Monthly Archives: March 2015

More thoughts on Cologuard reimbursement

Disclosure: Short EXAS.

A careful reading of the logistic regression in the Cologuard algorithm demonstrates that a positive result on the FIT component alone is sufficient to trigger a positive result on Cologuard.

After more than four months, I finally received a disappointing response from CMS to my FOIA request for copies of the other public comments submitted regarding the payment level for code G0464, which covers Exact Sciences’ Cologuard test. Had the FOIA request been processed within the statutory timeframe, I had hoped to incorporate counterarguments to any other concerns brought up by other stakeholders in my reconsideration request. As it turns out, that was unnecessary—there were none. CMS disregarded cost-effectiveness concerns on its own.

I was naïve enough to believe that CMS’ decision not to adjust its preliminary payment determination must have been at least partly guided by competing public comments. Indeed, this is what the former director of the Division of Ambulatory Services in the Hospital and Ambulatory Policy Group, which runs the CLFS rate-setting process, told me via email:

The comment period is an opportunity for commenters to share their thoughts with CMS.  We do not respond individually to these comment letters.  We take the information into consideration as we come to our final determinations.  And, as you can imagine, often there are competing comments provided to us. [...] 

In addition, we do not share proprietary information if it was shared with us in order to support a payment determination for either a crosswalk or a gapfill payment amount – that may or may not have been the case here.  In coming to the final determination for this test code, as well as all of the others that were brought through the Annual Public Meeting this year, we carefully considered everyone’s comments, but that does not mean that everyone was satisfied. 

I presumed that Exact Sciences must have submitted the purportedly more favorable cost-effectiveness study it has referenced in investor meetings but has never released to the public. (Investors learned on Exact’s Q4 conference call that its cost-effectiveness models are substantially identical to those used by CISNET, which determined that the test would not be cost effective nearly 8 years ago). At a minimum, surely Exact must have argued more than just the crosswalk rationale they presented at the public meeting in July.

As it turns out, there were only three comments submitted after the preliminary payment determination by CMS. One was my own. Another comment was from the manager of a well-known $400 million hedge fund. [Note: I've redacted his personal contact information by request]. His email concisely makes an important statistical point I did not reference in my own correspondence. Because Cologuard includes both a FIT component and a DNA methylation assay component, there is not actually sufficient evidence that Cologuard’s (very expensively reimbursed) DNA methylation assay adds any incremental effectiveness beyond its FIT component alone. Indeed, a careful reading of the logistic regression in the Cologuard algorithm demonstrates that a positive result on the FIT component alone is sufficient to trigger a positive result on Cologuard. In other words, if enough blood is present in the stool sample, Cologuard returns a positive result, irrespective of scores on the DNA assays.

If Cologuard has any sensitivity over and above its proprietary FIT component at all, it would be for detecting CRC-positive patients who for some reason do not have much blood present in their stool but have nevertheless begun to shed a significant amount of DNA. This is a point I plan to emphasize in presenting my reconsideration request at the upcoming public meeting this summer.

Against these two thoroughly documented critiques of the reimbursement level for G0464, Exact Sciences submitted a single-page letter thanking the Division for agreeing with its crosswalk proposal in the preliminary assessment. That was the full extent of Exact’s public comment on the issue. No pharmacoeconomic data was submitted to address cost-benefit concerns. Incidentally, Exact’s letter was dated October 30, two days after my comment letter was published and Cable Car’s website saw a spike in traffic from Madison, Wisconsin. If anyone at Exact Sciences is still reading: publish convincing evidence that Cologuard is cost-effective at its current reimbursement rate, and I will withdraw the reconsideration request.

Short selling can often feel like swimming against the tide, especially when regulatory agencies are involved. In this case, I have yet to hear a substantive response from Exact, CMS or any payor regarding the concerns raised in either letter. CMS is of course not obligated to act solely on the basis of public comments, but it is disheartening that the regulatory process did not directly address the merits of our commentary. Perhaps more concerning is that the politics of providing a ‘favorable’ reimbursement to the first test to go through joint CMS-FDA review appear to have outweighed any fiscal concerns. A different arm of CMS stated in its proposed coverage decision:

As permitted by §1861(pp)(1)(D) of the Act, we may consider the appropriate frequency and payment level in determining whether to expand coverage for new CRC tests. CMS has commissioned such analyses in all past determinations (FIT, stool DNA, and computed tomography (CT) colonography) and will re-analyze using the test parameters of Cologuard [emphasis added].

Yet Cable Car has recently learned that CMS and CISNET have yet to commence any new cost-benefit analysis of Cologuard. Why not?

Cable Car has a further FOIA request slowly wending its way through the queue (the FOIA staff at CMS have my sympathies—they are hard-working but underfunded), in hopes of understanding why NGS set a much lower payment level in one jurisdiction than the one in which Exact’s lab is located.

I look forward to discussing the appropriate reimbursement level at the public meeting.

Hanergy and the Shanghai-Hong Kong Stock Connect 2

Disclosure: No position in 566 HK.

Northbound short selling has been allowed for six trading sessions on the Shanghai-Hong Kong Stock Connect (沪港通) and so far not a single A share has been sold short. It’s not yet possible at Interactive Brokers, so apparently other brokers are also still figuring out the settlement logistics.

Hong Kong-listed shares, on the other hand, remain fertile hunting ground for short sale candidates. But seller beware! The influence of Shanghai’s frothy, retail-driven stock market may be beginning to be felt. One line in a WSJ article about the recent unexplained and unjustifiable rise of Hanergy Thin Film (566 HK) caught my eye: “The company’s share price has nearly quadrupled, to HK$6.80 on Friday, since the trading link opened on November 17, making it the best-performing and most widely held stock in the program known as Shanghai-Hong Kong Stock Connect.” To what extent might retail interest have driven such a massive change in the valuation of a loss-making solar panel manufacturer whose accounting and relationship with its parent has long drawn scrutiny from shortsellers?

