GBV-C Co-Infection Associated with Improved Ebola Survival

GB virus C (GBV-C or Human Pegivirus - or even Hepatitis G) is associated with high viremia but there is little evidence that it causes disease in humans. Back in 2001, Jack Stapleton and colleagues (including Dan) showed that GBV-C co-infection significantly improved survival in HIV+ patients. The thought behind this association is that GBV-C attenuates aberrant immune activation.

Given that GBV-C infects between 10-28% of individuals in the three countries that have experienced the highest level of Ebola infections in the recent outbreak, Michael Lauck and colleagues in Madison wanted to examine the influence of GBV-C co-infection on Ebola outcomes. Using a cohort of 49 Ebola infected patients with outcome, age and gender data available they assessed the association of GBV-C co-infection on mortality.

Overall, mortality in the cohort was 69%. However, while mortality was 78% (28/36) in GBV-C negative patients, it was "only" 46% (6/13) in GBV-C co-infected patients. The unadjusted and adjusted analyses are in the Table below. The higher p-value with unchanged OR in the multivariable model likely represents a loss in power and not age-related confounding as the authors claim. Minor quibble - they presented a case-control (OR) analysis for this cohort of patients with a significant p-value. Analyzed as a cohort study, the RR=0.59 (0.32-1.09), p=0.0950. Either way, if I had Ebola, I'd also want GBV-C.

Comments

  1. Interesting article. We should add GB virus to the statin/SSRI/sildenafil Polypill.

    Although the outcome of interest was common in this cohort study, and so RR would have been the preferable univariate outcome, since the authors prudently conducted multivariate logistic regression (whose output is odds), it seems more symmetrical to report the univariate analyses in the same way.

    Did you have an alternate approach in mind e.g. reporting both endpoints in univariate analyses?

    Tim

    ReplyDelete
    Replies
    1. Thanks for your comment Tim. It was a minor quibble, but I just thought it was important to mention, especially since the OR overestimated the benefit compared to the RR. For these types of studies, I think building survival models with each variable individually and reporting the univariable HR and then building a multivariable survival model is a good way to go. But there is never a single correct way to analyze data.

      Delete
  2. Yes, agree, survival models are best if follow up time is available. If not logistic regression is fine but agree the uncritical reader can be over-impressed.

    ReplyDelete

Post a Comment

Thanks for submitting your comment to the Controversies blog. To reduce spam, all comments will be reviewed by the blog moderator prior to publishing. However, all legitimate comments will be published, whether they agree with or oppose the content of the post.

Most Read Posts (Last 30 Days)