Dropout rates:
·
Through direct communication and omission of
material fact, Aegerion led investors to believe that the 2013 patient dropout
rate was a stable 15%.
·
Aegerion finally updated investors in late 2014,
claiming a “blended” dropout rate of 36%.
The stock price dropped 40% on the next day of trading.
·
To reach
that 36%, Aegerion distorted the rate downward
when it “blended” the patients
acquired in 2013 with new patients acquired in 2014.
·
If 2013
patient dropout rates are reliable and if such also applies to future first
year experiences, simple high school math shows that Aegerion’s relevant dropout rates are much worse
than we have been led to believe, most likely between 50% and 60%.Aegerion’s disclosures explicitly admit that the
dropout rate is material information, that they withheld this information, and
that they are continuing to withhold this information.If we accept Aegerion’s claims that the majority
of dropouts occur within the first two months of therapy, then we must abandon
our dropout estimate for first year patients in 2014 and we must accept that
Aegerion’s commercial enterprise is presently in the midst of an undisclosed
marketing crisis.
Solve the Aegerion Puzzle:
The following puzzle rests upon a few assumptions:
1.
That Aegerion did not misrepresent dropout rates
for first year experience with its drug in
2013.^{2}
2.
That that rate would serve as sample for the
first year experience of new patients in
2014. (Later, we will see what happens if we let go of this estimate.)
3.
Thus, dropout for first year experience in 2013
was 15%, and we use that percentage to estimate dropout for patients in their
first year in 2014.^{3}^{ }
4.
Given that Aegerion has not given investors an
update on patient counts, we use a revenue derived estimate of 508 for the
third quarter 2014.^{5}
All other data points have been provided by Aegerion or are
mathematically derived therefrom.
Skip to the next section to continue the report. For those
who enjoy puzzles, on
the right I present the kind of algebra required of any investor who wishes to
understand the true risks involved with Aegerion. (Scroll down for the solution.)

Aegerion dropout puzzle 
Basic Principle of Deception
Suppose an athlete ran the mile yesterday in 6 minutes.
Today he stopped halfway at 3 minutes. Could I really convince anyone that his
“blended” time of 4.5 minutes makes him a world class runner? Why wouldn’t I isolate yesterday’s distance
and time from today’s halfdistance
unless I wished to mislead someone? If I
wanted to include times for both days I must divide the data into three
distinct time periods.
Yesterday’s race

·
1^{st}
half of race: 3 minutes
·
2^{nd}
half of race: 3 minutes

Today’s race

·
1^{st}
half of race: 3 minutes.

Similarly, Aegerion is working with two distinct cohorts
over three time periods:
2013 Cohort

·
12
months: Patients acquired in 2013 tracked through 2013. The dropout rate
for the first year was said to be 15%.[2]
·
9
months: Patients acquired in 2013 tracked through the first nine months
of 2014.

2014 Cohort

·
9
months: Patients acquired in the first nine months of 2014 and tracked
through the first nine months of 2014.
If Aegerion’s claims with the first year of the 2013 cohort are
credible, we might use a 15% dropout rate for the 2014 cohort in their first
nine months. (See Endnotes 1 and 3 for a detailed explanation.[3])

