Free Expert Report - District Court of Delaware - Delaware

Case 1:05-cv-00059-JHR Document 23-7 Filed O4/26/2005 Page1 0f4
EXHIBIT 5

_ Case 1 :05-cv—OOO59-JHR Document 23-7 Filed O4/26/2005 Page 2 of 4
Academic Radiolony, V¤I11,No 12, December 2004 · · l" ‘. n - n n ` 1 n '=`. n - -‘ `LETTERS T0 THE ED¥TDR
irritialt reading was not compietely negative. However, the Laura S- Welch, MD, FAGP FACOEM
authors calculated their odds ratios of the differences be- Adjunct Professor
tween the initial and consultant B readers on the premise George Washington University Schooi of Public hearth and
that only if all 6 consuitants read the tibn as not completely Heatth Services
negative was the radiograph considered not compietely Washington, DG
negative by the consultants. lite odds ratios of the differ— Meolicai Director
ence would be appreciably less if the 143 radiographs Ggritgr to pyotoot Workers Rights
that were read as not completely negative by at least 3 of Stlvor Spring, MD
the 6 consultants were considered to be not completely
¤<·>g2liZiV€· ` Michael Fl. Harbut, MD, MPH, FCCP
Even with the re—calculation of the statistics in a more Ghgor _
ssuitabis manner, differences, th<>¤sh evt as Strikers, witi center ret occupational and environments: medicine
remain between the initial and the consultants readings. F-Qoyal Oak, M;
The process of selecting B readers for this type of study, C¤_D§,.&CtOr
and haw the data are emitted S0 the Gutters are slimi- retreat: ester at vaimeutrs site Asbestos-Related
nated, are of the utmost importance if we are to address Cammrs
the problem of how B readings are used in litigation and Detroit MI
which individual truly has an abnormal radiograph. Karmanos Cancgi, Institute
As a B reader and epidemiologist for 20 years, who chica; Assésmm Professor
bas mainly (> 90% of the time) read chest radiographs {mama, Medicine Wayne State University
for research studies and governmental programs, I share Bayou, M1
the concern expressed in the editorial on whether some-
thing is rotten or not and the need to have independent
studies performed to assess this issue. To the Editor:
In their article on the comparison of B readers interpreta-
tion of chest radiographs for asbestosrelated changes (1),
Authoids disclosure:. l ` l l _ -- · Y ‘ Gitiin and co-authors present the results of a rnultiaeader
comparison of radiographic readings for pneumoconiosis,
. The author has received fest from mth Piaimiffs and and conclude that “there is not support in the literature on
defense attorneys as an expert witness and for performing X_,.3y Studies Of workers exposed to asbestos and other
mdepmdcm medical BXami”m°nS` mineral dusts for the high level of positive findings re-
_ _ corded by the initial readers in this report?
’ nereaaucas · . ¤‘ ‘· - · ‘ - ` We take issue with many of the points in their article,
1. ewan Jn, cook tt., Linton ow, and earreremayer E. compresses of but wouid tike here to focus on the methods used in the
Z;;;‘;f":;;'g*‘°§§;?;?;?)%i;°;;igi;‘_;;§’g'“p“$ fm a$bES*°S"""md study they discuss. The study, done at Johns Hopkins
l 2. Janower Mt, and Berlin i,., "B" Readers radiographic interpretation in Medical Center, COt`lSiS£Gd of & 1"C—t`GHdiItg of chest t‘Hdi0—
ggziiifflgjffgigi '5 $°”“€"“"g '°“°" i" mg °°“""°°"'? Asst iéml graphs of plaintiffs in asbestos litigation, originally read
by doctors (B readers) selected by plaintiffs attorneys.
The rereading was done by a panel of B readers seiected
From, by the study’s authors; the study was done at the behest `
W ‘ of attorneys for defendants in asbestos iitigation. We ask
L Chrisum Owen M5, MS the following questions:
l Assistant Clinical Professor of Medicine
5-toward Moogool gotqoog l. I-low were the "consultant" readers selected by the
Boston, MA study’s authors'? There is no basis to judge the inde~
E President pendence of the consultant B readers. The paper
Occupational Health initiatives states that the consuttant readers were "cornpensated
Brookline, MA for their participation." Was this by defense attor-
E isa?

Case 1 :05-cv—OOO59-JHR Document 23-7 Filed O4/26/2005 Page 3 of 4
_ LETTERS T0 THE EDYTOH -‘__ n - n ‘ -_ ` in · '‘_. -‘ Z 0 Academic Radiology, Voi_11, No 12,. December 2004 -.
neys? It so, does that not create th appearance, if Table
not the reality, Of Conmct on the pm Of the Consult} Kappa Statistic for Agreement among Consultant Readers
ants who reread the radiographs? Kappa Statistic for
2. How was thc set of 551 cases chosen? Since these Agtéemém a“"*°“9 KaF’Pa
films were of individuals who had retained an attor-
ney in asbestos litigation, we can presume that all Film quality D.10(t}.0S-—U.12) Poor
of the films had been interpreted as showing some ls film G°mPl asbestos-related disease. This is confirmed by Tania ga Ally p&'9“°l‘}’mal _
_ _ _ _ abnormalities consistent
in Gitlin and colleagues’ article. There are thou- with pmumocomcsis? 0*31 {O-2B_U_33) Pwr
Sands cf Such cases filed each year- if these 551 Small Opaque-S we or greater? D.‘19(G.1B-0.21) naar
cases were ones consecutively received by a spe» Any pleural abnormalities
cilic asbestos trust fund, they could he considered °°”SlSi°“l Wllll _
representative. The authors have an obligation to pn8um°°°nl°SlS? DAQ l0'47`U`52) Fall
describe the method used for choosing the case
sample for the study, rather than stating that they
wwe no Idea how mg Samplc as chosclgllq addl In the study described by Gitlin and colleagues, there
tion, the reason for the reduction of the initial sam- , _ , _ _
pie Of 551 to the Hm} amiytjc Sample Of 492 is not is a clear difference between the initial reader and the
Chan consultant readers in the overall classification of films as
_ _ _ _ _ _ completely normal or not. However, additional statistical
3. Because this is a selected series of positive radio- . ,‘
_ _ _ _ analyses showed poor to fair agreement among the con—
gféphs Submited for c(iinpBnS?m0n’ H IS not _3ppm' sultant" readers for chesoradiograph tilm quality, any pa-
Rmm E0 conclude that lhcm IS mt Support in the renchymal abnormality, and profusion category. Adding
htafatum lm Xlmy Stlldms Of workers cxpljscd to the initial reader to the measure of agreement did not
asbestllsf and other mineral dum tm _th_€_h1gh hwg} change the value of the ec statistic for film quality, and
gf lxlsmva indian esetdel by its nasal Radars only changed it slightly for the finding of whether or not
Z m mls YGPQYEP The Fgmpic tells us seems about El the film was completely negative and for the iinding of
me sf l°°S“lV€ fmdmgs ammg an EXPOSEG Wmkgrs pleural abnormalities consistent with pneumoconiosis. In
S°f‘°“m°d· sr °V°“ the l°mp°m°“ wad as pmltivc by essence, the initial reader was in the same ballpark as the
; this Select gm'}? Of initial B Yeadcm The Smdy Sim"' consultant readers for many of the important parameters
Z PIY dccs lim have any densmlsatsr dam WGH which evaluated. Prior literature documents that there is a great
EU msc Suse 3 Conclusion- deal of variability in classiiication of radiographs using
gr- Alllmslt lhz Consullmll readers diseased with mc the International Labour Office (ILO}1980 system. But
` llllllal ¥€€ld€¥S» Hwy diseased among lhcmselycs as one cannot really conclude from this analysis that the
W€ll· FOY many nf the ¤0mP3Il$°¤$ made lll FMS 939% consultant readers were any more accurate than the initial
¥h€ ¤ST€€m€¤~l amimg me €0¤$l·llmm$ was at best fair readers. An analysis of each reader in comparison to the
E {T¤l¤l€>}- In m¤ld¤% @*6 kind Gl <>¤mP¤¤`i$<>¤$ math in other five consultant readers would have been illustrative
this study, some amount of agreement is expected by (3), to show the range of agreement among the consultant
_ chance alone. The K statistic adjusts for this chance readers 35 wg;] 35 with the initial made;
i agreement. Kappa is positive when agreement is better
E mm chance (3 `~’¤l¤€ of l Bflllals °0mPl€l€ 3B¥°’€m€l"*l}· 5. The ILO classification system was not designed to be
. is ZBYU when aéllwmtml Equals what Would be mls used as a diagnostic test, and the diagnosis of ashesto—»
i?>€€l¤d by ¤h3·¤¤€· and is ilcgallyc when Egfaemlml is sis requires more than a radiographic interpretation.
WOYSG than W0¤ld b¤= €XP€<5l€d by ¤l'l3¤¤€- Landis mid The American Thoracic Society recently issued a new
( Koch (2) have suggested that K values greater than set of giidelines for the clinical diagnosis of non—1na-
Q about 0.75 represent excellent agreement, values be- lignant lung disease related to asbestos (4). We agree
tween 0.40 and 0.75 indicate fair to good agreement vnth Gitlin and colleagues that classification based on
Z and values below about 0.40 represent poor agreement a a single radiograph is subject to observer variability,
i beyond chanoe. and should not he used as the only criterion for a di~—
§ tees

Case 1 :05-cv—OOO59-JHR Document 23-7 Filed O4/26/2005 Page 4 of 4
Academic Radiology, Vol 11, No 12, December 2004 -. n i n n .`.` s .I ‘ ‘:. ° I-fc'. .`.' 'I...E'¥’TEElS__TO THE EDITORV
agnosis of as’oestos·~related disease. It is impomt for To The Editor:
the physician diagnosing asbestosis not to rely on Gitlin and colleagues have revisited the problem of vari-
chest raehographs alone, and in fact the Association of ability in the interpretation of radiographs obtained for the
Occupational and Environmental Clinics has stated purpose of detecting asbestoswelated abnormalities (1).
that such an act is unethical (5). There are additional The problem that they address has persisted for too long,
and medically necessary steps to reach a diagnosis of and is worth roexamining sc long as it persists.
asbestosis. However, we cannot support the conclu— The documented iindings are expected. First, th au~
sions of Gitlin and colleagues study because of the thors used a methodology that is not designed to under- -
potential bias in the selection of the study sample, the state the problem, beginning with films interpreted by B
potential bias on the part of the consultant readers, and readers retained by plaintiffs’ attorneys. Second, very sub~
the clear high rate of variability among the consultant starltial differences in reader behavior were inferred be-
readers. tween readers in our much larger study two decades ago
This earlier study was DOE conducted in the context of
legal proceedings, and relied on very large numbers of
_Auih0rs’ disclosure statement: randomly distributed packets of radiographs obtained on
behalf of the U.S. Navy (and its substantial industrial op-
Tlw 3¤*hUTS of WS 1€‘»¤€Y have dmc medic-a1~¥¤~=sa¥ eratiorls, including shipyards). Despite the large numbers
work in the area of asbestos—related disease in the form of Of gyms {O; Bach made; and the absence Og a regu; Setting,
H`l€d`iC3.l YCPOKS and t€StiH'lO{ly f{)1’ patients S€€E] in th€-if it was Clem {hat epidemics Could appeal- and disappear
clinical practice, and in some cases for individuals not depending 0+; the Chorus, Of reader (2)
examined- The major t><>¤i<>¤ Of this Wvrk has been for The impact of B-reader behavior was atsaaaaat sia,
plaintiffs’ attorneys. For this work we have received con- amd my unidgmc;g0m]_ For 23 hgghwciume B-r€ad€rS’
sultation fees. Our research in the area of asbestosreiateci Navy dam abnormalities 2 yr) ranged from g_g5% {O
disease has not been funded by attorneys, nor have other ]g_g3% (median: ]_71%)_ and from 002% {0 755% fm-
relevant activities, such as testimony b€i°01‘B the U.S. Cow 2] [ 1, encompassing Several orders of magnitude (4).
ares or the drafting cf published suirielinss for the diss- white *·ava-raaaiagr is emphasized in the article by on-
· ¤¤5i$ slid ifaatmmi of ¤Sb€$Y0SiS· lin and colleagues and the accompanying editorial by
Ianower and Berlin (3), the failure to find existing disease
is generally regarded as the most serious problem in med-
1. Gillin JN, Cook LL, Linton OW, and Garretbtviayer E. Comparison of "l33" ical Surveillance €ffOnS· Because the normative film is
readers' interpretations of chest radiographs for asbestos related must; ggmpleg will bg ngmqai, gm "gvg;·-yg=;3_d3;·" may apa
2. EZZ2iL”$tiii°1£`2l§T'$2;§J£Z§iZQ?i?`tsttNt. egreerrtenttor tat- ssa me sulrstis ties ¤¤ "¤¤ds¤-¤¤¤ l german data. Biometrics %977; ss;1ss#x201e;174. statistical analyses that both we and Gitliri and colleagues
a W 3 `S§.iE2,}‘$j‘i.Z‘§2§.2%$'t§§"§Z',T2?EtiTt'i?'{Li;E;.i?J?§'ilE2i1£§$?.Z’§i2..E of sprite Bat ¤Ve·t<=s<*1¤s end ¤¤de·rs¤dmg as impe-
cneumoconioses using the 1980 International Labor Organization ciassi- Wim P1'0bl€Hl$L dlwidillg Whliili is {Eli? THOYB culpable •‘?·I`1“01’
ficatlon system for pneuomoconioses. Chest.1998; M4; 1740-1748. is a mana!. Of Pcrspcctiva Quamy assurance is a good fm.
rl. American Tlooraorc Society. {Diagnosis and inmat management ot nonma~
lignant diseases relatred to asbestos, Am J Respir Crit Care Nleci 2004; the EOOSE as WGH as {OY the gandmx
l6¤¤69l·?l5- It has been clear for two decades that the social utility
j 5. The Association ot Occupational and Environmental Clinics Policy State- . .
· mem on Asbestos Scragnm mw of the B-reading program as currently configured 1S prob-
p httpcffwww.avecorg/asbest¤s~scree¤-htm lematic. The assertion that the International Labour Oflice
system was "designed for research" is merely a failure to
address the social problem. That failure in tum affects the
From: health beliefs of our patients. The B-reading process is
also refetred to in U.S. Occupational Safety and Health
g Atan l\/E. Ducatrrtan, MD, it/ISc Administration law (l9l0. 1001). It is time to stop pre-
Professor anol Chair tending that the testing examination alone provides suffi-
i Department ot Community Medicine West Virginia cient quality assurance to address the needs of exposed
i University School of Medicine populations. The solution attributed to me by Gitlin and
; Morgantown, WV colleagues (1), of multiple readings, is likely to be an
; 1399