City of Huntington, West Virginia v. AmerisourceBergen Drug Corporation et al

Filing 1458

MEMORANDUM OPINION setting forth reasoning for the ruling that charts created by Dr. Craig McCann to summarize certain datasets are admissible under Rule 1006; overruling defendants' objection and admitting the charts. Signed by Senior Judge David A. Faber on 7/9/2021. (cc: counsel of record who have registered to receive an electronic NEF) (jsa)

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 1 of 26 PageID #: 62598

IN THE UNITED STATES DISTRICT COURT
FOR THE SOUTHERN DISTRICT OF WEST VIRGINIA
THE CITY OF HUNTINGTON,
Plaintiff,
v.

Civil Action No. 3:17-01362

AMERISOURCEBERGEN DRUG
CORPORATION, et al.,
Defendants.
v.

Civil Action No. 3:17-01665

CABELL COUNTY COMMISSION,
Plaintiff,
v.

AMERISOURCEBERGEN DRUG
CORPORATION, et al.,
Defendants.
MEMORANDUM OPINION
During the testimony of Dr. Craig McCann, plaintiffs
introduced charts that Dr. McCann created to summarize certain
datasets.

Plaintiffs contend that these charts are admissible

under Federal Rule of Evidence 1006.
they are not.

Defendants contend that

According to defendants, the summaries are merely

demonstrative exhibits, not substantive evidence.

On May 12,

2021, the court conditionally admitted the summaries under Rule
1006 and ordered briefing on the issue.

On May 18, 2021, the

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 2 of 26 PageID #: 62599

court ruled that the summaries are admissible under Rule 1006,
thereby overruling defendants’ objection and admitting the
charts. 1
I.

The court now sets forth its reasoning.

Background
a. Dr. McCann
McCann received a Ph.D. in economics from the
University of California at Los Angeles in 1989.
Throughout the 1990s, he taught graduate-level
economics and finance courses at the University of
South Carolina (1987 to 1992), Virginia Tech (1995 to
1997), Georgetown University (1996), and the
University of Maryland (1995 to 1998). In addition,
he served as an academic fellow for the U.S.
Securities and Exchange Commission.

In re Nat’l Prescription Opiate Litig., 2019 WL 3934490, at *7
(N.D. Ohio Aug. 20, 2019) (citations omitted).
McCann explained that during the time of his doctoral
studies, “some of [his] peers were what we would call
theoreticians and others empiricists.”
10, ECF No. 1333.)

(May 10, 2021 Trial Tr.

McCann describes himself as an empiricist.

Of his approximately 35 peer-reviewed papers, the “vast majority

The objection that the charts are improper under Rule 1006 is
overruled. This is the only issue that the parties were asked
to brief. This opinion does not deal with separate objections,
such as the objection that certain charts include pharmacies
outside the geographic scope at issue (Cabell County and the
City of Huntington). If the court later determines that certain
summaries (or certain portions thereof) are irrelevant, the
court will excise them from its consideration. But the court
does reject the argument that by including pharmacies outside
Cabell County and the City of Huntington, the summaries are
skewed or otherwise inadmissible under Rule 1006.
1

2

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 3 of 26 PageID #: 62600

of them” are “empirical” and involve “gathering data from
different sources, reporting summary statistics, generating
tables, figures, charts.”

(Id. at 13-14.)

McCann indicated

that those empirical papers tend to go “beyond the simple data
summary of the type [presented] here with the opioid data” by
including “some statistical analysis or some valuation.”

(Id.)

McCann has testified in the context of state or federal
litigation approximately 150 times, and in the context of
administrative proceedings or arbitrations approximately 450
times.

On May 10, 2021, this court found McCann to be “an

expert on data processing, validating, supplementing,
reconciling and summarizing large datasets as they relate to
ARCOS and the related governmental datasets that [plaintiffs]
used to supplement them.”

(Id. at 19-20.) 2

b. ARCOS Data
The court has previously described the Drug Enforcement
Administration’s Automation of Reports and Consolidated Orders
System (“ARCOS”) database.

(See ECF No. 1236, at 1-4.)

Generally speaking, ARCOS is a system that receives reports
regarding the flow of controlled substances, including opioids,
among different members of the supply chain.

ARCOS can be

Obviously, McCann’s expertise in processing large datasets is
not limited to ARCOS data and includes, for instance, the
transactional data that defendants produced in discovery.
2

3

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 4 of 26 PageID #: 62601

thought of as a “portal” through which the DEA gathers data
regarding opioid shipments.