But even without using specific numbers we know that
something is not right.
·
Because
dropouts occur over time, blending new patient
additions with old patient additions necessarily distorts the dropout average downward and misleads investors into
thinking that the problem is less than it really is.
Retailers, for example, use same store sales to isolate the mature portion of their business
from the immature. Blending this year’s newly opened stores with last year’s
established stores would necessarily distort the sales metric downwards.
Breaking down the case.
Aegerion’s Investors must resort to puzzlesolving in order
to fill in the missing pieces. In the puzzle
below, I will start with one set of inputs, as an illustration, but I encourage
the reader to estimate a range of
inputs and plug them into the equations. One will soon observe that even a wide range of estimates suggest that
Aegerion may have misled investors.
Numbers provided by
Aegerion
·
There were 430 patients in the USA by the end of
2013.[4]
·
The patient dropout rate through the first year,
2013, was 15%.^{2}
·
USA Revenue for the 3^{rd} Quarter 2014:
$39,767,000.^{5} (The USA portion was 91% of total revenue,
where product is sold month to month; international orders are multimonth
shipments and would unnecessarily complicate the estimates.)
·
Quarterly net cost of the drug at the end of
September 2014: $78,277 per US patient.[5]
·
The cumulative dropout for all patients over the
entire time period was 36%.^{1}
Numbers derived from
Aegerion’s reported numbers
·
Dropout rate for 2014 patients, 9 months into
2014 (Green Cell, between cells F and G):
A cumulative 15% dropout was claimed for first year patients at the end of
2013.^{2} If
Aegerion’s claim was credible, we can use this to estimate dropout for those
patients acquired in 2014. (See Endnote 3
for detailed considerations.)^{3}
·
Remaining
patients at the end of the cumulative (“blended”) time period from 1/2013 to
9/2014 (Last cell, Brown/orange, after cell J): Because Aegerion
stopped reporting patient counts, we derive an estimate by dividing the third
quarter 2014 revenue by the marketed price of the drug (IE, disclosed price
minus disclosed discount). Sales in the USA are month to month.^{5}
o
$39,767,000 in revenue for the quarter divided
by $78,277 cost per patient per quarter = 508 estimated patients at the end of
the third quarter, 2014.^{5}^{ }(“Value Insight” recently estimated
Aegerion’s patient count to be 533 in a SeekingAlpha article.[6])
Solving
the Puzzle that Aegerion has left investors
1. A  (A x .15) =
430
A = 430/.85
A x .15 = B
To say that 430 patients remain after a 15% dropout would be
consistent with saying that the 430 remaining patients were 85% of the total
patients introduced in 2013. A
x .85 = 430 is the same as 430/.85 = A. A = 506. Of 506 patients introduced,
15% dropped out: 76.
·
430/.85 = 506
·
506 x .15 = 76.
·
506 – 76 = 430.
2. I  (I x .36) = 508
I = 508/.64
I x .36 = J
Step 2 follows the same procedure as step 1. If 508 patients
remain after a reduction of 36%, then we would say that 508 was 64% or the original
total. I x .64 = 508 is consistent with
508/.64 = I. I = 794. 36% of 794 means
that we estimate 286 total dropouts over both 2013 and 2014.
·
508/.64 =794
·
794 *.36 = 286
·
794 – 286 = 508
3. I –A = F
If we have a total of 794 introduced to the drug over the
entire time period and 506 of them were introduced in 2013, then how many were
introduced in 2014? 288.
·
794506= 288 new patients introduced in
2014.
4. (F x .15) = G
F  G = H
Aegerion claimed a stable 15% dropout among first year
patients in 2013.^{2} If
this is reliable then in turn we use a 15% dropout rate for the first year
patients in 2014. (See Endnote 3
for detailed considerations.^{3}) If we’ve estimated that 288 patients were
acquired in 2014, we then calculate that 43 dropped out and 245 remained.
·
288 * .15 = 43
·
288 – 43 = 245
5. 508 – H = E
If there were a total of 508 remaining patients at the end
of September 2014, and 245 of them were from the 2014 cohort, then how many
from the 2013 cohort remained in 2014?
·
508 245 = 263.
6. J – G – B = D
Without “blending” in
new patients acquired in 2014, how
many of the patients who tried the drug beginning in 2013 dropped out in 2014? We
have a cumulative 286 dropouts by the end of the third quarter 2014. We
estimated that 43 of the patients acquired in 2014 dropped out in 2014, and 76
of the patients acquired in 2013 dropped out in 2013, which means that 167 of
the patients acquired in 2013 dropped out in 2014.
·
286 – 43 – 76 = 167
7. D / 430 = C
So what percentage of the patients who remained on the drug
at the end of 2013 dropped out by September 2014? Of the 430 patients remaining
on the drug by 2013, 167 dropped out in 2014. That is a 39% dropout rate for
the second year of use and in addition to the 15% reported dropout for the first year.
167/430 = .39
8. (D + B) / 506
If we isolated the patients acquired in 2013 and did not
“blend” in new patients from 2014, how many of the patients acquired in 2013
dropped out by September 2014? We
estimated that 76 dropped out in 2013 and 167 in 2014. That’s 48% of the 506
introduced in 2013.
(167+76)/506
·
It is easy to understand why Aegerion presented
the cumulative rate of 36% rather than rates from each isolated cohort. A responsible tally demonstrates that
Aegerion’s 2013 cohort dropout rate has accelerated to 50%.
Lots of room for error
Although the laws of mathematics are immutable, the results
are only as focused as the estimates we plug into them. Nonetheless, in Aegerion’s
case the room for error is very wide.
Two variables which are lacking but which would render the
puzzle a mathematically contained system are (1) the patient count as of
September 2014 and (2) the dropout rate for patients acquired in the first nine
months of 2014. If we found two reliable numbers here, the other numbers would yield
to the inexorable laws of mathematics. I invite the reader to plug in a wide
range of responsible estimates. It will become obvious that Aegerion has
clearly distorted the relevant dropout numbers by “blending” the 2013 cohort
with the newer patients acquired in 2014. There is a lot of room for error here:
# of estimated Patients
Sept 2014

Dropout rate over 21
months for patients acquired in 2013

400

41%

450

44%

508

48%

550

51%

600

54%

650

57%

There is some suspicion in regards to patient counts. For
example, Aegerion claimed to have 430 US patients on the drug at the end of
2013. However, deriving a patient count from revenue leaves us with an estimate
of 304.[7]
The difference, however, matters little.
Either number – 430 or 304  shows a severe loss of patients over time. If we
accept 430 as the number of patients at the end of 2013, then the dropout rate
for these patients over the 21 months to September 2014 would be approximately
50%; if we estimate there were approximately 304, then the dropout rate would
be approximately 60%.
I encourage the reader to take Aegerion’s disclosures and
attempt any responsible input.
Aegerion omitted these material facts in 2013 and continues to do so in
2014
Although Aegerion suggested, and still suggests, that they
know the dropout rate well enough to make stable projections they have not
provided investors with the relevant information.
The omission was deliberate. Here is a
rejection by Aegerion of a request for crucial data:
It is mathematically challenging for most dropouts to occur in the first
two months
Here are Aegerion statements about dropouts occurring within two months of therapy:
“Mirroring what we saw in our Phase III study, we see dropoffs happen most frequently during the first 1 to 2 months of treatment.” ~ Aegerion Pharmaceuticals Management Discusses Q4 2013 Results  Earnings Call Transcript^{}^{}
Craig E. Fraser  President of US & International and Global Supply: “…. As Mark mentioned, the majority certainly do come off early when they do drop, particularly between shipment 1 and shipment 2 and then some more between shipment 2 and shipment 3, and then it becomes a much lower factor as time goes on. So you do have a bit of a factor of time. Most of that is frontloaded, though, when you are getting the dropoffs. Marc, if you have any …”
Marc D. Beer  Chief Executive Officer and Director: “That's my answer, my complete answer.” ~ Craig E. Fraser, Aegerion Pharmaceuticals' (AEGR) CEO Marc Beer on Q4 2013 Results  Earnings Call Transcript
If
we accept that Aegerion’s disclosures in 2013 about patient dropouts were true
and that they serve as a reliable measure of future first year patients, then
the dropout for patients acquired in 2013 must
have accelerated through 2014. If I say that the first slice of a pie is
larger than the second slice, that second slice cannot be more than half of the
pie. A responsible set of inputs shows that Aegerion’s dropouts in the second
year exceed the first by more than 200%.
Returning to the 2013 cohort highlighted in this report, we estimate that
167 dropped in 2014, only 76 dropped out in 2013. Again, it is mathematically
impossible for the first slice of the 2013 cohort to be larger than the second
slice when that second slice is more than half of the pie.
76 dropouts / 12
months in 2013 = 6 per month.
167 dropouts / 9
months in 2014 = 19 per month
If I have a tendency
of 6 per month in 2013, how can I then say that it is more than the tendency of
19 per month in 2014?
To make Aegerion’s claim possible, at a minimum, the dropout
rate for the 2013 cohort in 2014 would
have to be less than it was in 2013,
not more.
Letting go of the dropout rate for the 2014 cohort
We shall see what can happen if we do not accept the 15%
dropout rate for the 2014 cohort. If we
take Aegerion’s claim that the majority of dropouts occur in the first two
months, then we would have to lower our estimated dropout rate for the 2013
cohort’s experience in 2014. As we saw
in the previous section of this report, it would have to be lower than the first year’s rate of
15%.
There are several problems with this.
First, if we lower the dropout rate for the 2013 cohort in its
second year to 10%, then in order to be consistent with the cumulative 36%
rate, we must significantly raise the
rate for the 2014 cohort  to nearly 60%.

