Yet Another Flawed Telehealth Study

A recent research paper on telehealth follow-up of heart failure patients, is being cited on Twitter and media sites as yet another case for telehealth preventing hospital admissions. As in other cases, the researchers reach this false logic due to a flaw in their thinking or their material.

In this case, the researchers state:

The research team exploited the within-person variation in each member’s heart failure telemonitoring program enrollment over time to examine whether there were any statistically significant associations between the program enrollment and the dependent variables. More specifically, member fixed effects were used in the multivariate regression models to remove variation across members in the sample by including a dummy variable for every member (ie, each member served as his or her own comparison). This approach removes all confounding related to any time-invariant factors, such as sex and race, as well as any underlying health conditions not directly observed from the data.

They are correct to assume that most confounders are removed by this method. As they did not include a control group, they haven’t accounted for the regression towards the mean which is inherent to following a population over time. They falsely conclude that the telegraph intervention resulted in the reduction in admissions. It is more plausible that with every re-admission, the population’s condition improved because of better medication, behavior change, and physical therapy during the hospital admission.

We can therefore not conclude that telehealth follow-up reduces admissions. I’m not entirely sure this study can even conclude that telehealth follow-up was cost-effective. Researchers should focus on whether telehealth is efficient in regards to the number of healthcare professionals needed per patients. Future healthcare is immediately threatened by the relative sparsity of doctors and nurses - not by the ever increasing economic expenses.