(See ECF No. 1333, at 21.)

The

repository of data that the portal collects can be referred to
as the ARCOS data.
The ARCOS data here is a subset of the entirety of the
ARCOS data in existence.
only 14 drug codes.

It covers only 9 years (2006-2014) and

The DEA produced this raw data in

“tranches,” with the production of the last tranche taking place
in approximately August 2018.

(Id. at 21.)

The raw ARCOS data

amounts to “roughly 500 million records of shipments,” which, if
printed, would fill up approximately 20,000 banker’s boxes.
(Id. at 28.)
There is no dispute that the ARCOS data is admissible. 3
There is also no dispute that it is unworkable in the manner in
which the DEA produced it.

The only way to comprehend this

admissible evidence in a meaningful way is to summarize it. 4
This is so at least as to the ARCOS data that Dr. McCann
summarized for this case. Plaintiffs’ attempt to enter (in the
form of a flash drive) the entirety of the ARCOS data that the
DEA produced was met with an objection on grounds that the data
was unworkable, cumulative, and beyond the geographic scope of
this case. (Id. at 39-43.) In a later colloquy, defendants
stated that they had no objection to the admissibility of the
subset of the ARCOS data that McCann summarized. (May 11, 2021
Trial Tr. 124-25, ECF No. 1342.) Defendants stated the same as
to the transactional data produced in discovery that McCann also
summarized. (Id.)
3

Defendants assure the court that a ruling in their favor on
this issue “would not keep ARCOS data entirely out of the case”
4

4

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 5 of 26 PageID #: 62602

c. Defendants’ Transactional Data
ARCOS data is not the only source of information pertaining
to the quantity and frequency of opioid shipments from
defendants into plaintiffs’ jurisdictions.

Defendants’ internal

records of these shipments, which they produced in discovery, is
another source.
ARCOS data.

These records go beyond the timeframe of the

AmerisourceBergen produced records for June 2002 to

December 2018.

(ECF No. 1333, at 81-82.)

records for 1996 to May 2018.
records from 2004 to 2018.

Cardinal produced

(Id. at 85.)

(Id. at 88.)

McKesson produced

The McCann charts

summarize these records (the “transactional data”) to the extent
that the charts at least partially cover time outside the 2006
to 2014 ARCOS window or, rarely, when there are slight gaps in
the ARCOS data. 5
because defendants have stipulated to its authenticity and
plaintiffs can use it “to examine any witness with whom
[plaintiffs] are able to lay a proper foundation.” (See ECF No.
1347, at 3 n.1.) This assurance is cold comfort. The
opportunity to piece together fragmented witness testimony
(assuming the admissibility of that testimony) concerning this
evidence is a far cry from being able to consider the evidence
itself (in summary form). See 5 Christopher B. Mueller & Laird
C. Kirkpatrick, Federal Evidence § 10:34 (4th ed.) (“Details
might emerge, but not the whole shape of things—the trees may be
clear enough, but not the forest.”).
As the most notable example of such a gap, the ARCOS data does
not include the relevant transactions for Cardinal for a single
month (March 2008), but the data that Cardinal produced in
discovery (the transactional data) does include them. (Id. at
50.) Such gaps in the ARCOS data account for significantly less
than 1% of the data, among the 3 defendants. (See id. at 49.)
5

5

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 6 of 26 PageID #: 62603

d. McCann’s Summaries
The

McCann produced summaries in the form of charts. 6

charts tabulate or illustrate a morass of data to make it
comprehensible.

These charts do not purport to summarize all of

the ARCOS data that the DEA produced.

Instead, for this case,

McCann narrowed his subject of summary to only the ARCOS data
that bears on this case.

He did not include distributions, for

example, bound for export or research, or to the state of
Washington (except to the extent such shipments are included in
national figures and averages), or to hospitals.
Just as McCann does not attempt to summarize all of the
ARCOS data, so too, he does not limit his summaries to the ARCOS
data alone.

Specifically, many of his summaries capture both

the ARCOS data and defendants’ transactional data produced in
discovery.

It is therefore both too narrow and too broad to

think of all of McCann’s charts as strictly ARCOS data charts.
Some are summaries of a subset of the ARCOS data, while others
are summaries of a subset of the ARCOS data, plus the
transactional data.

Some charts summarize shipments to

individual pharmacies; others, shipments to multiple pharmacies;
and still others, shipments into geographical areas.

Some

The court will use “charts” as shorthand for charts, graphs,
and tables.
6

6

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 7 of 26 PageID #: 62604

charts focus on distributions by defendants (or a single
defendant), while others include all distributors.

Many compare

distributions into different geographical areas.
II.

Discussion
Defendants argue that Rule 1006 does not embrace the McCann

charts for two main reasons.

First, they argue that what McCann

created are not proper Rule 1006 charts because (1) he used
information not available on the face of the raw ARCOS data
(such as U.S. Census Bureau data and information from the
Centers for Disease Control and Prevention) to calculate and
convey information such as per capita figures and morphine
milligram equivalent (“MME”) figures; and (2) he narrowed his
subject of summary and made certain corrections to the data.
Second, they argue that the charts are skewed.

The court is

unpersuaded by both arguments.
Rule 1006 provides:
The proponent may use a summary, chart, or calculation
to prove the content of voluminous writings,
recordings, or photographs that cannot be conveniently
examined in court. The proponent must make the
originals or duplicates available for examination or
copying, or both, by other parties at a reasonable
time and place. And the court may order the proponent
to produce them in court.
Fed. R. Evid. 1006.
Federal courts have wide discretion in determining whether
to admit evidence under Rule 1006.
7

United States v. Blackwell,

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 8 of 26 PageID #: 62605

436 F. App’x 192, 198 (4th Cir. 2011); United States v.
Milkiewicz, 470 F.3d 390, 398 (1st Cir. 2006) (appellate review
“highly deferential”); 31 Charles Alan Wright & Victor James
Gold, Federal Practice and Procedure § 8044 (2021) (“[T]he
convenience standard of Rule 1006 gives trial courts significant
discretion.”).

This discretion, however, is not unbounded.

First, the proffered summaries must meet certain basic
requirements.

They must be “an accurate compilation.”

States v. Janati, 374 F.3d 263, 272 (4th Cir. 2004).

United
They must

be based on records that are admissible and that have been
provided to the opposing party.

Id. at 272-73.

they must be produced in court.

Id. at 273.

And sometimes

Thus, there are

dual safeguards to ensure against bad summaries:

(1) a

threshold accuracy requirement and (2) a procedure to ensure the
opposing party an opportunity to bring inaccuracies to light.
See id. at 272-73.
Properly understood, the threshold accuracy requirement is
a modest one.

Charts are admissible despite mistakes.

United

States v. Hemphill, 514 F.3d 1350, 1359 (D.C. Cir. 2008) (“Even
if the calculations are mistaken, the chart is itself
admissible, since admissible evidence may be unpersuasive and a
defendant has the opportunity to rebut it.”).

Charts are

admissible despite their encompassing less than all the
underlying data.

United States v. Blackwell, 436 F. App’x 192,
8

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 9 of 26 PageID #: 62606

199 (4th Cir. 2011); United States v. Bentley, 825 F.2d 1104,
1108 (7th Cir. 1987) (“[C]harts need not be encyclopedic to be
admissible.”); Gill v. Arab Bank 893 F. Supp. 2d 523, 536
(E.D.N.Y. 2012) (Weinstein, J.).

And there is an important

distinction between an inaccurate chart and a chart that
supports only one side of the story.

See United States v.

Lynch, 735 F. App’x 780, 786 (3d Cir. 2018) (“A Rule 1006
summary chart need not accurately reflect all the facts in the
case; it merely must accurately represent the facts that it
purports to summarize.

‘So long as they are accurate, however,

such summaries may present only one party’s side of the case.’
The fact that G-94 did not include all the related accounts does
not undermine its admissibility because it accurately reflected
the accounts it did include.

It was up to Lynch, on cross-

examination, to show the jury that there was more to the story,
and he effectively did so.” (citations omitted)).
Besides these basic requirements, Fourth Circuit precedent
cabins the wide discretion that trial courts have under Rule
1006, in either direction.

On one hand, when proffered

summaries meet the basic requirements of Rule 1006, a district
court in this circuit abuses its discretion by cutting off both
the summary path and “the long way,” especially in complex
cases.

See Janati, 374 F.3d at 274.

On the other hand, a

district court abuses its discretion by admitting a proffered
9

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 10 of 26 PageID #: 62607

summary under Rule 1006 that amounts to “a skewed selection of
some of the [underlying] documents to further the proponent’s
theory of the case.”

United States v. Oloyede, 933 F.3d 302,

311 (4th Cir. 2019) (emphasis in original).

Together, Janati

and Oloyede act as guideposts for when it is permissible to
exclude a proffered summary and when it is permissible to admit
one.

This court must navigate between these guideposts by being

cognizant of the importance of summary evidence (especially in
complex cases) but on guard against skewed summaries.
The Janati opinion was the result of an interlocutory
appeal in a “complex healthcare fraud prosecution,” where the
government sought from the appeals court “the necessary
evidentiary latitude to prove its case.”

See 374 F.3d at 266.

The relevant question there was “whether the government [could]
use Rule 1006 charts in its case-in-chief to present its
experts’ conclusions about fraud with respect to some 1300
transactions alleged in furtherance of the conspiracy.”
270.

Id. at

The defendants there objected to the admission of the

charts on the grounds that the charts were “independent evidence
and not a summary of the evidence,” that the government did not
really need the charts, and that the summaries would
impermissibly contain opinion evidence.
original).

10

Id. at 272 (emphasis in

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 11 of 26 PageID #: 62608

In rejecting these arguments, the appeals court stressed
the need for summary evidence in complex cases.

Id. at 274-75.

In the absence of a definitive ruling by the trial court on the
admissibility of any chart, the appeals court found no abuse of
discretion.

Id. at 273.

But the appeals court made clear that

if the district court exercised its discretion to exclude the
proffered summaries, the district court “must accord the
government the right in some manner and to some reasonable
extent to prove its case the long way.”

Id. at 274.

Thus,

Janati stands for the principle that when presented with
admissible voluminous evidence and an acceptable summary of it,
a district court that rejects both abuses its discretion (at
least to the extent the excluded evidence is necessary for a
party to make its case).
Oloyede is the other side of the coin.

The appellants

there argued that the summaries admitted against them
“represented an ‘editorialized subsection of [banking]
transactions.’”

Oloyede, 933 F.3d at 310.

The appeals court

agreed that the summaries were the product of manipulated bank
records:
For example . . . while [the FBI forensic accountant]
included all cash deposits and all ATM deposits to the
extent that bank records did not identify whether it
was a cash or check deposit, he included wire
transfers only to the extent that the bank statement
identified the name of the sender and that name was on
a list provided to him by the case agent. Similarly,
11

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 12 of 26 PageID #: 62609

he included check deposits only to the extent that the
check was written by an individual on the list and
out-of-state teller transfers only to the extent that
a receipt associated with the transfer was included as
an attachment in a batch of emails that [the
accountant] was provided.
Id. at 310 (emphasis in original).

Accordingly, the summaries

represented “a skewed selection of some of the documents to
further the proponent’s theory of the case.”

Id. at 311.

“They

did not fully represent the accounts that they were purportedly
summarizing.”

Id. at 310.

Thus, the charts masqueraded as one

thing (comprehensive summaries of certain bank accounts), but
were in fact another (summaries of only some transactions).

The

purpose of the charts in Oloyede was to help the jury comprehend
a suspicious pattern based on selected transactions, not to
convey voluminous records to the jury in an intelligible form.
Id. at 311.

Because the charts had the airs of

comprehensiveness, the purpose was a hidden one.

See id.

The

appeals court drew the line at this sort of misleading chart.
See id.
This court must first determine whether the McCann charts
meet the basic requirements of Rule 1006.
is undoubtedly voluminous.

The underlying data

See United States v. Aubrey, 800

F.3d 1115, 1130 (9th Cir. 2015) (“Multiple bankers’ boxes of
bank statements constitute the type of materials anticipated by
Rule 1006.”).

The defendants received the underlying data with
12

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 13 of 26 PageID #: 62610

reasonable time to review it. 7

No party has requested that

plaintiffs produce the underlying data in court.
underlying data is, all parties agree, admissible.

And the
The

contention is over whether the charts have the proper inputs and
whether they are skewed.
a. The “Four Corners” Argument
As to the inputs, defendants say that the summaries run
afoul of Rule 1006 by “concededly go[ing] beyond the four
corners of the ARCOS database.”

(ECF No. 1347, at 7.)

They say

that this is contrary to the rule that Rule 1006 summaries may

At least one federal appeals court has deemed disclosure of the
summaries themselves unnecessary. See Colon-Fontanez v.
Municipality of San Juan, 660 F.3d 17, 30 (1st Cir. 2011). And
at least one federal appeals court has suggested that
fundamental fairness requires such disclosure. See Fid. Nat.
Title Ins. v. Intercounty Nat. Title Ins., 412 F.3d 745, 753
(7th Cir. 2005). Commentaries tend to support the latter view.
See 31 Charles Alan Wright & Victor James Gold, Federal Practice
and Procedure § 8045 (2021) (explaining that providing the
summaries to adverse parties is necessary to fulfill purpose of
Rule 1006 because, without the summaries, “adverse parties
cannot know what to look for in the source material to determine
if the summaries are accurate.”); 8 Graham, Michael H., Handbook
of Fed. Evid. § 1006:1 (9th ed.) (opining that not to require
disclosure of summaries “is ludicrous and clearly defeating of
the entire purpose of Rule 1006—to permit employment of a
summary, chart or calculation only upon providing the opponent
the opportunity to effectively explore objecting to its
admissibility and/or being placed in a position to effectively
meet the evidence being introduced.”); Krakauer v. Dish Network
L.L.C., No. 1:14-CV-333, 2016 WL 6775859, at *5 (M.D.N.C. Sept.
19, 2016). There appears to be no question, however, that
defendants received the McCann charts with ample time to review
them, check their accuracy, and prepare to cross-examine their
author.
7

13

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 14 of 26 PageID #: 62611

not “contain information not present in the original.”
Universal Furniture Int’l, Inc. v. Collezione Europa, USA, Inc.,
599 F. Supp. 2d 648, 656 (M.D.N.C. 2009), aff’d, 618 F.3d 417
(4th Cir. 2010).
The Universal Furniture rule regarding information that is
not in the original is easily overstated.

Delving this rule to

the root, which appears to be Standard Oil Co. v. Moore, 251
F.2d 188, 223 (9th Cir. 1957), is almost to watch it disappear.
In Moore, the plaintiff introduced, and the trial court
admitted, certain documents in support of Moore’s alleged
damages.

Id. at 222-23.

These included Moore’s tax returns (as

purported summaries of a business ledger) and summaries based
upon those tax returns.

Id.

The Ninth Circuit Court of Appeals

held that these documents were inadmissible as summaries because
they “contain[ed] computations which [were] not reflections of
ledger entries” and were never offered “as summaries of [the
plaintiff’s] accounting records.”

Id. at 223.

The tax returns

were hearsay documents that happened to contain some figures
reflecting admissible business records, and the summaries of the
tax returns had the same admissibility flaws.

See id. at 222-

23.
In Universal Furniture itself, the problem was not that the
purported summaries referenced outside information to translate
figures into different units of measurement or narrow data into
14

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 15 of 26 PageID #: 62612

a subset, but rather, that the author of the summaries had
included certain “averages” in them without a proper foundation.
Universal Furniture, 599 F. Supp. 2d at 656.

More bluntly, the

averages in the purported summaries were “unexplained and
unreliable.”

Id. at 658.

And in United States v. Drougas, upon which Universal
Furniture relies, the trial court only initially rejected the
proffered summaries, and even then, only “because they contained
argumentative inferences,” not because their author had
referenced government data or publications to determine a subset
or make the underlying data more comprehensible and the charts
more informative.

See 748 F.2d 8, 25-26 (1st Cir. 1984).

To the extent that the Universal Furniture rule disallows
proffered summaries like tax returns, or those containing the
products of unexplained and unreliable calculations, or those
laden with argumentative inferences, the rule is a sound one.
But defendants’ broad reading of the rule, which would
essentially force authors of Rule 1006 summaries to ignore all
information outside the four corners of the underlying
materials, goes too far.
What McCann did with information outside the four corners
of the underlying materials was permissible.

He applied

statements in government publications to narrow his subject of
summary and to make the data more comprehensible.
15

Defendants

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 16 of 26 PageID #: 62613

incorrectly characterize the addition of categories for MME and
per capita figures as “embellishments.” (See 1347, at 7.)
McCann did not so much add to the underlying data with the
government sources as use those sources to translate the data
into additional categories, for better comprehension.
“[n]othing [was] lost in the translation.”

And

See United States v.

Bray, 139 F.3d 1104, 1110 (6th Cir. 1998).
The inclusion of MME figures on the charts poses no
admissibility problem.

McCann applied information from the CDC

to convert the dosage strengths into MME figures.

After the

conversion, the summary is still of the ARCOS and transactional
data.

Nothing was added.

units of measurement.

It is just expressed in different

Stated differently, the raw data always

expressed the MME figures, just in different terms.

To the

extent defendants wish to challenge the CDC’s conversion
instructions, or McCann’s application of those instructions,
they had the opportunity to do so on cross-examination. 8

The

above analysis applies equally to the Food and Drug
Administration’s published list of drug labelers.
used the FDA key to unlock distributor names.

McCann simply

The names were

always there, just in number form.

They will also have the opportunity to present evidence in
their case in chief that the CDC or McCann is wrong about the
MME figures, to the extent such evidence exists.
8

16

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 17 of 26 PageID #: 62614

The data from the Centers for Medicare and Medicaid
Services also poses no problem.

McCann did not add this

information to the data he summarized.

Instead, McCann used

this information to better identify the buyer pharmacies that
would be the subject of his subset.

Because there is no general

prohibition on summarizing a subset, the use of this data to
help formulate a subset was permissible.

To the extent

defendants believe that the CMS data is wrong or that McCann
wrongly applied it, the appropriate remedy is cross-examination
or the production of defendants’ own charts.
McCann’s use of the Census data is a little different, but
still poses no problem.

To include per capita figures, McCann

had to do more than simply apply an abstract formula.
apply historical facts:

He had to

the number of persons within various

geographical boundaries during various times.

He used U.S.

Census data, which provides reliable estimates of these
historical facts.
Census data.

The court may take judicial notice of U.S.

See United States v. Cecil, 836 F.2d 1431, 1452

(4th Cir. 1988) (“[C]ourts may take judicial notice of official
governmental reports and statistics.”).

Thus, it does not bog

down the summaries with inadmissible material, as the purported
summaries did in Moore.

Neither does bringing Census data to

bear on the summaries result in unexplained, unreliable

17

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 18 of 26 PageID #: 62615

calculations or introduce argumentative inferences.

So even if

McCann added something here, the addition was permissible.
Defendants conclude that the portions of the charts
providing MME and per capita figures represent McCann’s expert
opinion.

But “[a] calculation does not constitute a conclusion

or opinion.”

United States v. Honeywell Int’l Inc., 337 F.R.D.

456, 459 (D.D.C. 2020).
rely are unavailing.

Moreover, the cases on which defendants

The first case dealt with expert testimony

that a party offered under Rule 1006.
994 F.3d 123, 138 (3d Cir. 2021).

United States v. Shulick,

The appeals court noted,

Judge Bartle then specifically asked defense counsel
whether they intended to offer Hamilton as an expert.
Defense counsel responded that they would not make any
further disclosure and that Hamilton would be offered
solely as a summary witness.
But when Hamilton was called to the stand at the
opening of the defense case, he attempted to offer
expert testimony. Specifically, Hamilton, a CPA who
acknowledged that he had previously served as an
expert “hundreds of times,” was asked why it was
appropriate to allocate certain shared costs across
Shulick’s business ventures to the budget for
Southwest—a post-hoc analysis, involving the
application of Hamilton’s own formulas and judgment to
facts. The District Court ultimately excluded this
testimony, accepting the Government’s argument that it
was undisclosed “expert opinion based on his years of
accounting experience, based on his forensic
background, and based on information that is not
available from the records.”
Id. at 138–39 (emphasis added) (record citations omitted).
The second case involved an expert who collected random
samples of phone calls from a collection of hundreds of millions
18

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 19 of 26 PageID #: 62616

of phone call records.

United States v. Dish Network LLC, 75 F.

Supp. 3d 916, 927, 933, 939 (C.D. Ill. 2014).

Doing so

“required him to use his expertise in statistical analysis and
was not a mere compilation of data.”

Id. at 933.

He also

“employed his expertise in analyzing large data sets” “to
calculate a valid Established Business Relationship.”

Id.

The

court determined that the “random samples” were “not a mere
compilation of data.”

Id.

The plaintiffs there argued that the

proffered expert’s “testimony and reports” were admissible as
summaries as an alternative to showing the expert was qualified
under Federal Rule of Evidence 702.
also stated legal conclusions.

See id. at 932.

Id. at 933, n.5.

The expert

The court

determined that some Rule 1006 did not cover “much of his work.”
See id. at 933.
Dish Network should be understood by reference to its
specific facts, not as a broad prohibition on charts that
require any expertise to create.

It was not just that expertise

was needed to process a large amount of data; rather, the
testimony and reports in Dish Network had inputs themselves that
could not have been determined without expertise.
922-23.

See id. at

The distinction is important because if the use of any

expertise in creating a chart renders the chart inadmissible,
then databases that by their sheer size require some level of
expertise to process into charts would never be admissible,
19

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 20 of 26 PageID #: 62617

rendering Rule 1006 self-defeating.

Ironically, under

defendants’ interpretation of Rule 1006, when the rule is needed
most (when the underlying evidence is most voluminous), it would
not apply.

Such a reading of the rule does not comport with the

purpose of Rule 1006 or with persuasive authority.

See

Honeywell, 337 F.R.D. at 459; United States v. Bentley, 825 F.2d
1104, 1108 (7th Cir. 1987) (charts using “statistically valid”
sampling of trading data admissible to show company’s “net
position in silver and copper futures” admissible); Sec. & Exch.
Comm’n v. Lek Sec. Corp., No. 17CV1789 (DLC), 2019 WL 3034888,
at *3 (S.D.N.Y. July 11, 2019) (describing as “classic summary
evidence,” testimony, tables, and charts conveying calculations
involving analysis of market data); United States v. Pomrenke,
198 F. Supp. 3d 648, 706-08 (W.D. Va. 2016); Sec. Inv. Prot.
Corp. v. Bernard L. Madoff Inv. Sec. LLC, 610 B.R. 197, 225
(Bankr. S.D.N.Y. 2019).
b. Arguments Regarding Narrowing and Corrections
Defendants also contend that the McCann charts are out of
bounds because they do not cover all of the ARCOS data.

They

say McCann erred by narrowing his subset to categories he deemed
relevant and by excluding what he deemed errant entries.

This

contention involves two quite different alleged deficiencies.
The relevance argument falls flat.

What defendants make much of

as “relevancy determinations” are no more than McCann’s
20

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 21 of 26 PageID #: 62618

narrowing of his subject of summarization. 9

Subject to the

limitations of Oleyede, there is generally no problem with
summarizing a subset of a dataset.

Making subsets off limits

under Rule 1006 would seriously undermine the purpose of the
rule, and the text does not require reading the rule to be selfdefeating.
Defendants’ specific suggestion that McCann narrowed his
subset to exclude (the relatively infinitesimal) 10 R (return)
transactions because he wanted to “maximiz[e] Defendants’
overall shipment volume” is baseless.

(See ECF No. 1347, at 4.)

First, because they are so small, McCann’s exclusion of the R
transactions does very little (and in most charts, nothing) to
accomplish his purported goal.

Second, in their

characterization of McCann’s motives, defendants ignore the
reasoned explanation that he gave for not including the R
transactions in his subset:

The problems with this category

“But what we did was, we took that data, narrowed it
down to shipments from manufacturers, to distributors, to
dispensers, and then checked internally that data to find
some potential data errors.” (ECF No. 1333, at 18.)
9

As McCann testified concerning the R transactions, “[T]he
magnitude of them is really quite small. You’ve
identified one month where there are perhaps 20 or 25% of
that month’s shipments but, overall, there were a truly,
truly tiny percent of the overall shipments.” (May 11, 2021
Trial Tr. 115, ECF No. 1337.) And further: “[T]hese R
Transactions account for approximately six tenths of 1% of the
shipment data and of the -- of the shipments from wholesalers
and distributors to dispensers.” (ECF No. 1342, at 114.)
10

21

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 22 of 26 PageID #: 62619

outweigh its usefulness.

Because it is often unclear which

distributor originally shipped the (later) returned pills, and
because the changes in almost all the charts would be
nonexistent or imperceptible, whatever completeness might be
achieved would be overshadowed by the uncertainty surrounding
the R transactions.

(ECF No. 1342, at 114-15.)

Moreover, to

the extent defendants feel that their cross examination was
insufficient to establish the purported significance of the R
transactions, they are presumably free to present their own
competing charts with the R transactions in their case in chief.
The corrections to the ARCOS data are a separate inquiry.
Some exclusions that McCann made are more in the nature of
corrections.

Dr. McCann excluded duplicate transactions,

erroneous transactions, transactions with obvious errors in
quantity, and transactions reported as lost in transit.
Additionally, McCann corrected miscalculated weight figures.
In some cases, corrections to underlying data may render
summaries inaccurate and inadmissible.
case.

