Let’s say you are a health policy expert and you are sitting in your office and you receive a phone call from a news reporter. The reporter asks you to comment for a story she is writing. She has learned from the chief of neurosurgery at Cooper University Hospital that the mortality rate for neurosurgery performed with their robotic system is 8%. This chief also states that the mortality rate for patients with neurosurgery conducted at University of Pennsylvania (which doesn’t have the robotic system) is 18%. The Cooper Hospital public relations department plan s a media campaign to promote the superior outcomes for neurosurgery with the robotic system and the reporter plans to write that patients who have neurosurgery without the robotic system are twice as likely to die as those who have robotic-assisted surgery. You explain that the direct comparison of the two mortality rates may not be adequate to determine success. You mention the comparison could be confounded.
a. Describe confounding as you might to a newspaper reporter.
b. Please list 3 possible (and reasonable) confounders and describe how they might influence the results in this example (meaning how would they affect the measure of risk or association)
a. Confounding refers to a situation where there is a mixing of effects, making it difficult to accurately attribute the observed difference in outcomes solely to one specific factor. In other words, confounding variables can affect the relationship between an exposure and an outcome, leading to a misleading interpretation of the results. It is essential to consider these confounding factors when comparing mortality rates between different treatment methods or settings.
b. Here are three possible confounders in this example and their potential influence on the results:
1. Age: If the patients undergoing neurosurgery with the robotic system at Cooper University Hospital tend to be younger on average compared to those having neurosurgery at the University of Pennsylvania, age could act as a confounder. Older patients generally have higher mortality rates due to factors such as comorbidities, so a difference in age distribution between the two groups could contribute to the observed differences in mortality rates. Adjusting for age would help isolate the effect of the robotic system on mortality rates.
2. Case complexity: If Cooper University Hospital primarily handles less complex neurosurgical cases while the University of Pennsylvania treats more complex cases, this could confound the mortality rate comparison. More complex cases typically have higher inherent risks, irrespective of the surgical method used. Therefore, if the mortality rate at the University of Pennsylvania is higher due to the concentration of complex cases, it would not necessarily imply that the lack of a robotic system is the cause. To address this, one would need to compare mortality rates after adjusting for case complexity.
3. Surgeon experience: The expertise and experience of the surgeons performing the neurosurgeries can heavily influence patient outcomes. If Cooper University Hospital has a team of highly skilled surgeons who are also experts in using the robotic system, while the University of Pennsylvania has less experienced surgeons, this could confound the mortality rate comparison. Higher surgeon skill and experience could lead to better outcomes regardless of the method used. To account for this, one could compare mortality rates while stratifying by surgeon experience or through statistical adjustment for surgeon characteristics.
By considering and accounting for these confounding factors, a more accurate understanding of the relationship between the robotic system and mortality rates in neurosurgery can be obtained.