Effectiveness and Safety of a Novel Care Model for the Management of Type 2 Diabetes at 1 Year: An Open-Label, Non-Randomized, Controlled Study
Sarah J Hallberg, Amy L McKenzie, Paul T Williams, Nasir H Bhanpuri, Anne L Peters, Wayne W Campbell, Tamara L Hazbun, Brittanie M Volk, James P McCarter, Stephen D Phinney, Jeff S Volek - Diabetes Therapy, 2018
In adults with type 2 diabetes, a continuous remote-care model based on nutritional ketosis markedly improved glycemic control and most cardiovascular risk markers at one year: HbA1c fell from 7.6% to 6.3%, weight dropped 13.8 kg, insulin was reduced or eliminated in 94% of users, triglycerides fell 24%, and HDL rose 18%. The inconvenient, skeptic-relevant signal is that LDL cholesterol rose about 10%. The study is open-label, non-randomized, and funded by Virta Health, the company selling the intervention.
Key findings
- HbA1c declined from 7.6% to 6.3% (59.6 to 45.2 mmol/mol; P < 1.0 x 10^-16).
- Body weight declined 13.8 kg, and insulin therapy was reduced or eliminated in 94% of insulin users.
- Triglycerides fell 24% and HDL cholesterol rose 18%.
- LDL cholesterol rose about 10% (P = 5.1 x 10^-5) - the inconvenient, skeptic-relevant lipid signal.
Effect measures
- Other: HbA1c declined from 7.6% to 6.3% (59.6 to 45.2 mmol/mol)P < 1.0 x 10^-16
- Other: -13.8 kgP < 1.0 x 10^-16
- Other: -24%P < 1.0 x 10^-16
- Other: +10%P = 5.1 x 10^-5
Why this evidence tier (Low)
- Risk of bias:
- High risk of bias by design: open-label and non-randomized, with an intervention cohort compared to a usual-care group rather than a randomized control, so selection and motivation differences are uncontrolled.
- Precision:
- Large within-group changes with extremely small p-values, but precision of effect cannot offset the lack of randomization for causal inference.
- Directness:
- Measures glycemic and lipid markers, not clinical cardiovascular events, in a type 2 diabetes population; the LDL rise is directly relevant to the cholesterol debate.
- Consistency:
- Glycemic and triglyceride/HDL improvements and the LDL rise are consistent with other ketogenic and low-carbohydrate studies.
- Funding / COI:
- Funded by Virta Health Corp, the commercial company that delivers and sells the continuous-care intervention being evaluated - a direct sponsor conflict of interest. Several authors (e.g. McKenzie, Bhanpuri, Volk, McCarter, Phinney) are Virta employees, and others (Hallberg, Volek) served as consultants or advisors, with several holding stock or equity. This is a substantial conflict that, combined with the open-label non-randomized design, limits causal strength.
Low certainty. Striking one-year metabolic improvements, but open-label, non-randomized, and funded by the company selling the intervention; the LDL rise is an important counter-signal.
- Population:
- Adults with type 2 diabetes; continuous-care intervention (nutritional ketosis plus remote care with medication management) cohort of about 262 completers at 1 year, compared with a usual-care comparison group. Open-label and non-randomized.
- Conflicts of interest:
- Funded by Virta Health Corp, the company selling the intervention. Multiple authors are Virta employees holding stock or equity (e.g. McKenzie, Bhanpuri, Volk, McCarter, Phinney); Phinney and Volek are company founders; Hallberg and Volek served as consultants or advisors. A substantial sponsor conflict of interest.
- Funding:
- Virta Health Corp - the commercial company that delivers the continuous-care intervention being evaluated.
Limitations
- Open-label and non-randomized; the intervention cohort was compared to usual care rather than a randomized control, so causal inference is weak.
- Funded by Virta Health Corp, the company selling the intervention, with several authors employed by or holding equity in the company.
- Measured glycemic and lipid markers, not clinical cardiovascular events.
- LDL cholesterol rose about 10% despite favourable triglyceride and HDL changes.
- An Author Correction/erratum exists (Diabetes Ther 2018; 9(2): 613-621; PMID 29508274); it is a correction, not a retraction or expression of concern.