Introduced to Juxtapid

Dropout rate

Dropout count

Remaining patients
 
2013 patients in 2013

506

15%

76

430

US Patient count as
claimed in 4th quarter & full year financial results

2013 patients in 2014

not
applicable

10%

43

387
 
2014 patients in 2014

288

58%

167

121
 
20132014

794

36%

286

508

US patients, Revenue
derived from 3rd Quarter 2014 income

       
Second, such an admission would mean that Aegerion’s current
growth story is false: it would actually be a marketing catastrophe in the
making.
Third, such an extreme and recent dropout acceleration would require specific disclosures of
both the sudden acceleration of the dropout rate itself and the factors
accounting for the extreme difference between this year’s experience and last
year’s. I have found no such disclosure.
Conclusion
1.
Aegerion’s use of a 36% “cumulative” rate distorts
dropout rates downwards. Dropouts are much worse than Aegerion has
made them appear. Relevant dropouts are probably
closer to 50% or 60%.
2. Aegerion’s
failure to disclose the relevant dropout rate constitutes an omission of
material fact.
3. If
we accept Aegerion’s claims that the majority of dropouts occur within the
first two months of therapy, then we must abandon our dropout estimate for first
year patients in 2014 and we must accept that Aegerion’s commercial enterprise
is presently in the midst of an undisclosed marketing crisis.
What is the purpose of disclosing patient metrics if not to
aid the investor in evaluating the business? Of two metrics available why
provide the one which decreases awareness of risk unless one’s aim is to
mislead investors?
In this last reporting period, Aegerion “blended” its 2014
patient data with ongoing data from patients acquired in 2013 and claimed a
“cumulative” dropout rate of 36%. Using
basic algebra, one can deduce the reason why different cohorts and time periods
were “blended” together: isolating patients acquired in 2013 and following them
from the onset to the first 9 months of 2014, we calculate a relevant dropout
rate between 50% and 60%.
Aegerion had previously led investors to expect the dropout
rate to be 15%. Being illprepared for even the 36% blended rate, investors
sold off their shares. The stock dropped over 40% on the day following the
announcement. Obviously investors regarded this news as material. And given the
facts presented in this report, it is clear that Aegerion has omitted material
facts in previous reporting periods and is continuing to do so now.
Sources:
[2] 2013 10% and 15% dropout rates:
“Approximately six months into the U.S. commercial launch of
JUXTAPID, the company has experienced a patient dropout rate of less than 10
percent …”
In the fourth quarter 2013 report, it updated the number:
[3]
Cumulative dropout rates have been accelerating as time passes. Thus, it is
probable that, contrary to Aegerion’s claims, dropouts within a single cohort increase over time as adverse events
accumulate. We follow this logic in the
reconstruction of what may be happening with Aegerion’s 2014 dropouts. If the
first year experience in 2013 was 15%, we use this rate to estimate the new
patients in their first nine months in 2014. With these estimates we try to
fill in the blanks: What is happening with the second year on the drug? What
can we say about the 2013 cohort over 21 months?
Because dropouts are obviously accelerating and because the
numbers for 2014 represent a shorter time period, at first glance it is
tempting to use a percentage smaller than 15% to estimate dropouts for the 2014
cohort. However, a second glance
suggests that 15% might be the better number after all. 15% is made up of an extremely small sample
of two elements. Dropouts for the first
six months of 2013 was said to be 10%.
The cumulative rate for all of 2013 was said to be 15%. This means that dropouts for the second half
of 2013 must have been around 20%. (10%+20%)/2
= 15%. In 2014 we are only working with
9 months, so it would seem reasonable to work with a number between 10% and
20%, given that dropout appears to be accelerating as time passes. Note however
that if we do lower the dropout rate below
15% (green cell in the puzzle), then the dropout for patients acquired in 2013
and tracked in 2014 (Cell C) increases in proportion. For example, a 10%
cumulative dropout for patients acquired in 2014 calculates a 42% dropout in
2014 for those patients who were acquired in 2013, while 15% calculates 39%. 15% here puts Aegerion in the better light
when we get to the final tally for the 2013 cohort.

Introduced to Juxtapid

Dropout rate

Dropout count

Remaining patients


2013 patients in 2013

506

15%

76

430

US Patient count as
claimed in 4th quarter & full year financial results

2013 patients in 2014

not
applicable

42%

181

249


2014 patients in 2014

288

10%

29

259


20132014

794

36%

286

508

US patients, Revenue
derived from 3rd Quarter 2014 income

On the other hand, if this dropout rate is higher than 15%, then either Aegerion’s
2013 disclosures are suspect or Aegerion has not fully disclosed new factors in
2014 dropouts.
[5] Patient
Count estimate for 3rd Quarter 2014:
“Net product sales for the
thirdquarter ended September 30, 2014 were $43.7 million, compared with $16.3
million in the thirdquarter ended September 30, 2013. Net product sales for
the nine months ended September 30, 2014 were $106.7 million, compared with
$24.1 million for the nine months ended September 30, 2013. 91% of net product
sales in the third quarter of 2014 were from prescriptions written for U.S.
patients, while 9% came from exU.S. countries, primarily named patient sales
in Brazil.” ~ Aegerion Pharmaceuticals Announces ThirdQuarter 2014 Financial
Results http://ir.aegerion.com/releasedetail.cfm?ReleaseID=879469
Price Hike mentioned in
Jefferies presentation:
Discount:
Mark J. Fitzpatrick: “Sure. The gross to
net in the quarter is running at 5%, Cory. That's pretty consistent with
what we're seeing yeartodate as well. It's still a little on the early side
to assess our government commercial impairments and to be able to assess the
Medicaid rebate in total, but we estimated conservatively as we book these
things month by month, quarter by quarter, so it is a 5% number for the
quarter and the yeartodate.” Aegerion Pharmaceuticals Management Discusses Q3 2013 Results  Earnings
Call Transcript
http://seekingalpha.com/article/1790502aegerionpharmaceuticalsmanagementdiscussesq32013resultsearningscalltranscript
To estimate
a patient count for the 3^{rd} quarter, 2014, we divided the revenue by
the net cost of the drug:




43,700,000

Revenue 3rd Q '14






39,767,000

91% US month to month sales





3,933,000

9% ExUS multimonth shipments

(Price was hike up twice since the launch)








329,587

5% discount

313107.7

4 quarters =

78,277

Per quarter per patient






39,767,000/78,277=508


