(ADINSERTER AMP)
Monday, September 26, 2022
HomePediatrics DentistryStability of diagnostic rate in a cohort of 38,813 colorectal polyp specimens...

Stability of diagnostic rate in a cohort of 38,813 colorectal polyp specimens and implications for histomorphology and statistical process control


Appendix A (Diagnostic Codes and Search Strings)

Dx code String
dx1 Suspicious for adenocarcinoma
dx1 Cannot exclude adenocarcinoma
dx1 Adenocarcinoma can’t excluded
dx1 Suspicious for invasive adenocarcinoma
dx2 Invasive adenocarcinoma
dx2 Adenocarcinoma
dx2 Invasive carcinoma
dx3 Intramucosal adenocarcinoma
dx3 Of high-grade dysplasia
dx3 Of high grade dysplasia
dx3 With high-grade dysplasia
dx3 With high grade dysplasia
dx3 Focal high grade dysplasia
dx3 Focal high-grade dysplasia
dx3 Showing high grade dysplasia
dx3 Showing high-grade dysplasia
dx4 Suspicious for lymphoma
dx4 Atypical lymphoid proliferation
dx5 MALT lymphoma
dx5 Mantle cell lymphoma
dx6 Tubular adenoma
dx6 Tubular adenomata
dx7 Hyperplastic polyp
dx7 Hyperplastic polyp
dx8 Hyperplastic changes
dx8 Hyperplastic mucosal changes
dx8 Hyperplastic features
dx8 Hyperplastic-like features
dx9 Tubulovillous adenoma
dx9 Tubularvillous adenoma
dx9 Tubulo-villous adenoma
dx9 Villotubular adenoma
dx10 Villous adenoma
dx11 Sessile serrated adenoma
dx11 Sessile serrated polyp
dx11 Serrated sessile adenoma
dx11 Serrated polyp, favor adenoma
dx11 Serrated polyp, favour adenoma
dx11 Serrated polyps, favor adenomata
dx12 Traditional serrated adenoma
dx13 Serrated adenoma
dx13 Serrated adenomata
dx13 Serrated polyp
dx14 No pathologic finding
dx14 Unremarkable fragment of large bowel mucosa
dx14 Unremarkable large bowel mucosa
dx14 Benign colonic mucosa
dx14 Benign large bowel mucosa
dx14 Normal colonic mucosa
dx14 NO DIAGNOSTIC ABNORMALITY
dx14 Unremarkable colonic mucosa
dx14 Mucosa without significant pathology
dx14 Mucosa within normal limits
dx14 No specific pathology
dx14 No significant histopathologic abnormality
dx14 NEGATIVE for evidence of significant pathology
dx14 No significant pathological changes
dx14 No pathological changes
dx14 No evidence of polyp
dx14 Polypoid mucosa
dx14 Colonic mucosa with prominent lymphoid follicles
dx14 Negative for apparent pathology
dx14 No pathological diagnosis
dx14 CAUTERY ARTIFACT, NOT FURTHER DIAGNOSTIC
dx14 Reactive changes, NEGATIVE
dx14 Colonic mucosa with lymphoid aggregate
dx14 Large bowel mucosa with no definite polyp
dx14 Non-diagnostic polypoid colonic mucosa
dx14 Bowel mucosa with no significant findings
dx14 Bowel mucosa with no evidence of polyp
dx14 No polyp
dx14 No specific polyp
dx14 No definitive polyp
dx14 No definite polyp
dx14 No specific pathologic diagnosis
dx14 No significant pathology identified
dx14 No significant pathological abnormalities
dx14 No findings
dx14 No histopathological abnormality
dx14 Prominent mucosal folds
dx14 Mucosa, likely mucosal fold
dx14 Without diagnostic abnormality
dx14 Unremarkable mucosal tissue
dx14 No significant findings
dx14 Colonic mucosa with no pathology
dx14 No significant inflammation or other findings
dx15 POLYPOID COLONIC MUCOSA
dx15 Prominent lymphoid aggregate
dx15 Lymphoid aggregate
dx15 Large intestinal mucosa slightly polypoid with lymphoid aggregates
dx15 Mucosa with lympho-follicular hyperplasia
dx15 Lymphoid follicle
dx15 Benign lymphoid aggregate
dx15 Mucosal germinal centre
dx15 Lymphoid hyperplasia
dx15 Lymphocytic aggregates
dx16 Inflammatory polyp
dx16 Inflammatory pseudopolyp
dx16 Inflammatory large bowel polyp
dx16 Inflammatory-type polyp
dx16 Inflamed polyp
dx17 Hamartomatous polyp
dx18 Granulation tissue
dx19 Active colitis
dx19 Active proctitis
dx19 Acute cryptitis
dx20 Poorly preserved colonic mucosa
dx21 Solitary rectal ulcer
dx21 Mucosal prolapse syndrome
dx21 Mucosal Prolapse
dx21 Mucosal prolapse-like polyp
dx21 Mucosa with prolapse like changes
dx22 Melanosis coli
dx22 Pseudomelanosis coli
dx22 Slight melanosis
dx23 Juvenile polyp
dx23 Juvenile type polyp
dx23 Retension polyp
dx24 Lipoma
dx25 Granular cell tumour
dx25 Granular cell tumor
dx26 Ischemic colitis
dx27 Leiomyoma
dx28 Xanthoma
dx28 Xanthoma/xanthelasma
dx29 Hemorrhoid
dx30 Prolapse changes
dx31 Fecal material
dx31 Fecal matter only
dx31 Vegetable fibres
dx31 Vegetable matter
dx31 Fecal matter
dx31 Feces
dx31 Food material
dx31 Polypoid vegetable resembling seed
dx31 Degenerated meat fibres
dx32 NEGATIVE for dysplasia
dx32 NEGATIVE for evidence of dysplasia
dx32 No neoplasia present
dx32 Negative for conventional/adenomatous dysplasia
dx32 Negative for adenomatous polyp or dysplasia
dx32 Negative for adenoma
dx33 Tissue not identified
dx33 No tissue is identified
dx33 No tissue present
dx33 No tissue was found
dx33 No microscopic assessment possible
dx33 Tissue did not survive processing
dx33 Did not survive tissue processing
dx33 No material present after processing
dx33 See gross
dx33 No specimen received
dx33 No specimen identified
dx33 Insufficient for evaluation
dx33 Insufficient for assessment
dx33 Insufficient tissue for histologic assessment
dx33 No colon tissue is observed
dx33 Mucosa, not diagnostic
dx34 Negative for high grade dysplasia
dx34 Negative for high-grade dysplasia
dx34 No evidence of high grade dysplasia
dx34 No evidence of high-grade dysplasia
dx34 No evidence of high dysplasia
dx34 No definite evidence of high-grade dysplasia
dx34 No convincing evidence of high grade dysplasia
dx34 Without high grade dysplasia
dx34 Without high-grade dysplasia
dx35 NEGATIVE FOR DYSPLASIA OR MALIGNANCY
dx35 Negative for high-grade dysplasia and malignancy
dx35 Negative for high grade dysplasia and malignancy
dx35 Negative for high-grade dysplasia or invasive malignancy
dx35 Negative for high grade dysplasia or invasive malignancy
dx35 Negative for high-grade dysplasia or invasive carcinoma
dx35 Negative for high grade dysplasia or invasive carcinoma
dx35 Negative for high-grade dysplasia and invasive carcinoma
dx35 Negative for high grade dysplasia and invasive carcinoma
dx35 No convincing evidence of high grade dysplasia or adenocarcinoma
dx36 Cautery/crush artifact
dx36 Cautery artifact
dx36 Cautery artefact
dx36 Cauterized tissue
dx36 Cauterized colonic mucosa
dx36 Polypoid cauterized mucosa
dx36 Crushed fragments of large bowel
dx37 Focal adenomatous changes
dx37 Focal adenomatous change
dx37 Fragments of adenoma
dx37 Possible adenomatous change
dx37 Adenoma
dx37 Suspicious for Adenomatous Changes
dx37 Adenomatous change
dx37 Adenomatous mucosal change
dx37 -ADENOMA(S)
dx38 Chronic inflammation
dx38 Chronic inflammation only
dx39 Dysplasia associated lesion or mass
dx39 Features of DALM
dx40 Carcinoid
dx40 Neuroendocrine tumour
dx40 Neuroendocrine tumor

Appendix B (Location Codes, Search Strings and Distances)

The location code was based on the location provided in the “source of specimen” section of the report. If the location was given in centimetres from the anal verge, it was converted to named location, based on approximate measures used by NCI.

Appendix B-1: Codes and location dictionary

Location code String
loc1 Rectal
loc1 Rectum
loc2 Rectosigmoid
loc2 Rectum—sigmoid
loc2 Rectal—sigmoid
loc3 Sigmoid
loc4 Descending
loc5 Splenic
loc6 Transverse
loc7 Hepatic
loc8 Ascending
loc9 Cecum
loc9 Cecal
loc9 Appendiceal orifice
loc10 Left colon
loc11 Right colon
loc12 Anastomosis
loc12 Anastomotic

Appendix B-2: Codes and location table

Location code Distance (cm)
loc1 < 13
loc2 < 18, ≥ 13
loc3 < 58, ≥ 18
loc4 < 79, ≥ 58
loc5 < 84, ≥ 79
loc6 < 130, ≥ 84
loc7 < 136, ≥ 130
loc8 < 147, ≥ 136
loc9 < 151, ≥ 147

Appendix C (Control charts/normalized funnel plots)

Based on the normal approximation of the binomial distribution:

$${\text{SE}} = \sqrt {\frac{{i \times (1 – i)}}{n}}$$

(1)

where:

SD = standard deviation.

i = ideal (diagnostic) rate †

n = number of specimens interpreted.

† The ideal rate in this study is approximated by the group median rate.

The healthcare provider rate (pathologist diagnostic rate) is normalized as follows:

$$N_{j} = \frac{{M_{j} – i}}{{{\text{SE}}_{j} }}$$

(2)

where:

Nj = healthcare provider rate for healthcare provider “j”.

Mj = measured rate for the healthcare provider “j”.

i = ideal (diagnostic) rate.

SDj = SD for healthcare provider “j”.

Equation (2) can be substituted into Eq. (1):

$$M_{j} – N_{j} \left[ {\sqrt[i]{{\frac{(1 – i)}{n}}}} \right] + {\text{i}}$$

(3)

To normalize we presume that the “SD” is equivalent and that only “n” changes. This amounts to forming two equations from Eq. 3 and solving for the normed Mj. After some rearrangement one can derive a conversion equation:

$$M_{{{\text{jnormed}}}} = \left[ {M_{{{\text{jmeasured}}}} – i} \right]\frac{{\sqrt[i]{{\frac{(1 – i)}{{n_{{{\text{jnormed}}}} }}}}}}{{\sqrt[i]{{\frac{(1 – i)}{{n_{{{\text{jmeasured}}}} }}}}}} + {\text{i}}$$

(4)

where

Mj normed = normed (diagnostic) rate for healthcare provider “j”.

Mj measured = measured (diagnostic) rate for healthcare provider “j”.

nj normed = normed number of specimens handled (interpreted) by healthcare provider “j”.

nj measured = number of specimens handled (interpreted) by healthcare provider “j”.

i = ideal (diagnostic) rate.

Appendix D (Probability Calculation)

The probability (Y) of n and < n outliers in k samples is dependent on the probability of the individual outlier (p), and the binomial cumulative distribution function:

$${\text{Y }} = {\text{ binocdf}}\left( {{\text{n}},{\text{ k}},{\text{ p}}} \right)$$

(5)

The probability (X) of n and > n outliers in this context is the complement of ‘n-1’:

$${\text{X }} = { 1} – {\text{ binocdf }}\left( {{\text{n}} – {1},{\text{ k}},{\text{ p}}} \right)$$

(6)

