MOVING DXCM to the BENCH
"Oy, Tom. I think your data must just be wrong. I don’t understand how you get your data, but ... the Dexcom numbers have spoken"
The quote above is from a client back in November 2019 after DXCM beat 3Q19, and I doubled down on more data and more analysis. Into the 3Q19 print, DXCM shares traded at $150 with a 9x EV/NTM Sales, and the consensus 2020 revenue estimate was $1,618M. Since then, the shares have doubled to $300, carry an EV/NTM Sales of 14.5X, and consensus 2020 is 10% higher (+$164M) at $1,782M.
The arithmetic of the +100% move in the stock price shows that the change in the estimate was worth $15, while the multiple added $135 per share. Effect size of small estimate changes seems to be a thing these days. Wrong is wrong regardless of the path.
Iterate or die | Yet another Claims analysis or just hurry up and die already?
After 3Q19, we expanded the features of our claims data set for a deeper understanding of our error. We found and spoke to physicians at large practices who had come up a steep CGM adoption curve. We segregated patients into T1 and T2 diabetic buckets, and tested multiple variations in an attempt to overcome the forecast limitations despite the availability of details on 10 million distinct patients per week.
As a data source, medical claims were never built to do what we are asking them to do. Yet compared to the alternatives such as physician and consumer surveys, app download data, and physician interviews, it seemed like a good bet. We came up short. We're not quitting.
The latest version we're holding out was developed with a reference data set, and we've moved to a method we hope improves the utility of the data in the case of DXCM and others. The correlation is impressive at 0.98, but there are still many important reasons to qualify the data. Combined with the app download data, anecdotes, and expectations, the data should be helpful as a directional tool, but point estimates are likely still beyond the use case.
Over the last 8 quarters, on average the estimate was within 5% of reported. Excluding the 2H18, the estimate was within ~3% from the actual.
TRYING A DEXCOM g6 ON myself, WIll people really use this thing?
On the 4Q19 call, management highlighted the increase in patient adoption in 2019 and the runway remaining in the current insulin intensive population, but also their intent of serving 30 million diabetics, and beyond. Management stated that "...work with UnitedHealth Group has progressed very well, and we look forward to the expanded use of DexCom CGM in their Type 2 populations in 2020. In addition to multiple partnerships with credible partners." Given the restrictive language in the UNH medical guidelines for CGM use, this is a big deal.
Others see a vast pool of patients beyond the insulin intensive diabetic patient population, but primary literature, calls with doctors, and managed care requirements suggest something more modest. I decided to to try a G6 to see for myself the impact of wearing a CGM device, and I am now more inclined to believe the upside scenario. I also need to lose a few pounds and trying the device for myself seemed like it could help.
Here are my conclusions from wearing a G6...
- Applying the device and getting started was very easy.
- I turned off the alarms in the first 5 minutes of using the device, which was not immediately obvious.
- I understand now what physicians meant by "sugar surfing," or compulsively checking the app reading, and why this may be a problem for some users.
- Watching my blood glucose levels is compelling. Health entertainment is a thing and likely something cool for the worried well.
- After 24 hours of use, I can already see that my diet will likely change over time as a result of seeing my blood glucose readings.
- A banana spiked my blood sugar, but remained stable when I ate blueberries and almonds. Furthermore, I did not eat a chocolate chip cookie, which is an incredible bit of behavior modification
- I don't expect to wear a CGM much longer than a few weeks, but checking in periodically as a routine part of medical care seems reasonable to me at this point.
FUNDAMENTAL EQUITY QUAD
Estimates are likely to continue to move higher from here. Consensus revenue sits slightly above the top end of management guidance for the year. Our fundamental equity quad method sorts estimate trends into 4 quadrants of growth and acceleration. Across a large number of companies and compared to many other style factors, fundamental equity quads back test well across all 4 Hedgeye Macro environments, or Macro Quads. Dexcom is likely to remain a Quad 2 stock as estimates trend higher despite the sharp deceleration in growth in 2020.
