quick pub and a flashy headline. If you read the article carefully, including the supplements, you will come to appreciate their conclusion, that potentially modifiable sleep duration probably explains most, but not all, of the association(s) between insomnia/hypersomnia and a wide range of biologic markers of accelerated aging. This is an important study.
Jonathan, thanks for reading so closely and for your comment. You make excellent points about this being a serious, sophisticated paper and that the authors worked deliberately on the question of reverse causation. I don't believe I made light of the study, and I respectfully disagree that my comment that long sleep is often a function of underlying illness, rather than the driver of accelerated aging, is unsupported by the research. In fact, the authors themselves spend considerable ink on the reverse causation question. They're careful to say their mediation analysis was "not for causal inference." To your point, they did say that their MR work "did not support a widespread causal effect of disease on sleep disturbances," but they also say plainly that reverse causality "cannot be fully excluded."
But they also acknowledge that long sleep "may not be a direct risk factor per se, but rather a marker of underlying physiological compensations or subclinical disease processes."
That said, your note made me rethink how I characterized my view as "the likeliest" explanation. I think I overstated that, in retrospect. I'm grateful for close readers like you.
Thanks for this. Sounds like we are in the same zone. I am sure the causality runs both ways. The article made me take more seriously that poor sleep can cause disease, and to empathize with the many folks whose family responsibilities, social positions, and economic imperatives make 7 good hours almost impossible.
Thanks also for all your work, which I only recently began to follow.