Our reference: PRP 51195 P-authorquery-v9

AUTHOR QUERY FORM

Journal: PRP Please e-mail or fax your responses and any corrections to:

E-mail: corrections.esch@elsevier.thomsondigital.com

Article Number: 51195 Fax: +353 6170 9272

Dear Author,

Please check your proof carefully and mark all corrections at the appropriate place in the proof (e.g., by using on-screenannotation in the PDF file) or compile them in a separate list. Note: if you opt to annotate the file with software other thanAdobe Reader then please also highlight the appropriate place in the PDF file. To ensure fast publication of your paper pleasereturn your corrections within 48 hours.

For correction or revision of any artwork, please consult http://www.elsevier.com/artworkinstructions.

Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags inthe proof. Click on the ‘Q’ link to go to the location in the proof.

Location in Query / Remark: click on the Q link to goarticle Please insert your reply or correction at the corresponding line in the proof

Q1 Please confirm that given names and surnames have been identified correctly.Q2 The country name has been inserted for the affiliation. Please check, and correct if necessary.Q3 Please check the fax number of the corresponding author, and correct if necessary.

Please check this box or indicate your approval ifyou have no corrections to make to the PDF file

Thank you for your assistance.

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

Pathology – Research and Practice xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Pathology – Research and Practice

journa l homepage: www.e lsev ier .com/ locate /prp

Original Article1

Communicating diagnostic uncertainty in surgical pathology reports:Disparities between sender and receiver

2

3

Sarah W. Lindley, Elizabeth M. Gillies, Lewis A. Hassell ∗Q14

Department of Pathology, The University of Oklahoma Health Sciences Center, BMSB 451, 940 Stanton L. Young Blvd, Oklahoma City, OK 73104, UnitedQ2States

5

6

7

a r t i c l e i n f o8

9

Article history:10

Received 23 October 201311

Received in revised form 13 February 201412

Accepted 15 April 201413

14

Keywords:15

Surgical pathology16

Communication17

Diagnostic Uncertainty18

a b s t r a c t

Surgical pathologists use a variety of phrases to communicate varying degrees of diagnostic certaintywhich have the potential to be interpreted differently than intended. This study sought to: (1) assessthe setting, varieties and frequency of use of phrases of diagnostic uncertainty in the diagnostic line ofsurgical pathology reports, (2) evaluate use of uncertainty expressions by experience and gender, (3)determine how these phrases are interpreted by clinicians and pathologists, and (4) assess solutions tothis communication problem. We evaluated 1500 surgical pathology reports to determine frequency ofuse of uncertainty terms, identified those most commonly used, and looked for variations in usage rateson the basis of case type, experience and gender. We surveyed 76 physicians at tumor boards who wereasked to assign a percentage of certainty to diagnoses containing expressions of uncertainty. We foundexpressions of uncertainty in 35% of diagnostic reports, with no statistically significant difference in usagebased on age or gender. We found wide variation in the percentage of certainty clinicians assigned tothe phrases studied. We conclude that non-standardized language used in the communication of diag-nostic uncertainty is a significant source of miscommunication, both amongst pathologists and betweenpathologists and clinicians.

© 2014 Published by Elsevier GmbH.

19

Background20

Communicating diagnostic uncertainty is an inherent part of all21

aspects of medicine. Pathology is presumed to be the final line in22

diagnosis, so when the pathologist expresses uncertainty in their23

diagnosis it could potentially lead to delayed treatment, repeat24

biopsy, and other interventions which increase medical expendi-25

tures and may negatively impact patient care.26

It is common practice in the pathology community to use27

phrases of uncertainty in the diagnostic line, most commonly when28

dealing with biopsy specimens. This may understandably be due to29

inadequate tissue, or extensive artifact that makes definite inter-30

pretation impossible. Other cited reasons for uncertainty include31

nonstandard histomorphology, ambiguous immunohistochemical32

stains, lack of clinical information, uncertain criteria in the liter-33

ature, lack of experience with the diagnosis, and hope (however34

unsubstantiated) to avoid legal liability for misdiagnosis.35

As pathologists we take pride in our linguistic acumen. When36

it comes to expression of uncertainty, pathologists are both very37

∗ Corresponding author. Tel.: +1 405 271 5653; fax: +1 4052712524.Q3E-mail addresses: Lewis-Hassel@ouhsc.edu, sachmacmon@gmail.com

