Quora .

This question originally appeared on Quora. Answer by Jane Chin, PhD, and president of the Medical Science Liason Institute.

Short answer: The researchers who authored the original observational study do not know why later bedtimes in preschoolers would correlate with higher obesity as teens.

Long explanation: The researchers were specific to state that their findings do not show causation and even the way it measured “sleep” is admittedly “imperfect” — I have emphasized the specific points the researchers raised about the interpretation of these results from their paper below:

Our results should be interpreted in the context of the following limitation. First, observational studies like ours cannot establish causality, and it is possible that underlying biological mechanisms drive both a child's obesity risk and sleep requirements. However, randomized intervention studies have included recommendations about bedtime as part of multifaceted approaches to childhood obesity prevention.32 Results from such studies are promising and support a causal relationship between late bedtime and obesity, but follow-up has been short term. Second, children in our study were born in 1991. We cannot be certain that associations would generalize to children born more recently. Third, measurement in our study was imperfect. Bedtime was recorded from mothers' phone-interview response to a single question about their child's typical weekday bedtime. It is unclear how mothers would respond for children with inconsistent bedtimes. Further, we do not know children's sleep duration, the quality of their sleep, or whether they have a different bedtime on weekends, and these aspects of sleep may also be important.33 Fourth, our results were robust to adjustment for multiple additional variables, but confounding because of unmeasured or poorly measured variables cannot be excluded. Finally, we used multiple imputation to increase the size of our analytic sample. Compared with a complete-case analysis,multiple imputation may reduce bias but it also increases variance.

Here are the key points about how to interpret or draw your own conclusions from this research:

1. This research is an observational study. Observational studies are useful to formulate new ideas about possible avenues of research to follow (new hypotheses). The biggest weakness in observational studies is factors that can skew (confound) observations aren’t apparent or detected in the study. In terms of levels of evidence (robustness of scientific conclusion you may draw from the research), Observational studies are Level 3, which should be viewed with scientific caution when trying to draw any conclusions about what is actionable from a general application perspective. (To learn more about levels of evidence, click here.)

2. Existing scientific research may show an association between sleep disorders/sleep deprivation with cardiovasular and metabolic disorder-related syndromes. For example, one study published in the journal Sleep reads:

Epidemiological studies have shown that short sleep duration is associated with increased incidence of cardiovascular diseases, such as coronary artery disease, hypertension, arrhythmias, diabetes and obesity, after adjustment for socioeconomic and demographic risk factors and comorbidities.

3. Or not! There may not be an association! Less sleep = More protection against obesity. In another report, another set of researchers say:

Collectively, cross-sectional and longitudinal studies on self-reported sleep duration and obesity do not show a clear pattern of association with some showing a negative linear relationship, some showing a U-shaped relationship, and some showing no relationship.

…It is still too early and a too easy solution to suggest that changing the sleep duration will cure the obesity epidemic.

Correlation and causality are different questions

We don’t yet know whether one causes the other — whether other risk factors including genetic risk factors that predisposes individuals to metabolic diseases also CAUSE sleep disorders, or whether sleep disorders CAUSE metabolic diseases. We only know that we tend to see both comorbidities. In the same line of analysis, we also don’t know whether treating the sleep issue will improve outcomes for the metabolic issues, or vice versa. These are all questions still under scientific investigation, in research designs that need to be more robust (Levels 1 and 2 of Evidence) than observational studies.

One of the biggest and most obvious question has to do with the sleep duration in these children

Does the health effects of 10 hours of sleep from a 8pm bedtime equivalent to 10 hours of sleep from a 9pm or 10pm bedtime? There is no way to answer this question based on the study design in the Preschooler-Later-Bedtime-Correlate-Higher-Obesity research, and the researchers know this and address this in their discussion portion.

One of the obvious confounding factors in this study is that when kids are in bed earlier, there is less chances of wanting a snack before bedtime, which may translate to habits of late evening snacks during adolescence and adulthood. This would be the dietary confounding factor that has a direct impact on obesity, but is not assessed in a meaningful way in this observational study, considering that the research relied heavily on parental reporting, which may have incorrect reporting or biased reporting on the parents’ part. Even if reporting is anonymized (and it isn't in this study bc answers were collected via phone interviews), a parent may not want to admit, “Yes the kid slept at 2am and I fed him 2 slices of pizza at midnight when he got hungry and juice at 10pm when he was thirsty.”

Another factor that cannot be addressed here is that in the 1990s, iPads and comparable electronic devices haven’t been invented. The usage of these devices could show an effect on sleep disruption (the light emissions from these devices have been shown to interfere with REM sleep), which means even if preschoolers go to bed at the early bedtimes, the quality of their sleep may have changed from their 1990s-peers, and this will change the way we even ask questions about the links between sleep and obesity.

Finally, it is useful to look at the lead researcher/author’s research history. Dr. Anderson deals with public health questions in particular populations and particularly with nutritional epidemiology. Her last few papers cover very broad questions about diet/nutrition and specific populations (low income communities, autism spectrum disorders, Ohio’s adults) that are meant to serve as points of forming hypotheses for more robust well-designed clinical trials looking to answer outcome based questions.

Therefore, a sensible response to an observational study may be: “This is an interesting observation. Now they need to figure out what the specific questions should be, that can give us useful, actionable information.”

In the meantime, healthy daily routines are important for children in general.

More from Quora: