On 2014 Jul 12, David Keller commented:

The USPSTF statement confuses sensitivity and specificity

The following error appears in the USPSTF Recommendation statement posted on their own website (1) as well as in the identical document published online by Annals of Internal Medicine (2):

"Although screening with ultrasonography has few direct harms, all screening strategies, including those with or without confirmatory tests (that is, digital subtraction or magnetic resonance angiography), have imperfect sensitivity and could lead to unnecessary surgery and result in serious harms, including death, stroke, and myocardial infarction."

Correction: The author meant to write that these screening tests "have imperfect specificity, and could lead to unnecessary surgery..."

Rationale: Imperfect specificity is defined as a rate of false-positive test results greater than zero, which can lead to a diagnosis of disease which does not exist, and hence to unnecessary surgery. In contrast, imperfect sensitivity is defined as a rate of true-positive test results less than 100%, which does not lead to unnecessary surgery - but it can lead to the opposite kind of harm, which is a missed diagnosis, which can lead to the failure to perform necessary surgery.

The correct use of these terms, and the distinction between their definitions, is important in any discussion of public health measures.

The erroneous word was bolded by this commentator.

9/22/2014: As of today, the above error remains uncorrected on both the USPSTF and Annals websites. I have notified the editor of Annals by email. I encourage readers to submit their comments regarding this error to USPSTF, Annals, and PubMed Commons.

11/9/2014: As of today, the above error remains uncorrected on the Annals of Internal Medicine website. However, it has been corrected on the USPSTF website, at the following new web address:

http://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/carotid-artery-stenosis-screening

I will again attempt to have it corrected by the editors of Annals.

12/26/2014: I exchange emails with the editor-in-chief of Annals, who assures me that the matter will be looked into.

1/21/2015: A recheck of the Annals website discloses that the error has been corrected, and now reads "Although screening with ultrasonography has few direct harms, all screening strategies, including those with or without confirmatory tests (that is, digital subtraction or magnetic resonance angiography), have imperfect sensitivity and specificity and could lead to unnecessary surgery and result in serious harms, including death, stroke and MI".

To date, I have received no acknowledgement or thanks for pointing out the above error, from either the USPSTF or from Annals. This is the typical response I have received when correcting errors, including other, more serious errors in JAMA and the NEJM, which were eventually corrected after I expended almost preposterous efforts. Errors in the scientific literature can propagate unless corrected, leading to further errors. The NIH should establish an impartial board of experts, to whom outstanding errors in PubMed-indexed publications can be appealed, in cases where readers have spotted significant errors and the responsible editors refuse to correct them.

References

1: U.S. Preventative Services Task Force website, accessed on 7/12/2014, search for the second instance of "imperfect sensitivity" http://www.uspreventiveservicestaskforce.org/uspstf13/cas/casfinalrs.htm

2: Annals of Internal Medicine website, accessed on 7/12/2014, search for the second instance of "imperfect sensitivity" http://annals.org/article.aspx?articleid=1886690

On 2014 Jul 12, David Keller commented:

This USPSTF statement contains two different recommendations, which must be discussed separately

1) The USPSTF recommends against screening for asymptomatic carotid artery stenosis in the general adult population. Grade D

2) The USPSTF recommends against invasive mechanical treatment (ie: surgery or angioplasty) for asymptomatic carotid artery stenosis in the general adult population. Grade D.

The USPSTF statement convinced me that their grade of "D" is valid for recommendation #2 above; invasive treatments of asymptomatic carotid stenosis probably do more harm than good.

However, the USPSTF is clearly wrong on recommendation #1 above. Their statement admits that there is insufficient data to determine whether high-dose statin therapy will prove to be as beneficial for carotid atherosclerotic disease as it has proven for coronary atherosclerotic disease. Therefore, by their own grading standards, recommendation #1 should be classified as Grade "I" for indeterminate.

Further, many experts maintain that atherosclerotic stenoses, whether discovered in a carotid artery or a coronary artery, are manifestations of the same underlying disease entity, systemic atherosclerosis, which if present in a significant degree, should be treated with high-dose statin therapy. For those who hold this view, screening asymptomatic patients for carotid stenosis should be given a grade of "C", meaning each individual should discuss with their physician whether they should get screened, based on their risk profile.

I screen asymptomatic patients who have a reasonable risk of atherosclerosis, to identify systemic atherosclerosis in a convenient artery which is accessible to ultrasonic evaluation. If carotid stenosis is present, I treat is as a coronary artery disease risk equivalent, and initiate treatment with high-dose statin therapy. This strategy is reasonable, based on extrapolation of the findings of benefit in every cardiac statin trial.

The arguments supporting carotid screening for the purpose of risk-stratification are biologically plausible and lead to active changes in therapy for patients without known coronary artery disease, who are found to have asymptomatic carotid stenosis.

The USPSTF should rewrite this statement to narrowly focus on recommending against invasive treatment of carotid stenosis, and encourage physicians to continue screening appropriate patients for asymptomatic carotid atherosclerosis as a way to identify additional patients who can benefit from the use of high dose statin therapy.

On 2014 Jul 12, David Keller commented:

This USPSTF statement contains two different recommendations, which must be discussed separately

1) The USPSTF recommends against screening for asymptomatic carotid artery stenosis in the general adult population. Grade D

2) The USPSTF recommends against invasive mechanical treatment (ie: surgery or angioplasty) for asymptomatic carotid artery stenosis in the general adult population. Grade D.

The USPSTF statement convinced me that their grade of "D" is valid for recommendation #2 above; invasive treatments of asymptomatic carotid stenosis probably do more harm than good.

However, the USPSTF is clearly wrong on recommendation #1 above. Their statement admits that there is insufficient data to determine whether high-dose statin therapy will prove to be as beneficial for carotid atherosclerotic disease as it has proven for coronary atherosclerotic disease. Therefore, by their own grading standards, recommendation #1 should be classified as Grade "I" for indeterminate.

Further, many experts maintain that atherosclerotic stenoses, whether discovered in a carotid artery or a coronary artery, are manifestations of the same underlying disease entity, systemic atherosclerosis, which if present in a significant degree, should be treated with high-dose statin therapy. For those who hold this view, screening asymptomatic patients for carotid stenosis should be given a grade of "C", meaning each individual should discuss with their physician whether they should get screened, based on their risk profile.

I screen asymptomatic patients who have a reasonable risk of atherosclerosis, to identify systemic atherosclerosis in a convenient artery which is accessible to ultrasonic evaluation. If carotid stenosis is present, I treat is as a coronary artery disease risk equivalent, and initiate treatment with high-dose statin therapy. This strategy is reasonable, based on extrapolation of the findings of benefit in every cardiac statin trial.

The arguments supporting carotid screening for the purpose of risk-stratification are biologically plausible and lead to active changes in therapy for patients without known coronary artery disease, who are found to have asymptomatic carotid stenosis.

The USPSTF should rewrite this statement to narrowly focus on recommending against invasive treatment of carotid stenosis, and encourage physicians to continue screening appropriate patients for asymptomatic carotid atherosclerosis as a way to identify additional patients who can benefit from the use of high dose statin therapy.

On 2014 Jul 12, David Keller commented:

The USPSTF statement confuses sensitivity and specificity

The following error appears in the USPSTF Recommendation statement posted on their own website (1) as well as in the identical document published online by Annals of Internal Medicine (2):

"Although screening with ultrasonography has few direct harms, all screening strategies, including those with or without confirmatory tests (that is, digital subtraction or magnetic resonance angiography), have imperfect sensitivity and could lead to unnecessary surgery and result in serious harms, including death, stroke, and myocardial infarction."

Correction: The author meant to write that these screening tests "have imperfect specificity, and could lead to unnecessary surgery..."

Rationale: Imperfect specificity is defined as a rate of false-positive test results greater than zero, which can lead to a diagnosis of disease which does not exist, and hence to unnecessary surgery. In contrast, imperfect sensitivity is defined as a rate of true-positive test results less than 100%, which does not lead to unnecessary surgery - but it can lead to the opposite kind of harm, which is a missed diagnosis, which can lead to the failure to perform necessary surgery.

The correct use of these terms, and the distinction between their definitions, is important in any discussion of public health measures.

The erroneous word was bolded by this commentator.

9/22/2014: As of today, the above error remains uncorrected on both the USPSTF and Annals websites. I have notified the editor of Annals by email. I encourage readers to submit their comments regarding this error to USPSTF, Annals, and PubMed Commons.

11/9/2014: As of today, the above error remains uncorrected on the Annals of Internal Medicine website. However, it has been corrected on the USPSTF website, at the following new web address:

http://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/carotid-artery-stenosis-screening

I will again attempt to have it corrected by the editors of Annals.

12/26/2014: I exchange emails with the editor-in-chief of Annals, who assures me that the matter will be looked into.

1/21/2015: A recheck of the Annals website discloses that the error has been corrected, and now reads "Although screening with ultrasonography has few direct harms, all screening strategies, including those with or without confirmatory tests (that is, digital subtraction or magnetic resonance angiography), have imperfect sensitivity and specificity and could lead to unnecessary surgery and result in serious harms, including death, stroke and MI".

To date, I have received no acknowledgement or thanks for pointing out the above error, from either the USPSTF or from Annals. This is the typical response I have received when correcting errors, including other, more serious errors in JAMA and the NEJM, which were eventually corrected after I expended almost preposterous efforts. Errors in the scientific literature can propagate unless corrected, leading to further errors. The NIH should establish an impartial board of experts, to whom outstanding errors in PubMed-indexed publications can be appealed, in cases where readers have spotted significant errors and the responsible editors refuse to correct them.

References

1: U.S. Preventative Services Task Force website, accessed on 7/12/2014, search for the second instance of "imperfect sensitivity" http://www.uspreventiveservicestaskforce.org/uspstf13/cas/casfinalrs.htm

2: Annals of Internal Medicine website, accessed on 7/12/2014, search for the second instance of "imperfect sensitivity" http://annals.org/article.aspx?articleid=1886690