(L.A. Hassell).

particular and very inventive in the phrases that they use. A 2004 38

survey of sign-out practices of 96 veterinary pathologists found 39

they were using at least 68 unique terms to describe uncertainty [1]. 40

No comparable study has been published in the human pathology 41

literature. 42

Unsurprisingly, clinicians and others in the health professions 43

interpret and act upon these phrases in different ways based on 44

their understanding (or misunderstanding) of the intent of the 45

pathologist. To the pathologist “consistent with” and “worrisome 46

for” may be intended to mean different things and direct different 47

courses of action, perhaps expressing a graded continuum of diag- 48

nostic certainty corresponding to an internal scale on the behalf of 49

the observer; however if this difference is not being clearly per- 50

ceived by the clinicians, then we are doing a disservice, both to 51

ourselves and to our patients. This study sought to clarify and quan- 52

tify this potential gap between intent and perception and diagnostic 53

language, and to begin to seek means to narrow this chasm. 54

Methods 55

We determined the incidence of usage of phrases of diagnostic 56

uncertainty in our institution by reviewing 1500 sequential surgi- 57

cal pathology reports and tallying both the occurrence of phrases of 58

http://dx.doi.org/10.1016/j.prp.2014.04.0060344-0338/© 2014 Published by Elsevier GmbH.