Appendix E: Understanding statistical process control

Overview

Statistical process control is a cyclical process that involves:

  1. (1)

    Repeated measurement

  2. (2)

    Assessment of the variation in the measurement (using statistics)

  3. (3)

    Possible adjustments in the process to ensure that future measurements fall within a prescribed range, as defined by the so-called “control lines”

Statistical process control applied to diagnostic pathology at a conceptual level

In the diagnostic pathology context—if all the following are true:

  1. (1)

    Cases are assigned randomly to pathologists from a given population

  2. (2)

    Pathologists see large numbers of a particular type of case (e.g. 200 cases)

Then:

It is likely that the diagnostic rate (of say ‘tubular adenoma’) is similar for different pathologists (e.g. Pathologist ‘A’ diagnoses 102 tubular adenoma in 200 cases, Pathologist ‘B’ diagnosies 105 tubular adenomas in 200 cases) ****

**** Statistically, it can be stated that there is range in which the diagnostic rate (number of diagnoses/total cases interpreted) will fall 95% of the time.

If pathologists have significantly different diagnostic rates:

It is likely that they interpret cases differently and it may be possible to reduce diagnostic variation, via diagnostic calibration.

Diagnostic calibration is not new

Pathologists can change their diagnostic rates over time and may do this when (1) their cases are reviewed and found discrepant, (2) they review cases of (trusted) other (more experienced or subspeciality trained) pathologists, and (3) when new diagnostic entities are discovered or diagnostic criteria revised.

If a pathologist consistently (over time) has a diagnostic rate for tubular adenoma (e.g. 20/200 = 0.2) that is outside the range expected (in relation to the diagnostic rates of all their colleagues (~ 100/200 = 0.5)), one can infer that it would be possible to “re-train” that individual (or all the other pathologists) to arrive at the same diagnostic rate.

SPC in a nutshell is: systematically looking at the data (with statistics) to calibrate a process; in anatomical pathology it would be a process to look at diagnostic rates and feed those diagnostic rates back to the pathologists—such that they can adjust to a target rate/find agreement on what the target rate should be.

Statistical process control is predicated on two conditions

Condition (1) the process that one wants to control is stable over time (such that it is possible to predict the future) or can be made stable.

Condition (2) there is an ability to adjust the control variable * in a meaningful way—in relation to the control lines **.

* a control variable is: a parameter that one wants to control, e.g. the diagnostic rate of ‘sessile serrated adenoma’.

** control lines in SPC are determined by the “expected” statistical variation—when the process is in control/running optimally, e.g. the control parameter falls within a given range 95% of the times. Control limits are confidence intervals and are directly analogous to the funnel lines on funnel plots.

“Ability to adjust the control variable in a meaningful way” implies the following:

  1. (1)

    the variation (between providers) one expects to see due to chance is smaller than the (actual) variation that is observed; this implies that improvement is possible ***

  2. (2)

    an intervention can change the control variable in a substantive way, such that the variation is reduced

*** If the variation is less than the variation by chance the process is in control (or one may need a larger sample size).

The conditions for statistical process control and the objective of the manuscript

‘Condition 1’ for SPC is met if there is diagnostic stability.

‘Condition 2’ for SPC is met if there is significant diagnostic variation—that is stable (e.g. one pathologist is a consistent outlier in relation to the median diagnostic rate *****), and it is assumed that pathologists want to improve their practice/can be encouraged to make positive changes (see section “Diagnostic Calibration is Not New”).

‘Condition 1’ and ‘Condition 2’ are sufficient to infer that SPC should be feasible and could be used to improve care.

***** It should be noted that: the ‘median diagnostic rate’ may not be the ideal diagnostic rate for a given population. It is possible that an ‘outlier’ pathologist represents the ideal diagnostic rate.

In SPC, one talks of variation due to an “assignable cause” [a modifiable factor] and “common cause” [un-modifiable factors]. In the language of SPC, the question succinctly is: Is the pathologist an assignable cause?

If diagnostic rates are stable [Condition 1], and the pathologist is an “assignable cause” [Condition 2], SPC should be feasible.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments