IQVIA White Paper on Time in Range: Improving US Diabetes Population TIR to 70% Could Save At Least $2-$4 Billion Over Ten Years – November 7, 2019

Download; Up to $9.7 billion in cost savings for 80% TIR and 40% Reductions in Hypo; Uses IQVIA’s Core Diabetes Model to Assess Cost Savings; Why is it conservative?

Healthcare consulting company IQVIA published an exciting whitepaper today, “Advancing Glycemic Management in People with Diabetes,” focused on Time in Range (TIR). The 22-page paper describes the limitations of A1c, the benefits of measuring TIR with CGM, and estimated cost reductions if average TIR for people with diabetes in the US (type 1 and type 2) reached 70% or even 80%. The cost savings estimates range from $2.1-$9.7 billion over ten years (!), depending on the assumptions used within IQVIA’s Core Diabetes Model (read more below). Many of the assumptions are rightly acknowledged as conservative, leaving a lot of upside here to show even more meaningful, population-wide cost savings.

Momentum behind the “Beyond A1c” movement is at an all-time high – especially with the establishment of consensus TIR targets earlier this year – and we’re elated to see a move towards proving cost-effectiveness.

Cost Reductions if US Population Average TIR Reached 70% and 80%

Using IQVIA’s Core Diabetes Model, IQVIA “conservatively” calculated cost reductions of $2.1-$4.2 billion over ten years if average population TIR reached 70%. These calculations were based on the assumption that the current average TIR for the US diabetes population is 58% - the authors rightly point out that this is the TIR seen in four large CGM studies, so it’s highly likely that average time in range is much lower than 58%. If average population TIR moved from 58% to 80%, the model estimates $4-$6.9 billion in total savings over ten years. Both of these cost reduction calculations came from improvements in complication rates. Further, if hypoglycemic events could be reduced by 40% in type 1s (often seen with CGM or AID), the paper estimated a further $2.8 billion in cost savings over ten years. Altogether, the model estimates that improving population TIR from 58% to 80% (type 1 and type 2), along with reducing rate of hypoglycemic events by 40% (type 1-only) could generate a total savings of $6.7-$9.7 billion over ten years!

  • The IQVIA Core Diabetes Model can only use A1c as an input (as opposed to TIR directly), so in order to calculate cost savings, TIR values had to first be converted to A1c values. This TIR/A1c conversion was calculated using two different linear correlations proposed by Drs. Roy Beck (2019) and Bob Vigersky (2019). With the Beck model, a TIR of 70% correlates to an A1c of 6.8% and with the Vigersky model, 70% TIR correlates to a 6.7% A1c. In other words, the calculated cost reductions from raising the population’s average TIR to 70% would be the same as those calculated if the population’s average A1c reached 6.7%-6.8%.

  • Unsurprisingly, the greatest cost reductions were seen in people with the highest baseline A1c levels. While a 5% increase in TIR for a patient with baseline A1c <7% is estimated to generate just $20 in per-person cost reductions over ten years, the same increase in TIR for a patient with baseline A1c >8% would generate $1,470 in cost reduction over ten years – a 74x difference! Ensuring that TIR and TIR-based interventions reach those with less well-managed glycemia will be critical to delivering the greatest health and cost benefits. We note that the picture below is only type 1 diabetes and does not appear to include hypoglycemia reductions, so they are also a cost underestimate.

What is the Population’s Actual Time in Range Right Now?

The paper’s estimate of 58% for population TIR average came from Beck et al.’s 2019 review of four major CGM studies: JDRF CGM Trial (2008), DIAMOND (2017), REPLACE-BG (2017), and HypoDE (2018). The mean TIR across those four studies was 58%, though we would note that those studies were made up of mostly white (>90%), low baseline A1c participants (7.5%) and took place at top diabetes centers – in other words, the actual population average TIR is likely much lower than 58%. Given that improving glycemia for patients with higher A1cs and lower TIR generates greater cost savings, we think these cost reductions noted above could be severely underestimated – especially the impact of reducing extreme BGs >250 mg/dl and <54 mg/dl. All that said, this is a valuable, conservative starting point from which to build – especially once more healthcare claims data is available from CGM users.

  • To date, Abbott has the largest set of real-world CGM data – n=592,328 readers (5.8 million sensors) –  with the median FreeStyle Libre scanner reporting a TIR of 56% (see poster 972-P at ADA 2019). While this is actually quite close to the 58% assumption used in the IQVIA White Paper, the estimated A1c in Abbott’s data set ranges from 6.7%-8.2% - i.e., still lower at the midpoint than the diabetes population’s average A1c. In other words, people on CGM who upload their data are probably a self-selected group already based on what we know about overall population A1c trends. To this end, a population TIR below 56%-58% seems likely.

  • The IQVIA White Paper cites the mean A1c of a cohort of adults with type 1 diabetes as 8.4%, coming from the T1D Exchange’s 2016-2018 update. With this value for population average A1c, we can use the same linear correlations used in the paper to estimate the population’s average TIR. Using Dr. Beck et al.’s model, an A1c of 8.4% corresponds to a TIR of 30%; with Dr. Vigersky et. al’s model, it corresponds to a TIR of 49%. While these numbers are quite different, they are both significantly lower than the 58% TIR estimate used in the paper.

    • Using a similar method, interpolation of Dr. Roy Beck and colleagues’ re-analysis of DCCT, would give a population average TIR around 39%. In that study, which used fingerstick data to calculate TIR, the intensive group recorded a TIR of 52% (A1c: 7.3%) and the conventional group recorded a TIR of 31% (A1c: 9.1%). Interpolating from these two data points, an average population A1c would correspond to an average TIR of 39%. Offhand, we think that sounds much closer to what is probably the actual population A1c but this of course is major speculation.

Approaches to Further Use of Time-in-Range

The paper, independently written by IQVIA with funding from Lilly, has eight pages on approaches to further the use of TIR at multiple levels, targeting healthcare policymakers, healthcare providers, and people with diabetes. The authors outline three “stages of maturity” for the use of TIR and key actions to help advance stages.

  • Establish the importance of TIR across stakeholders. A major part of this stage has already been achieved with the development of consensus TIR goals earlier this year, which were quickly incorporated into the ADA’s Standards of Care. To raise the awareness of TIR and CGMs, the paper suggests a campaign similar to the “Know Your Number” campaign that was designed to raise patients’ awareness of A1c. While evidence supporting the importance of TIR already exists (e.g. Beck et al., 2018), the authors propose the creation of a “TIR registry” that includes demographic, care, and outcomes data to further advance the importance and understanding of benefits of improved TIR. We believe pairing healthcare claims (cost) data with CGM data could be very powerful for building the case with payers.

  • Advance the importance of TIR and promote technology that enable TIR. The authors emphasize that the use of TIR is currently limited “due largely to the limited access to CGMs by the majority of [people with diabetes], who are treated predominantly by primary care physicians.” Developing an improved reimbursement framework and EHR integrations to support use of CGM data will help further use of TIR. We do point out that TIR can be determined by “connected” meters in those taking six or more measurements a day though this will become less and less popular. The paper also notes the potential for value-based contracts for CGMs to increase adoption. Lastly, educating nurses, diabetes educators, pharmacists, and others will be critical to support people with diabetes’ use of TIR. The move of CGM to the pharmacy in the US should certainly help with this – see Dexcom 3Q19 and Abbott 3Q19.

  • Perpetuate the use of TIR across all patient populations. Access to CGMs (and TIR) is largely limited to type 1s and a small, but growing, number of type 2s. Getting CGMs (and presumably connected care meters) to all people with diabetes will obviously require investment from payers and manufacturers. Easy-to-use devices and software must be designed for people of all backgrounds; to this end, Helmsley Charitable Trust funded the STEPP-UP project, to develop lower-literacy technology guides. The rise of mobile-delivered diabetes coaching could also bring CGM access to a wider audience, especially new wear models (e.g., Omada Health with Abbott, Onduo with Dexcom).


--by Albert Cai, Adam Brown, and Kelly Close