But this is not such a

McCann was completely transparent about the corrections

he made, and the corrections were appropriate.

Perhaps most

significantly, however, the corrections were microscopic, and no
magnification of them renders the charts inadmissible.
McCann testified as follows regarding these corrections:
I’m going to talk a little bit about these -22

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 23 of 26 PageID #: 62620

these corrections or adjustments that we made. They
don’t really deserve the weight of the testimony I
give them because they’re on the order of one
hundredth of 1% or two hundredths of 1%, but for
completeness, I describe the corrections that we made.
(ECF No. 1333, at 22.)
And the corrections account for something like less
than a tenth of 1%, maybe less than five one
hundredths of a percent. So, . . . what I summarize
here is ARCOS data with very minimal, very minimal
corrections, trivial.
(ECF No. 1342, at 112.)
Defendants’ argument appears to be that any correction to
summarized material, no matter how miniscule, infects the
material such that no summary based upon it is admissible.

Rule

1006 is not so impractical.
c. The “Skewed” Argument
Finally, defendants argue that the charts are inadmissible
because they are “skewed” under Oloyede.
problems.

They identify two

First, they say that the charts impermissibly feature

“a cherry-picked set of non-representative pharmacies that have
no geographical nexus to the City of Huntington or Cabell
County” in an attempt to show “bad behavior.”
9, (quoting ECF No. 1342, at 55-56).)

(ECF No. 1347, at

Second, defendants renew

their contention regarding the exclusion of R transactions.
(See ECF No. 1347, at 10 (“Dr. McCann excluded all transactions
other than sales to customers.”).)
the charts are skewed.
23

Defendants fail to show that

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 24 of 26 PageID #: 62621

In making the first argument, defendants conflate the
selectivity of the individual charts with selectivity of the
underlying data used in formulating the charts.

Had McCann

relied on only the selected pharmacies in formulating average
shipment numbers, for example, the summaries could be skewed.
It could also be a problem if McCann had held out the pharmacies
as representative of other pharmacies in general.
not.

Each chart is what it purports to be.

But he did

And no chart

presents the product of manipulated data, as in Oloyede.

There

is no requirement that plaintiffs seek to admit summaries of
only “representative” pharmacies.

So long as the allegedly

high-volume pharmacies are not the product of manipulated data,
they are admissible.
The “matrix” of the various pharmacies appears to be the
only chart that comes close to being susceptible to the charge
of being skewed.

This chart lists shipments to pharmacies both

inside and outside Cabell County and the City of Huntington.
McCann’s understanding is that the selection of these pharmacies
were “not intended to be selected randomly” but instead were
meant to show excessive shipments to pharmacies.
55-56.)

(See ECF No.

In other words, McCann understood that plaintiffs

wanted him to include these pharmacies on the matrix because
they were examples of alleged “bad behavior.”

24

(See id.)

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 25 of 26 PageID #: 62622

The problem with defendants’ argument is that the matrix
does not purport to be random or representative.
misleading, as the chart in Oloyede was.

It is not

Although the

pharmacies were apparently selected to show examples of
excessive shipments, the figures for those pharmacies are not
skewed.

That other pharmacies may have received fewer opioids

means that other charts may help complete the overall picture,
but it does not mean that these charts are skewed.

Again,

defendants presumably can present those other charts that they
think are necessary to round out the picture.

But there is no

problem with plaintiffs’ use of charts to produce evidence
regarding certain pharmacies so long as those charts are not
masquerading as something they are not.
The argument regarding the exclusion of the R transactions
also fails.

As explained above, it was permissible for McCann

to focus on a subset of the ARCOS data.

The charts do not

purport to account for net shipments after returns.

Moreover,

given the extremely small fraction of the ARCOS data for which R
transactions account, and given McCann’s reasonable explanation
for why he excluded them, the exclusion of the R transactions is
far different than the exclusion of select banking transactions
in Oloyede.

25

Case 3:17-cv-01362 Document 1458 Filed 07/09/21 Page 26 of 26 PageID #: 62623

III. Conclusion
For the reasons expressed above, the court OVERRULED the
objection that the charts are improper under Rule 1006 and
admitted the charts into evidence under Rule 1006.
The Clerk is directed to send a copy of this Order to those
counsel of record who have registered to receive an electronic
NEF.
IT IS SO ORDERED this 9th day of July, 2021.
ENTER:
David A. Faber
Senior United States District Judge

26