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

2 S.W. Lindley et al. / Pathology – Research and Practice xxx (2014) xxx–xxx

uncertainty in the diagnostic line and the frequency of use of each59

term. These sequential reports were completed between August60

and October of 2011 (1000 reports) and April and May of 200961

(500 reports.) For the latter series of 500 cases, specifics of case62

type (biopsy, resection, etc.) category of question (neoplastic, med-63

ical) as well as additional determination as to gravity of issue64

was determined. Cases where use of the uncertainty phrase cen-65

tered around a peripheral or subclassification rather than the core66

(malignant/not-malignant) were also noted and quantitated.67

In order to investigate the trends of usage of uncertainty terms68

by practitioner, a separate series of 200 sequential reported cases69

for each of the 14 actively practicing surgical pathologists at our70

institution were evaluated. The incidence of use of uncertainty71

terms for each pathologist was calculated. This data was then used72

to assess the effect of age and gender.73

To assess how various phrases were interpreted by clini-74

cians, we administered an anonymous survey of attendees at75

multi-disciplinary tumor boards. The survey asked respondents to76

estimate the degree of certainty associated with eight diagnos-77

tic scenarios (Fig. 1). One diagnosis contained no expression of78

uncertainty while the other seven contained the following phrases:79

“cannot rule out”, “consistent with”, “highly suspicious”, “favor”,80

“indefinite for”, “suggestive of”, and “worrisome for”. The order of81

presentation of each phrase and the specific content or specimen82

type associated therewith was randomized between surveys. The83

clinical context of the diagnoses was also customized according to84

the specialty of the tumor board where the evaluation was per-85

formed. A total of 76 responses were received. Statistical analysis86

was by Student’s t-test and ANOVA. Subgroup analyses were per-87

formed based on level of training and clinical specialty (medical88

students, residents, fellows, attendings, medicine/medical subspe-89

cialists, pathologists/radiologists, and surgeons).90

Looking for viable solutions to reporting of uncertainty, we con-91

ducted a focus group by sending a more detailed survey to seven92

senior physicians in various departments (surgery, oncology, radi-93

ation oncology, gynecologic oncology, and otolaryngology). In this94

survey, respondents rank ordered eight phrases from least to most95

certain. We also asked respondents what their opinion for mov-96

ing forward to resolve this communication problem would be and97

assessed their response to examples of certain proposed solutions.98

Finally, to gain further input into possible solutions to this prob-99

lem, we held an open discussion with attendees at a short course100

at a national pathology meeting in fall 2012.101

Results102

Of 1500 surgical pathology reports, we found expressions of103

uncertainty in 529 (35%). The most commonly used phrase at our104

institution was “consistent with” (50%), while the other oft-used105

phrases included “suggestive of”, “worrisome for”, “cannot rule106

out”, “highly suspicious for”, “favor”, and “indefinite for” (Fig. 2). We107

found no statistically significant difference in incidence of uncer-108

tainty phrase usage by either age or gender (Fig. 3).109

Uncertainty phrases were used more often in biopsy cases (96110

of 149 incidences, 64%) than in resection cases. Most often these111

involved a question of neoplasic or pre-neoplastic (83 of 149, 56%)112

rather than medical (66 of 149, 44%) disease. About one-fifth (22%)113

of incident usage dealt with a “trivial” matter (e.g. “consistent114

with lipoma”, “favor ganglion cyst”) and a similar number (29 of115

149, 19%) dealt with a sub-classification issue (e.g. “serrated polyp,116

favor serrated adenoma” or “spindle cell sarcoma, consistent with117

undifferentiated pleomorphic sarcoma.”) Interestingly, only a small118

number of reports containing uncertainty phrases included a com-119

ment or clarifying note (20 of 149, 13%) to either explain the cause120

of the uncertainty or further direct management, and few if any121

Table 1Standard deviation of percent of perceived certainty, as a measure of the degreeof consensus regarding the level of certainty, for common uncertainty phrases insurgical pathology reports. Higher numbers indicate wider variability in the level ofunderstood certainty.

(a) Deviation by specialty

Medicine Pathologist/radiologist Surgeons

No phrase 30 6 13Consistent with 16 25 13Highly suspicious 26 19 27Worrisome for 22 23 22Favor 24 25 23Suggestive of 26 23 29Cannot rule out 31 21 30Indefinite for 21 24 31

(b) Deviation by level of training

Medical students Residents Fellows Attendings

No phrase 8.3 8.1 30 15Consistent with 16 21 8.9 24Highly suspicious 27 23 27 18Worrisome for 22 24 19 23Favor 10 23 23 24Suggestive of 22 26 23 25Cannot rule out 18 25 25 27Indefinite for 29 19 25 28

