About Those Skipped Heart Test Results

Harlem Hospital Center stands just three miles or so north of my home. I know the place from the outside glancing in, as you might upon exiting from the subway station just paces from its open doors. The structure seems like one chamber of its neighborhood’s heart; within a few long blocks’ radii you’ll find rhythms generated in the Abyssinian Baptist Church; readings at the Schomburg Center and artery-clogging cuisine at the West 135th Street IHOP.

So I was saddened to hear about the missed heart studies. Or should I say unmissed? No one noticed when nearly 4,000 cardiac tests went unchecked at the Harlem center, a public hospital managed by the city’s Health and Hospitals Corporation. The skipped beats began sometime in 2007.

According to the Times report, that’s when hospital administrators, hurting perhaps for doctors sufficiently skilled in reading echocardiograms, OK’d a process by which technicians scanning the images would alert the responsible physicians if they noticed abnormalities. Otherwise they stored the results – pictures of the heart’s contractions, wall thickness and size, valves and some large vessels – for review, later.

Usually when a person gets an echocardiogram there’s a reason. Mine, for example, was done before I received a chemotherapy drug, adriamycin that can affect the heart’s function and, another time, before I had a major operation – basically to make sure my heart was strong enough to handle the stress of surgery. Years earlier, I’d had an echo (as doctors sometimes call these tests) to evaluate shortness of breath I experienced while pregnant. I like echocardiograms, as cardiac imaging methods go, although I must admit I find the blobby representations cryptic if not frankly rorschachian. These tests rely on ultrasound, the same technology we routinely use to examine unborn fetuses by projecting and canvassing sound waves. There’s no radioisotope or x-rays. Not even a magnet’s involved.

Echocardiogram reveals 4 heart chambers - adapted from Wikimedia Commons

What generally happens is that after the procedure a doctor, usually a cardiologist, inspects the images and provides a written assessment. Ideally, the test report reflects the reason for doing the procedure. So if a teenage soccer player has an echo to evaluate an episode of fainting on the field, the physician-reviewer would focus on structural heart abnormalities associated with sudden death in some young athletes. Sometimes the studies reveal enlargement of the heart; this can occur in alcoholics, in people with chronic forms of severe anemia like sickle cell disease, and in other conditions. For patients with atrial fibrillation – a disorder in which the heart flutters irregularly – doctors might look to see if there’s clot inside the heart’s walls that might, unmitigated, migrate through the arteries to the brain. Echocardiogram can assess the heart’s condition after a heart attack or in congestive heart failure. They can visualize holes in the heart chamber walls of infants, lapsed valves and more.

The Times story indicates that doctors didn’t review images for over half of the echocardiograms performed at Harlem Hospital since 2007. The medical center, staffed by doctors from Columbia University, had six attending cardiologists and six fellows in 1999, according to the paper. Now the hospital has only three full-time cardiologists and lacks a fellowship program. The hospital runs approximately 2,500 echocardiograms each year. Among those 4,000 patients whose tests went unread, some 200 have died since the time of the procedure. Hospital officials say it’s unlikely that any deaths are attributable to the lapse.

Since the story emerged last week, a squad of doctors has been scrambling to review the images. Heads rolled at Harlem Hospital: the clinical director was fired and the medical director has been demoted. An investigation, led by Dr. John N. Morley of the State Health Department, is underway. The press, or at least my local newspaper, is all over the matter.

So what’s to be learned from this oversight? My take’s two, so far:

1. It appears that at least some physicians working at Harlem Hospital felt it was understaffed and that they were too overworked to meet their clinical responsibilities, and that the administration did not adequately address their concerns. And while Health and Hospitals Corporation has indicated this problem is unique to that particular department – the echo lab – at one hospital, I’m not convinced.

Having worked for years in hospitals where cardiologists, gastroenterologists, hematologists and even pathologists spend much of their time putting out fires, so to speak, it’s scarily easy for me to envision how non-urgent tests could pile up without review. When hospitals operate with money as a bottom line, the difficult work doctors do doesn’t get easier. So we might blame individual physicians for not signing those reports. But I’d take the system to task, and not just at one Harlem hospital.

2. No one’s mentioned the patients’ role in all of this, which seems strange to me. These days, we expect that most patients will enter into discussions with their physicians about what tests they need done. Maybe at a medical center like Harlem Hospital, which serves a relatively poor population, the expectations differ regarding patients’ involvement in medical decisions. But if that is the case, those separate standards reflect another problem – of poor communication between physicians and their patients – equally demanding of our attention.

Lastly, as I’ve said previously here and elsewhere, we waste a lot of medical resources by ordering procedures without thinking. If a person undergoes a medical test there should be a reason for it, sufficient that either the doctor or the patient cares enough to find out the results.

Related Posts:

DNA Comes Home, or Maybe Not

Earlier this month employees at most of 7500 Walgreens pharmacies geared up to stock a new item on their shelves: a saliva sampler for personal genetic testing. On May 11, officials at Pathway Genomics, a San Diego-based biotech firm, announced they’d sell over-the-counter spit kits for around $25 through an arrangement with the retailer. A curious consumer could follow the simple package instructions and send their stuff in a plastic tube, provided in a handy box with pre-paid postage, for DNA analysis.

DNA helix structure (Wikimedia Commons)

Once the sample’s in the lab, the cost of genomics testing depends on what, and how much, you want to know. Pathway offers a variety of options. A pre-pregnancy planning report would check to see if you carry mutations for each of 37 inherited diseases – conditions as varied as beta thalassemia, cystic fibrosis and familial Mediterranean fever – for $179. A profile of tests for genes involved in metabolizing specific drugs, such as Plavix and Coumadin, goes for $79. A vaguer, health conditions panel suggests a propensity to develop particular diseases including Type II diabetes and melanoma. This series runs $179 if purchased separately, but might be had for less through a discounted package rate. A genetics ancestry evaluation lists for $249.

Within two days, after some controversy and a receipt of a letter from the FDA Office of In Vitro Diagnostic Device Evaluation and Safety addressed to James Plante, Founder and CEO of Pathway (dated May 10), Walgreens nixed the plan. Now, Congress wants to know more about direct-to-consumer personal genomics testing. On May 19, the House Committee on Energy and Commerce sent letters to Plante and the CEOs of two major competitors – Navigenics and 23andMe. House Committee Chairman Henry Waxman and colleagues have some questions about how samples are processed and the accuracy of the results:

“The Committee is requesting information from the companies on several aspects of the tests:  How the companies analyze test results to determine consumers’ risk for any conditions, diseases, drug responses, and adverse reactions; the ability of the companies’ genetic testing products to accurately identify any genetic risks; and the companies’ policies for the collection, storage, and processing of individual genetic samples collected from consumers.

The Federal Trade Commission has cautioned consumers about genetic testing kits since 2006.  Still, personal genomics tests are readily through available on-line sales. You can get the 23andMe “DNA Test for Health and Ancestry Information” from the manufacturer or at Amazon.com for $499. Navigenics takes a distinct approach by marketing its genetic tests strictly as a laboratory service for medical practitioners and so, thus far, avoids some rules regarding in vitro diagnostic tools.

New York State, my home, is one region where Walgreens wouldn’t have sold the kits in stores. That’s because of stricter state laws regarding genetic testing.

Dan Vorhaus, writing for the Genomics Law Blog, provides a considered analysis:

At present, whether a genetic test is subject to FDA regulation largely depends on how it is developed and marketed. The literature, as well as current FDA regulatory policy, divides genetic tests into two primary categories:

(i) in vitro diagnostic test kits (also sometimes referred to as IVD kits or, simply, as genetic test kits), which may be sold by their manufacturers directly to consumers, testing laboratories, clinicians or other approved recipients, depending on the device; and

(ii) laboratory developed tests (or LDTs, also sometimes referred to as “home brew” assays), which are not sold directly to the general public or to physicians; rather, a testing service (as opposed to the actual test itself) is marketed, and samples (e.g., of saliva) are collected and submitted to the laboratory for evaluation.

The FDA regulates IVD kits as medical devices…

Up until now personal genomics testing companies have had few constraints, and little profit. What’s clear from the recent news is that we’ll be hearing more about these kits – their manufacture, distribution, accuracy and interpreting results. And that doctors, for our part, have some serious studying to do. Whether the test results go directly to patients, or not, they’re sure to raise many legitimate questions. We’ll need some solid answers about the testing process in itself, besides meaningful responses about what’s found in our DNA.

Related Posts:

Why Blog on OncotypeDx and BC Pathology?

