Reconciling Quality metrics and patient choice

As readers of my blog probably realize, I have become increasingly an advocate of data collection and data/metric driven medicine. At this point in my career, I have come to realize that you can't improve if you don't know where yo u are starting from. In order to define baselines you need to measure something(s). In order to develop feedback loops we need to embed data collection into what we do.

I was doing my usual scan of the literature this weekend and I came across an article in the Annals of Internal Medicine looking at the benefits and risks of screening mammography.

Overdiagnosis of Invasive Breast Cancer Due to Mammography Screening: Results From the Norwegian Screening ProgramAnn Intern Med April 3, 2012156:491-499;

Yet again the results of large scale screening fail to demonstrate broad and unambiguous benefits of using mammography to prevent death from breast cancer and appear to result in over diagnosis and over treatment of what appears to be biologically benign disease. In the same issue were a series of letters addressing an earlier study of screening for prostate cancer with PSA with similar results.

The Annals has created a library of patient materials based upon Annals studies and in this issue of the Annals a breast cancer and mammography is also published. http://www.annals.org.proxy.library.emory.edu/content/156/7/I-58.full.pdf+html

The text is copied below....

What is the problem and what is known about it so far? - Women are advised to get routine mammograms to find breast cancer early, before it would have been noticed on examination. This is called screening mammography. However, some cases of breast cancer found with screening mammography would never have been noticed and therefore would not have led to death.

 Why did the researchers do this particular study? -To see how common screening mammography leads to overdiagnosis of breast cancer.

 Who was studied? -Women in Norway who had a diagnosis of invasive breast cancer from 1986 to 2005.

How was the study done? - In Norway, screening mammography was introduced or “rolled out” over a decade, county by county. The number of cases of breast cancer found in counties with screening was compared with the number of cases found in counties without screening during the same period.

What did the researchers find? - About 15% to 25% of cases of breast cancer found through routine mammography were overdiagnosed; that is, they would never have been found on examination or led to death. 

What were the limitations of the study? - The study included women diagnosed with invasive breast cancer, not those with ductal carcinoma in situ, a different form of breast cancer. The study was not a randomized trial where women were or were not given mammograms depending on chance (like flipping a coin).

What are the implications of the study? - For many women, screening mammography may lead to a diagnosis of breast cancer that would never have been noticed or caused harm during their lifetime. These women would have been unnecessarily treated with therapies, including surgery, radiation, and chemotherapy, that have adverse effects.

The underlined text identifies a principle which can be broadly applicable to almost everything we do. We make recommendations based upon information which ideally we share with our patients and they make decisions regarding whether they will adopt our recommendations. If physicians are rewarded for patients making particular decisions, it creates an incentive for physicians to frame decisions so they benefit. For example, the NQMC has a quality measure entitled:

Preventive care and screening: percentage of female patients who had a mammogram performed during the two-year measurement period.

The description of the measure is the following:

This measure is used to assess the percentage of female patients aged 50 to 69 years who had a mammogram performed during the two-year measurement period.

The math is straight forward:

Denominator Description

All female patients aged 50 to 69 years at the beginning of the two-year measurement period

Numerator Description

Female patients who had a mammogram performed

And there is an out for those who decline, although the documentation requirements are vague. My guess is it will be much easier hide the patient materials from the Annals and frame the discussion where everyone agrees to be screened.

There are a host of other measures relating to issues such as prostate cancer screening, hypertension control, anti-coagulation associated with atrial fibrillation, and colon cancer screening to name but a few.  In each of these contexts, physicians performance will be assessed by measuring what percentage of their patients were treated or screened. The interventions in question often are unlikely to directly benefit those to which they are applied and it is not unreasonable to predict that if the possible benefits are framed in such a way to convey that information, many patients would decline the interventions if they are fully briefed.

The data may show that much of what we push is essentially worthless but that may not matter. "Quality" metrics will reward physicians for promoting dubious interventions. There is no question in my mind that we can talk most patients into basically anything and we are most inclined to talk them into things that can be explained quickly and are financial winners. We are least inclined to spend time explaining nuanced concepts which undermine the economic basis for our practices. Few are inclined to spend time telling patients that the very tools heavily sold as saving lives may not be worthwhile. The people you explain this to may get cancer. Send  them for a mammogram and find a cancer you are a hero.  It appears the genie is out of the bottle and will not be put back in.