On 2014 May 29, David Keller commented:

Replace the USPSTF with several committees composed of experts in the disease being considered

In medicine, as in other sciences, open questions regarding Nature are settled by experimentation. If the data from an experiment yield unclear results, the outcome of the experiment should not be declared by a committee. The experiment should be refined and repeated until the data clearly indicate the answer to the question being examined.

If we have clear, unambiguous results from well-designed experiments, we do not need a committee like the USPSTF to interpret those results and declare what they mean. Conversely, if the data is so ambiguous, arguable or suspicious that a committee is required to interpret it, then that data is inadequate to base important recommendations on.

Equipoise is the state of not being certain which of two opposite courses of action is correct. Take multivitamins or not? Screen for carotid stenosis or not? Screen for prostate cancer using PSA or not? These are all very important questions with life-altering consequences for many people if we get it wrong. Equipoise exists as long as the data are so unclear that the correct answer is not obvious, and certainly as long as a committee is required to interpret the results.

The USPSTF members are part-time volunteers who are mainly occupied by other important jobs, such as running large healthcare organizations, academic departments and research enterprises. How can a pediatrician or a gynecologist on the USPSTF, for example, be expected to master, in a few hours per week, the art and science of an unfamiliar specialty such as urology or vascular surgery, which they do not practice and to which others devote their lives? They cannot, nobody is that smart or that efficient. This is why the pronouncements of the USPSTF often meet with disbelief and non-compliance among practicing physicians. When our accumulated experience informs us that certain screening tests or treatments are helpful to patients, we require convincing data to change our practices. And, given data which is convincing, alert clinicians do not need a committee to interpret it for us. The consensus of editorials and reviews in medical journals is a reliable source of guidance when it is required.

In the years following the widespread introduction of PSA screening, there was a drop of over 35% in deaths due to prostate cancer. While this is merely observational data, from which no conclusions can be drawn regarding cause-and-effect, it is still an impressive fact, which must be explained if we are asked to stop screening with PSA. One of the main clinical trials which the USPSTF relied on to arrive at their "D" recommendation for PSA screening was PLCO (for "Prostate, Lung, Colorectal and Ovarian" cancer). PLCO was hindered by the fact that 52% of the men in the control arm, who were not supposed to get PSA screening, had PSA testing done off-protocol. Yet the rigorously-applied intention-to-treat analysis demanded that these men be counted as if they had not been screened with PSA tests. With the majority of the "control" group thus contaminated, no wonder PLCO did not discern a benefit to screening versus not screening PSA (1).

In addition to giving too much credence to the flawed PLCO results, the USPSTF did not place enough emphasis on the Goteburg study, which showed robust benefits to PSA screening (1), since it was conducted prior to widespread access to off-protocol PSA testing; it may be the last trial which can ever be conducted under such pristine conditions.

There is a great deal we must do to improve how we use the PSA test, but it remains the best hope we have for maintaining the dramatic improvements in prostate cancer mortality we have seen since its widespread introduction.

Reference

1: Allan GM, Chetner MP, Donnelly BJ, Hagen NA, Ross D, Ruether JD, Venner P.Furthering the prostate cancer screening debate (prostate cancer specificmortality and associated risks). Can Urol Assoc J. 2011 Dec;5(6):416-21. doi:10.5489/cuaj.11063. PubMed PMID: 22154638; PubMed Central PMCID: PMC3235209.

On 2014 May 29, David Keller commented:

Replace the USPSTF with several committees composed of experts in the disease being considered

In medicine, as in other sciences, open questions regarding Nature are settled by experimentation. If the data from an experiment yield unclear results, the outcome of the experiment should not be declared by a committee. The experiment should be refined and repeated until the data clearly indicate the answer to the question being examined.

If we have clear, unambiguous results from well-designed experiments, we do not need a committee like the USPSTF to interpret those results and declare what they mean. Conversely, if the data is so ambiguous, arguable or suspicious that a committee is required to interpret it, then that data is inadequate to base important recommendations on.

Equipoise is the state of not being certain which of two opposite courses of action is correct. Take multivitamins or not? Screen for carotid stenosis or not? Screen for prostate cancer using PSA or not? These are all very important questions with life-altering consequences for many people if we get it wrong. Equipoise exists as long as the data are so unclear that the correct answer is not obvious, and certainly as long as a committee is required to interpret the results.

The USPSTF members are part-time volunteers who are mainly occupied by other important jobs, such as running large healthcare organizations, academic departments and research enterprises. How can a pediatrician or a gynecologist on the USPSTF, for example, be expected to master, in a few hours per week, the art and science of an unfamiliar specialty such as urology or vascular surgery, which they do not practice and to which others devote their lives? They cannot, nobody is that smart or that efficient. This is why the pronouncements of the USPSTF often meet with disbelief and non-compliance among practicing physicians. When our accumulated experience informs us that certain screening tests or treatments are helpful to patients, we require convincing data to change our practices. And, given data which is convincing, alert clinicians do not need a committee to interpret it for us. The consensus of editorials and reviews in medical journals is a reliable source of guidance when it is required.

In the years following the widespread introduction of PSA screening, there was a drop of over 35% in deaths due to prostate cancer. While this is merely observational data, from which no conclusions can be drawn regarding cause-and-effect, it is still an impressive fact, which must be explained if we are asked to stop screening with PSA. One of the main clinical trials which the USPSTF relied on to arrive at their "D" recommendation for PSA screening was PLCO (for "Prostate, Lung, Colorectal and Ovarian" cancer). PLCO was hindered by the fact that 52% of the men in the control arm, who were not supposed to get PSA screening, had PSA testing done off-protocol. Yet the rigorously-applied intention-to-treat analysis demanded that these men be counted as if they had not been screened with PSA tests. With the majority of the "control" group thus contaminated, no wonder PLCO did not discern a benefit to screening versus not screening PSA (1).

In addition to giving too much credence to the flawed PLCO results, the USPSTF did not place enough emphasis on the Goteburg study, which showed robust benefits to PSA screening (1), since it was conducted prior to widespread access to off-protocol PSA testing; it may be the last trial which can ever be conducted under such pristine conditions.

There is a great deal we must do to improve how we use the PSA test, but it remains the best hope we have for maintaining the dramatic improvements in prostate cancer mortality we have seen since its widespread introduction.

Reference

1: Allan GM, Chetner MP, Donnelly BJ, Hagen NA, Ross D, Ruether JD, Venner P.Furthering the prostate cancer screening debate (prostate cancer specificmortality and associated risks). Can Urol Assoc J. 2011 Dec;5(6):416-21. doi:10.5489/cuaj.11063. PubMed PMID: 22154638; PubMed Central PMCID: PMC3235209.