A few days ago I wrote on a relatively new pathology tool called OncotypeDx. This device measures expression of 21 genes in tumor cells to establish the likelihood a cancer will recur. For women with early-stage, ER+ tumors that haven’t yet spread to the lymph nodes, the OncotypeDx results stratify patients into three groups – those having a low, intermediate or high risk for recurrence at 10 years.

As things stand, chemotherapy is routinely prescribed for most women with early-stage BC after initial treatment by mastectomy or lumpectomy and radiation. But the overall relapse rate is fairly low (around 15% at 5 years, higher over time depending on other factors) for women who take an anti-estrogen pill like tamoxifen. Chemo reduces the recurrence rate by approximately one third. The problem is that women and their doctors don’t know in advance who’s likely to benefit.

Here’s why this is important:

What happens now is that most women choose to undergo treatment even though it’s unlikely their cancer will come back. This – the problem of overtreatment – was one of the main concerns to emerge from the mammography screening debate.

The original OncotypeDx data, which have been considered here and elsewhere, support that most women with low recurrence scores are unlikely to benefit from chemo. So if women and their doctors could access the kind of information provided by OncotypeDx, at a cost of ~$3800 each, tens of thousands of women with BC and low risk scores might opt out of chemo treatments each year.

For example, if a woman’s recurrence score is less than 18, the likelihood of a relapse within 10 years is only 7%. Such a patient might happily and rationally choose not to take adjuvant chemotherapy.

I can’t even begin to think of how much money this might save, besides sparing so many women from the messy business of infusions, temporary or semi-permanent IV catheters, prophylactic or sometimes urgent antibiotics, Neulasta injections, anti-nausea drugs, cardiac tests and then some occasional deaths in treatment from infection, bleeding or, later on, from late effects on the heart or not-so-rare secondary malignancies like leukemia. And hairpieces; we could see a dramatic decline in women with scarves and wigs.

So why doesn’t every woman with eligible (ER+, node-negative) BC get an OncotypeDx readout, or some other modern pathology report, such as Mammaprint, that’s available and already FDA-approved? (OncotypeDx is just an example, really, of an advance in science that’s moving at a snail’s pace into the clinic.)

One issue, perhaps, is that it’s challenging for some doctors to learn about this test sufficiently that they’re comfortable with it. Quantitative RT-PCR, the method by which RNA is measured in the assay, wasn’t invented until around 1990, long after many practicing oncologists completed school. And as for the particular 21 genes measured – they’re unfamiliar to most physicians I know. Now, you might say that it doesn’t matter – if the device works, the doctor doesn’t have to understand the underlying technology. But a black box-like approach to clinical cancer decisions is far from ideal.

From the physician’s perspective, it may be easier, and perhaps legally safer, simply to prescribe the chemo – which she knows well and uses all the time – than to engage in a decision-making process using new methods and terms she doesn’t fully command.

Besides, there’s a conflict of interest: many oncologists, hospitals and infusion centers make money by giving infusions of chemotherapy. Identifying a large subset of patients who wouldn’t benefit from chemo may not be a priority for some clinicians. In a recent JCO paper, half of the oncologists’ initial recommendations for a combination of chemotherapy and hormonal treatments changed to hormonal treatment (without chemo) upon seeing the patients’ OncotypeDx scores.

An ongoing large, NIH-sponsored TailorRx trial involving 10,000 patients aims to clarify the potential benefit of this test. I’m concerned that by the time those results are available, with relapse rates and survival curves at 5 and 10-years, the technology in science and availability of new treatments may persuade doctors, then, to question the trial’s relevance. Meanwhile, hundreds of thousands of women will be treated off protocol, many without this sort of information, in a sort of blinded guessing game about the chances of recurrence and whether or not they should take chemo.

For now I hope that all women with newly-diagnosed BC, and their physicians, know about OncotypeDx and other tools, and their potential to inform decisions regarding chemotherapy.

Related Posts:

More News, and Considerations, on OncotypeDx

This week I’ve been reading about new developments in breast cancer (BC) pathology.

At one level, progress is remarkable. In the 20 years since I began my oncology fellowship, BC science has advanced to the point that doctors can distinguish among cancer subtypes and, in principle, stratify cases according to patterns of genes expressed within tumors. This sort of information – cancer cell profiling – might inform prognosis and influence treatment decisions that BC patients and their doctors, usually oncologists, make every day.

