Dave's Posts

Deb Ziegler asked me on my guestbook page about where researchers get their information about morbidity and complications following heart valve surgery. Let me first say that there are privacy rules1 that generally prevent associating patients with data without consent. So the statistics are usually just that: statistics, rather than full records. Next, there are 2 basic kinds of data sets: retrospective and prospective. Retrospective studies use medical records that are generated as a matter of course by physicians regarding their patients. They cull the relevant data and assemble it into tables etc. The medical records are a hodge-podge of formats, notes, images, and so on and expensive to work with. Often studies using them will be of patients at a single hospital where there is some uniformity of format, but then the sample size is relatively small. The Society of Thoracic Surgeons maintains a database for cardiac surgery2. From their website: "The Adult Cardiac Surgery Database, now containing more than 4.5 million surgical records, represents an estimated 94 percent of all adult cardiac surgery centers across the U.S." This represents a large set of statistics. It is hard to use though, requiring a proposal, evaluation/screening, and queries performed by the database administrators. Its large size and uniformity make it a potential gold mine for generating "evidence-based" knowledge. Surgeons submit data to the database, but once patients pass out of the reporting surgeon's care I'd assume it becomes rather incomplete. Another example of a retrospective database, but grim, is the Social Security Deaths Database3. It lists the names and social security numbers of people whose deaths have been reported to the Social Security Administration. It is often used to assess survivorship in retrospective studies when patients have been lost track of in the medical records themselves. You do not want to be in this database! Then finally, there are "case studies" which dive into a particular patient's details. Case studies are a popular feature in the medical journals. Prospective studies are designed before the fact and patients are "enrolled". Consent is often obtained (required in fact under many circumstances). The data formats are rigorous and uniform. A famous example is the Framingham Heart Study4, which first showed the high heart disease risk of cigarette smoking. In prospective studies patients typically receive special exams, undergo interviews, and, in clinical trials, receive a variety of treatments to test their effects. Clinical trial studies are subject to rigorous oversight to ensure the Hippocratic oath - "do no harm" - is being upheld. In general, retrospective studies cover more patients (millions in some cases), are less specific, less costly, and result in associations between risk factors and health effects. Prospective studies cover fewer patients, are more specific, more costly, and can result in direct cause and effect relationships. Authors of each paper in a medical journal describe in detail where they got their data from. Part of understanding the results is understanding the source of the data. Obviously data quantity and quality are important issues. There is a whole field of biostatistics whose subject is the degree to which reliable conclusions can be drawn from medical data. We are all unique and our health trajectories are especially so. Our own medical data is highly valuable to each of us personally as it guides our own treatments. But it can also be used to guide the treatments of others when combined into large datasets and properly understood. So think of your medical data as a gift to the world to make life better all around. Wow Deb, I wasn't expecting to turn this into a full-fledged journal post, but your question was interesting and needed more than a superficial answer. -- DVB Footnotes 1. In the USA, the Health Information Privacy Protection Act governs the collection and dissemination of patient medical information. 2. Society of Thoracic Surgeons Database (https://www.sts.org/national-database) 3. Social Security Administration's Death Master File (https://www.ntis.gov/products/ssa-dmf.aspx) 4. Framingham Heart Sudy Wikipedia artcle. (https://en.wikipedia.org/wiki/Framingham_Heart_Study)

If you've been diagnosed with severe mitral regurgitation, raise your hand. This post is for you. When they do various tests on you and tell you the results you are probably thinking something like: lots of people before me had these same test results, what happened to them? It's a darned hard question to answer, but you are in luck because there's been some work on exactly this question for mitral regurgitation patients. These studies are hugely useful for planning your future, so let's dive in to one1 that has been influential. First off, this was a prospective study, which means that the study participants were chosen before being poked, prodded, and results tallied. 456 patients were given echocardiograms and then their medical histories were followed for years. This is a powerful method for exploring correlations between diagnostic data and outcomes, in this case the correlations between various quantitative measures from their echo exams and their heart health. You can measure a lot of things off an echo and some of these are more useful than others. A useful number is one that can be used to guide treatment for good outcomes. And according to this paper the most useful number2 is the "effective regurgitant orifice" area (ERO), which is a measure of the size of the opening between your mitral leaflets that the wrong-way blood is going through. It makes sense that the bigger this opening the worse things are, and this is borne out by the statistics. When the study participants are stratified, or grouped in ERO ranges, the picture that emerges is dramatic. Here I am linking to Kaplan-Meier aka survival curves3 for the study participants in various ERO bins. Figure 1. Kaplan—Meier Estimates of the Mean (±SE) Rates of Overall Survival among Patients with Asymptomatic Mitral Regurgitation under Medical Management, According to the Effective Regurgitant Orifice (ERO). Values in parentheses are survival rates at five years. Credit: The New England Journal of Medicine © 2005, used with permission. These curves are for overall survival, which means that those that dropped out did so for heart-related reasons as well as other causes of death. They show very clearly that larger ERO values are worse, and that those of us with ERO values larger than 40 square millimeters have only a slightly better than even chance of being around five years on. A companion figure in the paper zeroes in on the rates of death in the study group from cardiac causes only (below)4. Figure 2. Kaplan—Meier Estimates of the Mean (±SE) Rates of Death from Cardiac Causes among Patients with Asymptomatic Mitral Regurgitation under Medical Management, According to the Effective Regurgitant Orifice (ERO). Values in parentheses are the fraction of deaths after five years. Credit: The New England Journal of Medicine © 2005, used with permission. If your ERO is less than 20 square millimeters, then you are running a 3% risk of death from cardiac causes over the next 5 years. 40 or over? 36% chance. That doesn't sound so hot if your ERO is over 40, like mine was. But the good news is that surgical mitral valve repair prior to the onset of symptoms restores normal life span5. So there's the value of your ERO measurement. It lets you know which of these cohorts you are in, and then looking at the survival curves what the historical experience of that cohort is. The bottom line here is be sure your cardiologist gives you your ERO number since it contains high quality information needed to guide your course of treatment. Keep the beat -- DVB Footnotes 1. The paper is freely available: Enriquez-Sarano, Avierinos, Messika-Zeitoun, Detaint, Capps, Nkomo, Scott, Schaff and Tajik, Quantitative Determinants of the Outcome of Asymptomatic Mitral Regurgitation, New England Journal of Medicine volume 352, page 875 (2005). (https://www.nejm.org/doi/pdf/10.1056/NEJMoa041451) 2. The authors considered a large number of quantitative measures and determined which had the best utility in terms of predicting outcomes using statistical methods. 3. See my journal entry on July 3, 2012 titled "I was expecting more time with you my love". 4. If you carefully compare these two figures you will also note that the death rate from non-cardiac causes is significantly higher in the larger ERO groups, even though the average age was younger. This means that severe mitral regurgitation makes you more susceptible to other potentially fatal problems. 5. See my journal entry on July 15, 2012 titled, "What happens if I don't get my mitral valve fixed?"