First, an anecdote

I am reminded of a conversation I had with my uncle in Shanghai earlier this year, a few days after the Shanghai Composite fell nearly 8% in one day. Not to pick on him, but my uncle is the quintessential retail daytrader. He and my aunt have long enjoyed telling family about their occasional speculative stock trades, but as the Chinese A-share market has heated up, they have become more serious about actively trading their savings. My 16-year-old cousin proudly told me about her first real-money three-bagger last fall, which she had picked at random. A sign of the times, perhaps?

Since I invest for a living, we were naturally very interested to hear one another’s perspectives on the stock market. It was fascinating to learn about the range of financial products and securities lending-related innovations available to retail accounts in the mainland, many of which are quite complex. But I must confess; the conversation suffered from more than the usual language barriers. It wasn’t just that I don’t recognize the Chinese name for, say, Bollinger bands (“布林线”) off the top of my head—we were speaking entirely different languages. I told him about the meeting we had had with NetDragon the day before and how I think the business trades well below asset value despite exciting growth prospects over the next 3-5 years. He responded by explaining how he only buys a stock when it has had at least 3 “red” days in a row, and that after a couple of “green” days it was time to sell. (In China, the color red is considered good luck, so the red/green negative/positive association the rest of the world uses is inverted—a small part of my motivation for choosing neutral blues for Cable Car’s logo! See the tickers in the photo below from January.)

View of stock tickers in the Shanghai financial district January 2015

View of stock tickers in the Shanghai financial district January 2015

To his credit, my uncle understands the dangers of trading on margin, unlike much of his competition, and he is quite thoughtful and careful about his position sizing. But like most of his fellow retail traders in China, he is not evaluating stocks on any sort of fundamental basis. He is a momentum trader and a speculator with a long bias, and he knows that eventually what goes up must come down.

Back to Hanergy

In my opinion, this sort of non-fundamental trading creates opportunities for fundamental investors with a longer time horizon. Conversely, it can also be a reason why mispricing persists in markets that remain dominated by retail investors. Estimates vary, but Shanghai trading is thought to still be at least 50% retail-driven, with Hong Kong around 30% according to INSEAD. In US markets, it is often clear when an illiquid, low-capitalization security has been pumped by uninformed retail speculators. There is typically a paper trail of newletters or cheerleading posts on message boards. When it comes to Hanergy, I may not be looking in the right places, but I haven’t seen the usual signs of a pump-and-dump in Chinese media.

It’s possible that interest may be arising through boiler room-style marketing by brokerage firms, but it seems at least plausible that the simple fact of the recent rise in the share price has created a momentum bubble that feeds on itself. My experience with Chinese retail investors is that they like to buy stocks that have gone up recently. On the surface, it seems improbable for a company with a US$36 billion market capitalization and meaningful float to be influenced by retail volume. However, retail interest—especially through the Stock Connect Program—has contributed a significant amount of the trading volume during the rise. Southbound trade in Hanergy has exceeded 25% of trading volume on some days, and it has represented net purchases (buy volume minus sell volume) every day except one since the beginning of February.

The chart below shows the percentage of total trading volume represented by estimated gross Stock Connect volumes as the price has increased.

Southbound Stock Connect volumes have represented 7-26% of daily Hanergy trading volume since February

Southbound Stock Connect volumes have represented 7-26% of daily Hanergy trading volume since February

For better or worse, I have been unable to secure a borrow on Hanergy, so this is more of an academic exercise for me. But it is fascinating to witness the formation of a valuation anomaly and speculate as to its cause.  Perhaps the bubble will continue as long as mainland retail money continues its net inflow into the stock.


The raw daily data below is from FactSet and the HKEX Daily Top 10 Southbound securities, available for trading sessions beginning February 2, 2015. *Estimated from February monthly figure.

Date Buy
(HKD m)
(HKD m)
(HKD m)
Est. Stock Connect Volume
(m shrs)
Total Volume
(m shrs)
Stock Connect %
2-Feb 13 6 7 $3.61 5 95 5.6%
3-Feb 18 12 6 3.64 8 80 10.1%
4-Feb 19 13 7 3.77 9 116 7.4%
5-Feb 67 21 47 4.09 22 187 11.5%
6-Feb 39 18 21 4.20 14 127 10.6%
9-Feb 34 23 11 4.22 13 79 16.9%
10-Feb 46 21 24 4.31 16 91 17.3%
11-Feb 102 25 77 4.51 28 167 16.9%
12-Feb 75 7 67 4.47 18 116 15.8%
13-Feb 55 8 47 4.53 14 104 13.3%
*17/18-Feb *10 *18 (8) 4.35 7 73 9.0%
25-Feb 58 7 51 4.48 15 60 24.4%
26-Feb 25 9 16 4.52 8 63 12.1%
27-Feb 48 7 42 4.51 12 64 19.0%
2-Mar 123 7 117 4.69 28 107 25.9%
3-Mar 138 37 101 5.10 34 178 19.2%
4-Mar 239 64 175 5.92 51 414 12.3%
5-Mar 491 337 154 7.58 109 583 18.7%
6-Mar 250 182 68 6.84 63 322 19.6%
9-Mar 154  66 87 6.65 33 152 21.7%
Total 2,005 887 1,118 $5.50  506 3,180 15.9%