What disappoints is the slow pace by which this knowledge infiltrates the clinic. In practice, women and their physicians rarely have much more information on BC pathology than what was available two decades ago – the tumor size in its largest dimension (crudely measured in centimeters), whether it’s spread to the lymph nodes (and if so, how many nodes), the type of cancer (based on the cells’ appearance under a light microscope: infiltrating ductal, lobular carcinoma and other BC forms) and whether the cells express a few key molecules including estrogen receptors (ER).

In the past five years, more laboratories are offering data on Her2 in BC samples. This complex molecule, an epidermal growth factor receptor, normally transmits signals from a cell’s surface to the interior. Her2 expression dictates the BC subtype in some newer classifications of the disease and usually determines the cells’ responsiveness to Herceptin, a monoclonal antibody treatment. Still, there’s been some controversy, in part due to variation among lab facilities in the reproducibility of Her2 testing results.

The problem is this: if pathologists don’t provide accurate, valid results on Her2 expression in BC cells – which can be measured by various methods – it’s hard for women and their physicians to make sound decisions based on the molecule’s expression. And Her2 is just one of dozens of molecules that can be measured in BC. The reason it’s tested, for the most part, is to foster decisions on Herceptin treatment and also, perhaps to a lesser extent, to provide prognostic information.

What puzzles me is why so few use better, modern pathology and other decision tools. Technologies like Mammaprint, Adjuvant! and OncotypeDx have been available for years but aren’t used routinely in most clinical settings. So I thought I’d do some more research and, in future posts, will consider each of these and other, relevant technologies.

For today I’ll focus on OncotypeDx.  This test, manufactured by the Redwood City, California-based Genomics Health, assesses BC recurrence risk in cases that are ER+, node negative (see below). As cancer gene testing panels go, OncotypeDx is a baby, based on expression of just 21 genes by a two-decade old method called quantitative RT-PCR. The test intrigues me; I’ve posted on it once before.

No doubt, my interest in OncotypeDx is intensified by my personal history of BC. My case was exactly the sort of ER+, node-negative tumor for which OncotypeDx is intended; often I’ve wondered what would have been my tumor’s recurrence score (RS) and if knowing that would have affected my decision to undergo treatment with adjuvant chemotherapy.

Some background terms –

ER+ means that the cells express hormone receptors, for estrogen, at the surface;

Node negative means that the breast cancer has not spread to the lymph nodes, or glands, of the armpit. (Axilla is the medical term for armpit. Axillary lymph nodes are normal immune organs that drain fluid including potentially foreign particles from the breast, chests and nearby arm. The nodes can swell if there’s an infection to which the body reacts, if malignant cells infiltrate the gland and sometimes due to autoimmune diseases like lupus.)

So an ER+, node negative breast tumor is one in which the cancer cells are sufficiently differentiated, or mature, to produce and bear hormone receptors at their surfaces and in which the tumor cells haven’t yet migrated to the armpit (or at least haven’t done so at a level that can be detected by a pathologist).

Real-Time, Reverse Transcriptase (RT) – Polymerase Chain Reaction (PCR) is a standard method for amplifying tiny amounts of nucleic acids such that they can be measured and sequenced. Standard PCR usually amplifies DNA whereas in RT-PCR, RNA transcripts are converted to DNA before amplification in a machine. This method can assess the amount of RNA, or message for a particular gene, that’s expressed in a pathology sample.

Adjuvant therapy refers to additional, or extra, treatment that’s given after initial cancer surgery to reduce the chances of the tumor’s recurrence.

Back to OncotypeDx –

This pathology tool predicts the likelihood that ER+, node-negative BC tumors will come back within 10 years of a woman’s primary treatment (mastectomy, or lumpectomy with radiation) usually followed by tamoxifen. The assay measures each of 21 genes in a panel and, using those results, calculates a “recurrence score” (RS) between 1 and 100. The higher the RS, the more likely the cancer will re-emerge after treatment.

According to the Genomics Health website, the test measures RNA in BC tumor specimens for the following transcripts:

Groups of genes measured in OncotypeDx assay, according to the manufacturer
cell proliferation tumor invasiveness growth factor receptors hormone responsiveness other genes of interest reference** genes



Cyclin B1


Stromelysin 3

Cathepsin L2




PR (progesterone receptor)






Beta actin





*In my opinion, survivin and bcl-2 might be better classified distinctly; the products of these genes inhibit apoptosis (programmed cell death).

**These “housekeeping” genes are not of known significance in BC pathology. Rather, they serve as controls in the assay for the quality of the RNA sample, and for comparison to other measured genes.

The OncotypeDx results are reported by risk group, as follows:

Low risk (RS <18, the 10-year recurrence rate was 7% in NSABP study – see below)

Intermediate (RS 18 – 30, the 10-year recurrence rate was 14%, in same);

Higher risk (RS >31, the 10-year recurrence rate was 30%, in same).

The tool has been tested in multiple clinical trials for its capacity to predict BC recurrence after surgery and tamoxifen in women with ER+, node-negative tumors. The study most-cited, and from which the above statistics are drawn, was published in the New England Journal of Medicine in 2004, based on a retrospective analysis of 668 cases by Genomics Health in collaboration with investigators of the National Surgical Adjuvant Breast and Bowel Project (NSABP, a large, NIH-sponsored, multicenter cancer research effort).

OncotypeDx has been on the market since 2004. The cost of one assay runs near $3800, and most U.S. insurance plans including Medicare will cover it. Tumor samples, set in fixative, are sent to a single lab – a Genomic facility – that’s regulated according to the Clinical Laboratory Improvement Amendments of 1988 (CLIA). The whole process takes 10-14 days. Still, the FDA has not approved the test for use as a decision-making tool.

Meanwhile, an NCI-sponsored trial called TAILORx will recruit and evaluate 10,000 women with ER+, node negative disease. Those investigators will determine, prospectively, if decisions based on OncotypeDx results can safely spare women with low RS the side effects and toxicity of chemotherapy without compromising their survival.

Why Oncotype and other new BC pathology tools matter –

In the U.S., the number of women who learn they have an ER+, node-negative BC approximates 100,000 per year. The question of adjuvant therapy – whether a woman should take tamoxifen or another hormonal agent and/or chemotherapy after surgery to reduce the risk of recurrent disease – is crucial.

If patients and their doctors could access more detailed molecular information about each case, they’d have a better sense of whether adjuvant treatment is likely to help in their particular situation. This approach would, potentially, spare many individuals with early-stage BC the costs, toxicity and hassle of unneeded chemotherapy. At the same time, it would help patients with small but riskier tumors by informing them that they have a high RS and thereby would more likely benefit from added therapy. Fewer women would receive chemotherapy, driving down costs, and the risks of additional treatment would be assumed only by those with a high likelihood of recurrence.

Some numbers here might help:

Overall, for women with ER+, node-negative tumors the chances of cancer recurring five years after primary treatment (mastectomy, or lumpectomy and radiation) followed by tamoxifen are around 15%. Over time that risk rises – BC can strike back after 10, 15 years or even later; the recurrence rate is said to approach 30% over time. In general, a basic chemotherapy regimen – something like CMF – cyclophosphamide (Cytoxan), methotrexate and 5-fluorouracil (5FU) reduces the probability of recurrence by about a third.

So if 100 women with node-negative tumors have to decide whether to take chemotherapy after surgery +/- radiation, or not, without a tool like OncotypeDx or another modern pathology test, they’re making those decisions based on very crude approximations of their odds. Because they don’t know whose tumors will recur, most if not all of their oncologists will recommend chemotherapy. And most women do choose to undergo the extra treatment because they’re afraid that, otherwise, there’s a greater chance the cancer will come back.

This is exactly the situation I faced in November, 2002, when I had an ER+, node negative, 1.5 cm tumor. Then, I reasoned that BC tends to be more aggressive in younger women. With hopefully more decades ahead in my life – more time, in effect, for the disease to recur – an 85% disease-free rate at 5 years wasn’t good enough. So I went for the chemo and upped my chances to the 90% range. Not a big difference in the stats, but I wanted to position myself on the upper branch of that Kaplan-Meier curve. Now, had I known my recurrence score based on the pattern of gene expression in the tumor cells, that information would have been useful. But it wasn’t an option then and, unfortunately, it’s still rarely available to most women who are undergoing treatment for BC in 2010.

The slow pace of progress, science in hand, is kind-of shocking.

So what’s new with OncotypeDx?

Two months ago, I reviewed a small study published in the ACS Cancer journal on the experiences of most of 100 women with newly-diagnosed breast cancer whose oncologists used the OncotypeDx assay to evaluate their cases. In that, two-thirds of the women reported they “understood a large amount or all” of what the doctors told them about the results and nearly all said they would undergo the test if they had to decide again.