of these offered specific suggestions beyond “clinical correlation.” 122

Surprisingly, none of the 149 incident cases in our review of 500 123

sequential cases appeared to be due to ambiguous or inconclusive 124

special stains. 125

Clinical respondents demonstrated wide differences in the 126

assigned level of certainty perceived to be associated with hedge 127

words in the diagnosis, with overall certainty scores of 91% for no 128

waffle phrase, 79% for “consistent with”, 71% for “highly suspicious 129

for”, 61% for “worrisome for”, 73% for “favor”, 50% for “indefinite 130

for”, 62% for “suggestive of”, and 48% for “cannot rule out”. The 131

variations within the level of perceived certainty (representing a 132

measure of the clarity of the phrase) are quantified by the standard 133

deviations from the means (Table 1). The average percent certainty 134

of the various groups were compared, both by level of training 135

(Fig. 4) and by specialty (Fig. 5). ANOVA analysis of the certainty 136

per phrase yielded statistically significant differences between all 137

phrases except “indefinite for”, “suggestive of”, and “worrisome 138

for”. When these phrases were compared to each other, the means 139

were not statistically different (p = 0.05). 140

In our focused study of seven senior clinicians, we found marked 141

variability in the way that the clinicians ranked the certainty asso- 142

ciated with various phrases. We also found varied opinions as to 143

how we should resolve this communication problem from the dif- 144

ferent clinicians surveyed. Many of the free text comments we 145

received were illuminating, reflecting their own preferred manner 146

for resolving such issues. For example, one surgeon emphasized the 147

need to review the slide directly with the pathologist, or at a min- 148

imum have a direct phone conversation. Another emphasized that 149

the issue was not so much grading the degree of uncertainty as it 150

was determining the threshold to treat or pursue further diagnos- 151

tic evidence. Our initial survey also sought to assess which phrases 152

could be linked to various levels of action, but the data is not pre- 153

sented here. From the majority of comments in the focused survey, 154

only an unqualified diagnosis or the phrase “consistent with” were 155

deemed actionable for definitive therapy. 156

Discussion 157

In our review of surgical case reports we were surprised by 158

the 35% incidence of expression of diagnostic uncertainty. Some 159

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

S.W. Lindley et al. / Pathology – Research and Practice xxx (2014) xxx–xxx 3

Fig. 1. Example survey question.

of this represents common institutional or individual phraseol-160

ogy; e.g., “consistent with lipoma” and “focal changes suggestive161

of HPV cytopathic effect” and may not truly represent significant162

diagnostic uncertainty but are reflected in the overall incidence163

nonetheless. However, these kinds of trivial uses only accounted164

for 22% of cases. Also of note is that all of the pathologists in our165

institution used some phrases of uncertainty in the relatively small166

number of cases studied. This highlights the breadth of the prob-167

lem of expressing uncertainty in surgical pathology. The variation in168

usage between pathologists (ranging from less than 5% to over 35%)169

is also an issue of concern with regard to both quality and consis-170

tency of communication and may warrant monitoring. Expressing171

a level of uncertainty out of habit or extreme caution when none172

is truly present dilutes the value of the phrase when perspicuity173

is warranted or essential. Not surprisingly, biopsies accounted for174

nearly two-thirds of the instances of use, and the majority of these175

were questions of malignancy or dysplasia, areas well known to be176

prone to interpretive variability. Medical disorders however, also 177

accounted for a significant number of cases (44%) which seems to 178

be reflective of imperfect or overlapping histopathologic criteria for 179

entities such as chemical gastropathy, inflammatory bowel disease, 180

or the many inflammatory dermatoses. 181

We considered a number of potential reasons commonly 182

asserted to be associated with a hedged diagnosis. Analysis of 183

reporting pathologists’ usage of uncertainty phrases by both age 184

and gender revealed no statistically significant differences. This 185

refutes the notion that expression of uncertainty is correlated with 186

lack of experience or even more archaically, with the gender of 187

the pathologist. Our data does not support either of these ideas. 188

Other possible rationales for expressions of uncertainty in diag- 189

nostic lines may include contradictory or low probability staining 190

results, lack of or inconsistent clinical information, uncertain crite- 191

ria in the medical literature, quantity of sample or abnormality, and 192

possibly a desire to avoid legal liability for an over- or under-stated 193

62%38%

Frequency of Expressions of

Uncertainty

No

caveats

Waf�led50%

39%

Phrases Used Favor

Consistent

With

Suggestive of

Suspcious of

Cannot rule

out

Inde�inite for

Worrisome for

Fig. 2. Expression of uncertainty use at our institution.

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

4 S.W. Lindley et al. / Pathology – Research and Practice xxx (2014) xxx–xxx

52

41

13

35

10 9

17

26

10

29

37

68

20

33

0

10

20

30

40

50

60

70

80

60+60+60+60+30-4030-4030-4040-5040-5040-5040-5050-6050-6050-60

Num

ber o

f Waffl

e W

ords

Use

d pe

r 200

Sig

n-O

uts

Age

Fig. 3. Waffle word usage by age and gender.

Fig. 4. Average percent certainty by level of training.

