Surgery Blog for Surgeons

(Ab)Using scientific journals for political purposes

I just read the article "What Liver Transplant Outcomes Can Be Expected in the Uninsured Who Become Insured via the Affordable Care Act?" in the last edition of AJT and would like to share some thoughts about it.

Cohort selection

On page two the authors write:

All charity care candidates were rigorously screened by social workers and neuropsychiatric professionals to ensure that they have 1) a strong social support system, 2) a failsafe transportation mechanism to facilitate access to posttransplant clinics, 3) adequate cognitive function to understand the transplant process, and 4) a personality that predicts that the candidate will be reliable to take medications as prescribed and come to clinic appointments as required.

Besides suggesting high preselection bias of the patients in this group, one would wonder whether the protocols in Alabama don't include this kind of evaluation of all patients and not just of the ones who are uninsured.


The "charity care" patients are 23 (compared to 630 private and 233 publicly insured)! While we don't expect a comparable cohort size, retrospectively following 23 patients over a period of more than 10 years is clearly not a method able to deliver any usable data allowing conclusions about the patients which would be covered by the Affordable Care Act.

Table 1 shows some data differences. The MELD score of the "charity care" patients is not only significantly higher. The significance here does not reflect the difference of the MELD which is substantial in this retrospective observance. MELD differences definitely influence outcomes as we very well know by the published data. Considering the small size of the "charity care" cohort, this significance and MELD differences are expected to impact the outcome in a way which is not reliable.

Figure 1 shows a difference of treatment based on insurance status which the authors disclose as being highly significant. This fact may reflect general policies in the way patients are treated at UAB. The obvious preference for privately insured patients may influence the overall treatment of these patients and the respective treatment of the ones from the other two groups. This would undoubtedly impact the overall outcomes and its comparison between the groups.

The reported significant differences of partner status between the groups (with the "charity care" group being most negatively presented) not only impact the expected outcome but one would wonder how this may have happened during the rigorous screen reported in the "Cohort selection" part. There is enough data that partner status does impact transplant outcome despite the fact that this might not be the case in Alabama as reported further down.

Describing the deaths is where the cohort size demonstrates its significance. Three of 23 patients died of for non-compliance reasons (however these may be defined in Alabama?!). These 3 patients account for 37.5% of all deaths! This is why the data is so unreliable.


The following sentence is picking my attention right at the beginning of the discussion:

The public health debate is whether insurance status is causally associated with patient outcomes or just an aggregate surrogate marker of socioeconomic status.

This is profoundly wrong! There is no debate about insurance status being associated with patient outcomes as this is something we know for many years.

It is unclear why the authors expect the "charity care" patients to have the lowest income. Patients with lower income should be covered by the public insurances or at least this was the original purpose of those insurances. People without insurance are usually the ones who work and have low income hoping for better days to buy an insurance. But this income should be above the cutoff point for receiving public insurance as otherwise they would have been publicly insured. At least this is my understanding of the things which admittedly could be very wrong.

The fact that a group of highly selected patients without insurance in Alabama still has a lower income than the ones who have a public insurance is at the least interesting. The authors did not investigate why these people don't have an insurance as they should be eligible for Madicaid/Madicare by definition, unless access to healthcare is utterly wrongly organized in Alabama.

The authors were not only unable to determine the household income per capita but also the educational level of the patients. Furthermore, educational level was not even included in the criteria for defining the socioeconomic status. Commenting on this might make me use inappropriate and politically incorrect language.

Case control analyses do not reduce the likelihood of type 1 statistical error as the authors write! Playing with the numbers does not reduce the need for a bigger cohort.

The following sentence at the end of the discussion is interesting:

These data, albeit limited in number, raise concerns regarding expected survival outcomes in newly insured patients via the Affordable Care Act.

The question here is whether someone who has any kind of insurance has better a outcome than someone who does not. We know that the not insured patients with ESLD and an average MELD of 27 would have a one year survival of about 15% and a 5 years survival of about 0% without treatment. The "charity program" of the UAB and the Affordable Care Act (as suggested by the authors, would) therefore tremendously increase the outcome of these patients.

One would wonder why then the authors suggest that these patients would have a decreased outcome. The answer to this question can be probably found in the overwhelming evidence of biased look at patients not only in one department (Transplant Surgery) but in the whole institution of UAB as the multidisciplinary group of authors suggests.

Why did AJT publish this scientific junk? Probably because opposing Obamacare is so very popular among US surgeons. If you go into surgical lounge rooms across the country and see FOX News running all the time, then you might understand the bias. I don't have any better explanation for this kind of politics making scientification.