In its April 1 issue the Journal of Clinical Oncology (JCO) published two relevant reports and an editorial. These papers support that OncotypeDx offers useful information to women with early-stage breast cancer and that it can assist patients and doctors in care decisions, in some cases providing support for them to choose a chemotherapy-free treatment regimen.

One study, a “Prospective Multicenter Study of the Impact of the 21-Gene Recurrence Score Assay on Medical Oncologist and Patient Adjuvant Breast Cancer Treatment Selection” by Dr. Shelly Lo and colleagues, followed the analysis and prescribing patterns of 17 medical oncologists at 3 diverse academic medical centers and one community hospital. Genomic Health, provided free OncotypeDx kits and testing at their central lab for all 93 patients with ER+, node-negative BC who enrolled in the trial.

The mean age of the women was 55 years (range 35 – 77). The oncologists were asked to state their treatment preferences (hormonal treatment with or without chemo) before and after receiving the OncotypeDx results for their patients. What happened was this:

Before seeing the OncotypeDx results, the oncologists recommended chemo and hormonal therapy (CHT) to 42 of the 89 women for whom the study was completed. In 20 of those 42 cases (22% of the total, and nearly half of those women who were to receive chemo) the doctors changed their recommendation from CHT to HT (hormones only) upon reviewing the OncotypeDx report. In 8 cases, the oncologists switched their recommendation to include chemotherapy. In total, the OncotypeDx results influenced the oncologists’ preferences in 31% of the cases – nearly a third.

As for the patients – 74 of the 89 (83%) said the OncotypeDx results influenced their treatment decision. The assay report persuaded 9 patients in the group to opt for a less aggressive (chemo-free) approach. The majority (78 women, 95% of those responding) said they were glad they used the OncotypeDx assay. According to the paper, many patients felt reassured by the assay findings and benefited from a diminished perceived risk of recurrence (less worry, in effect).

The upshot is that the OncotypeDx assay – which costs around $3800 per evaluation – led to significantly fewer women with early-stage breast tumors getting chemotherapy in this trial of 89 patients. The doctors were more confident in their decisions to not give chemotherapy in cases with low RS and, overwhelmingly, the women felt glad about the decision-making process.

In the second JCO study in the April 1 issue, the number of patients evaluated was much greater – over a thousand. But this was a more complicated analysis in which the investigators applied OncotypeDx to old tumor samples and then, upon reviewing those cases in some well-documented randomized trials, examined how the cases fared in relation to the RS. What they found was that OncotypeDx score predicted the likelihood of loco-regional recurrence (LRR) in women who had node-negative, ER+ disease.

Bottom line –

The OncotypeDx tool has been on the market for 6 years. It has, in multiple and well-done studies, identified patterns of BC gene expression that accurately predict the likelihood of recurrence in women with early-stage, ER+, node-negative tumors. This should, in principle, reduce administration of chemotherapy – along with its attendant risks, costs and side effects – to women whose tumors are unlikely to relapse. Recent trials show that doctors find the results useful and that patients find it helpful in their decisions.

I can’t know for sure why the tool’s not used more often. But I have some concerns:

1. It takes time for doctors – even knowledgeable oncologists – to learn about this device, to know how it differs from other BC pathology tests like Mammaprint and decision tools (like Adjuvant!) and then it takes even more time for those physicians to discuss the results with their patients.

From the perspective of a physician sitting behind her desk or at a table with a newly-diagnosed BC patient, saying “this is what I think, you need treatment X” may be a lot easier than “well, let’s go over these OncotypeDx results…”

2. If the OncotypeDx report does indeed identify large subgroups of early-stage breast cancer patients who don’t need chemotherapy, the use of this test would reduce the number of patients who get chemotherapy. Oncologists, infusion centers and others generate income by prescribing chemotherapy. So there’s a potential conflict of interest.

3. Perhaps some physicians fear lawsuits for not giving chemotherapy to women who, without the OncotypeDx results, would receive it.

4. Some doctors might not recommend OncotypeDx because they don’t really understand the test, its merits and limitations.

5. Maybe OncotypeDx isn’t the best of the new BC adjuvant therapy decision tools. For this reason, among others, I will consider some of the other methods available in future posts.

Related Posts:

The Checklist and Future Culture of Medicine

Like many New Yorkers, I learned about the checklist in a magazine. I remember thinking, in late 2007, that maybe I’d seen something on the subject in The New England Journal. Indeed, a year earlier Dr. Peter Pronovost and colleagues reported on a simple, inexpensive strategy to save lives in a now-landmark article, “An Intervention to Decrease Catheter-Related Bloodstream Infections in the ICU.” Still, I’d missed the paper. Or at least I’d overlooked the significance.