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

S.W. Lindley et al. / Pathology – Research and Practice xxx (2014) xxx–xxx 5

Fig. 5. Average percent certainty by specialty.

diagnosis. These latter motives were not fully investigated in our194

study but may bear further scrutiny.195

While acknowledging that our method of sampling (written196

survey given at tumor boards) has limitations, including potential197

sample bias and response bias; we feel that this method was the198

most time and cost effective way to get a cross sectional study of199

clinicians at all levels of training and in a wide variety of specialties.200

Our questionnaire design incorporated elements of customization201

and presentation randomization to limit these biases.202

Overall, we found that the phrases “consistent with”, “highly203

suspicious”, and “favor” are perceived to be associated with more204

certainty in the diagnosis. The latter term is a surprise to be included205

in this group since it is regularly interpreted by pathologists as206

less certain than the other two and quite similar to “suggestive207

of”. But the surgical group ranked it more certain than “highly sus-208

picious” by almost 10 percentage points. The phrases “suggestive209

of”, “worrisome for”, and “indefinite for” were all less certain. The210

finding of no statistically significant difference in comparing the211

phrases “indefinite for”, “suggestive of”, and “worrisome for” across212

all groups of respondents tells us that these phrases are equiva-213

lent in terms of communicating uncertainty. So to the pathologist214

obsessing over a subtle internal rank order of phrases with which215

to exactly convey what they are seeing, for approximately 50–60%216

certainty in diagnosis, should probably relax and use one or any as217

our data shows them to communicate an equivalent message. This218

may be driven by the equivalent nature of the clinical response each219

phrase is likely to produce.220

To move toward at least a local solution to this problem, we221

conducted the focused survey of our senior clinicians. All but one222

of our respondents felt that only “carcinoma” and “consistent with223

carcinoma” were sufficient to treat. One respondent felt that even 224

“worrisome for carcinoma” was enough to treat given the right 225

clinical circumstance. 226

We posed some potential solutions to the focus group clinicians 227

at our institution and to a group of approximately 30 practic- 228

ing pathologists at a national forum on the topic. One option is 229

to develop a national consensus categorization with data-driven 230

guidance, similar to the Bethesda systems in cytology [2]. Less 231

ambitiously, we could develop a local departmental or institu- 232

tional consensus on usage communicated monolithically to users, 233

more gestalt-driven, perhaps based on cytology model with a tiered 234

system. So for example, a diagnosis of a malignancy without any 235

qualifiers would lead to definitive action; “suspicious for” or “con- 236

sistent with” would lead to definitive action if clinical story agrees; 237

and “atypical”, “favor”, “cannot rule out”, “suggestive of” would 238

be accepted to merit additional evaluation or follow-up. Alter- 239

nately, we propose an outcomes data driven solution based on 240

analysis of reports with various phrases from which a quantitative 241

qualifier could be appended (e.g., diagnoses containing the phrase 242

“suggestive of” are associated with an 80% probability of a posi- 243

tive diagnosis). An individually assigned, subjective quantization 244

of the intended degree of certainty (gestalt-based only) included 245

as a note or other element of the report itself might also close the 246

gap between sender and receiver, but would be subject to variable 247

usage and experience. The last and least rigorous option is to make 248

no reporting or usage change, but just build awareness amongst 249

pathologists and clinicians that use of these phrases leads to misun- 250

derstandings, and so might best trigger a phone call to the clinician 251

by the pathologist or vice versa to discuss the case and subsequent 252

actions. 253

Please cite this article in press as: S.W. Lindley, et al., Communicating diagnostic uncertainty in surgical pathology reports: Disparitiesbetween sender and receiver, Pathol. – Res. Pract (2014), http://dx.doi.org/10.1016/j.prp.2014.04.006

ARTICLE IN PRESSG ModelPRP 51195 1–6

6 S.W. Lindley et al. / Pathology – Research and Practice xxx (2014) xxx–xxx

Our focus group found elements of each of these proposed solu-254

tions attractive and useful, though they recognized the magnitude255

of the challenge in arriving at a data-driven solution given the num-256

ber and variety of causes for the problem, tissue sample types,257

locations and professional stakeholders potentially impacted. In258

presenting these various possible solutions to our forum on the259

topic at a national meeting, we again found no clear consensus on260

the best approach. Most importantly, this discussion highlighted261

the prevalence of this particular communication problem across262

all pathology practices.263

Some aspects of this issue have been previously studied in the264

British literature. Attanoos et al., studied phraseology in surgical265

reports and communication of uncertainty between surgeons and266

pathologists at the University Hospital Wales [3]. Galloway and267

Taiyeb examined the interpretation of phrases used to describe268

uncertainty amongst pathologists, other doctors, and medical stu-269

dents online and at the University College London Medical School270

[4]. In both of these studies, akin to our findings, there was wide271

variance in the interpretation of phrases between the groups stud-272

ied. They similarly concluded adoption of a limited number of273

descriptive phrases that are mutually understood and accepted by274

both pathologists and clinicians is needed to avoid ambiguity in275

surgical pathology reports. An additional study addressed the need276

for uniformity in reporting cancer for the British National Cancer277

Registry [5]. In his 2000 commentary on individuality in surgical278

pathology, Dr. Foucar aptly concluded, “. . .There is no place for the279

pathologist who expresses individuality by subjecting unsuspect-280

ing patients to uncontrolled diagnostic self-expression” [6].281

Although a clear consensus solution, either at our institution or282

among our colleagues elsewhere remains elusive; we have reached283

several important conclusions. Like the British studies, communica-284

tion of uncertainty indeed is a common practice and an unexamined285

source of possible medical error in the United States. We plan286

to study this possible relationship more fully. Our own anecdotal287

experience in tumor boards and an array of practice settings have288

provided several “near miss” examples, and more than a few need-289

less repeat biopsies or other procedures due to cautiously worded

reports with these phrases. Further study is needed to further refine 290

the specimens and diagnostic settings in which diagnostic uncer- 291

tainty is most commonly expressed in order to encourage improved 292

diagnostic criteria and provide better follow-up guidance when 293

such are not fully present and an uncertainty phrase mandated. 294

Additionally, it would be helpful to be able to calculate the possi- 295

ble cost to the health care system due to repeat biopsies in specific 296

cases. Secondly, action needs to be taken to address the issue of the 297

gap between uncertainty intention and perception at least locally 298

and preferably at a national level. An interesting trend appears to 299

be emerging from both our discussion at our institution and those 300

at the national meeting: more recently trained pathologists more 301

fully support national guidelines on terminology while more senior 302

pathologists tend to resist this loss of individuality in reporting. In 303

this and so many other aspects, it will be fascinating to see where 304

the new generations of pathologists take our field. Inasmuch as 305

interpersonal communication is a core competency for physicians 306

and training programs, the issue herein raised centers around a 307

critical practice skill for pathologists, and one where the data might 308

indicate we are not yet fully competent. In regards to clarity of com- 309

munication, our data suggests that moving toward uniformity both 310

in reporting style and language is the right direction. 311

References 312

[1] M.M. Christoper, C.S. Hotz, Cytologic diagnosis: expression of probability by 313

clinical pathologists, Vet. Clin. Pathol. 33 (2) (2004) 84–95. 314

[2] E. Cibas, S. Ali, The Bethesda system for reporting thyroid cytopathology, Am. J. 315

Clin. Pathol. 132 (2009) 658–665. 316

[3] R.L. Attanoos, A.D. Bull, A.G. Douglas-Jones, et al., Phraseology in pathology 317

reports: a comparative study of interpretation among pathologists and surgeons, 318

J. Clin. Pathol. 49 (1) (1996) 79–81. 319

[4] M. Galway, T. Taiyeb, The interpretation of phrases used to describe uncertainty 320

in pathology reports, Pathol. Res. Int. 2011 (2011), Article 656079. 321

[5] P. Silcocks, M. Page, What constitutes a histological confirmation of cancer? A 322

survey of terminology interpretation in two English regions, J. Clin. Pathol. 54 323

(March (3)) (2001) 246–248. 324

[6] E. Foucar, “Individuality” in the specialty of surgical pathology: self-expression 325

or just another source of diagnostic error? Am. J. Surg. Pathol. 24 (11) (2000) 326

1573–1576. 327