Fortunately I had the opportunity to hear Pronovost, a still-youngish Professor at Johns Hopkins and recent MacArthur award recipient, speak at the annual meeting of the Association of Health Care Journalists in Chicago. After hearing his talk, I couldn’t wait to read more.

The checklist refers to 5 steps doctors can take to reduce the likelihood of patients getting serious infections from catheters placed in the ICU. One problem with Pronovost’s quintet is that it’s, well, unexciting. In his book Safe Patients, Smart Hospitals he reveals what a person shouldn’t forget before inserting an intravenous (IV) tube through a vein to the heart’s entryway:

1. wash hands with something like soap before the procedure;

2. set up a clean work area by covering the patient with a sterile drape and donning a gown, cap, mask and sterile gloves;

3. insert the catheter in a place other than the patient’s groin, if possible;

4. wipe down the patient’s skin with antiseptic fluid, chlorhexidine;

5. remove catheters that are no longer needed.

Pronovost, an intensive care specialist who holds an MD and a PhD in Public Health, first tested the checklist on his home turf, the surgical ICU at the Johns Hopkins Hospital in Baltimore, MD, in 2001. At the start, he distributed the list and asked ICU staff nurses to mark physicians’ compliance. It turned out the doctors skipped at least one step in over a third of central catheter placements. Next, he upped the list’s power by talking to Hopkins administrators. Nurses, they said, could call out a physician if they didn’t stick to the rules.

“This was revolutionary,” said Atul Gawande in the New Yorker.

What’s the big deal, you may wonder. It’s this – first, in the usual culture of medical practice, doctors don’t follow orders but give orders. And second, what’s implicit in the checklist is that physicians – even at one of the world’s most renowned medical facilities – are fallible to such a degree that their work can improve, and measurably so, by using something as ordinary as a checklist. It’s humbling.

“We don’t use checklists in health care because we still have his myth of perfection,” Pronovost said at the journalism conference.

In the year after Pronovost’s team implemented the checklist at Hopkins, the rate of central catheter infections there dropped from eleven percent to zero. As for how much good this did – the estimate runs at 43 infections spared, 8 deaths avoided and $2 million saved in one year at that hospital alone. The work expanded, soon to cover ICUs in most hospitals in the state of Michigan. There, after a lot of fuss, administrative hurdles and number crunching of results for some 375,757 catheter-days’ worth of infection data, the incidence of central line-associated bacterial infections snapped from 2.7, on average, for every 1000 days a patient was in a Michigan ICU with at least one central line, down to 0 (zero!).

These numbers are supported by impressive stats, with p-values falling below 0.002 in the original study. Estimates for the Keystone Initiative render some 1000 lives saved and $175 million in hospital costs reduced in a single year in Michigan. What’s more, all of this was accomplished without the use of expensive technology or additional ICU staffing.

This is a win/win intervention with huge implications. Every day some 90,000 people receive care in ICUs in North America. The annual incidence of catheter-related blood infections is 80,000 per year in the U.S.; the cost of treating each line infection runs around $45,000. In the U.S., we might save over $3 billion in expenses per year.

So why aren’t more hospitals and states adopting these and other, similar measures? Gawande addresses this, to some extent, in the New Yorker piece and in his book, The Checklist Manifesto. “There are hundreds, perhaps thousands, of things doctors do that are at least as dangerous and prone to human failure as putting central lines into I.C.U. patients,” he writes. “All have steps that are worth putting on a checklist and testing in routine care. The question – still unanswered – is whether medical culture will embrace the opportunity.”

Poka-yoke, a Japanese term for rendering a repetitive process mistake-proof, may be familiar to business students and corporate executives.  This concept, that simple strategies can reduce errors in highly complex works, is not the kind of thing most doctors pick up in med school. Rather, it remains foreign.

Pronovost is unusual because he examines health care delivery, in itself, rather than attempting an innovative cure for cancer or surgical method. His work just isn’t sexy enough to sell. I suspect that’s the reason he came to the health care journalism conference in Chicago and gave such an impassioned talk about the checklist, so that a few of us might help get the word out.

Things change, after all, and sometimes they do get better.

Related Posts:

newsletter software
Get Adobe Flash player