CardioExchange Archive

CardioExchange, an NEJM practice community for medical professionals dedicated to improving cardiac patient care, was active from 2009 to 2015. CardioExchange fostered discussion between clinicians from across the globe.

December 2nd, 2013

Left Ventriculography: Procedural Variation Writ Large

The Dartmouth Atlas Project has documented that the use of many procedures varies widely by region and by institution, raising the question of whether much of the care in high-use areas may be unnecessary. At Palo Alto’s Veterans Affairs Health Care System, where I practice, our cardiology fellows noticed that the use of left ventriculography varied markedly depending on whether coronary angiography was performed at the VA itself (lowest use), at the university hospital by academic staff, or by private physicians (greatest use).

So we conducted a study, now published in Circulation: Cardiovascular Quality and Outcomes, of the use of left ventriculography during coronary angiography across the VA Health Care System nationwide from 2000 to 2009. We found an overall rate of 58% but with dramatic variation: as high as 95% in some facilities and lower than 1% in others. Differences in reimbursement do not explain the variation, given that VA providers are salaried and that most private payers don’t reimburse much for left ventriculography. Other factors must, therefore, be at work.

We have heard, anecdotally, that one VA facility reduced its use of left ventriculography by more than 80% within 8 years despite turnover in just a few interventional cardiologists. However, no published performance measures, guidelines, or appropriateness criteria exist for left ventriculography, so we can’t systematically assess where care is “best.” We did manage to document that even in cases where an echocardiogram was performed in the prior 30 days with no intervening admission for a cardiac cause that might prompt a change in LV ejection fraction, left ventriculography was still performed in 50% of patients. Although that seems like inappropriate use of the procedure, if the clinician does not trust the echocardiogram, he or she may feel compelled to do a left ventriculogram.

So I ask you: Does the variation in use of left ventriculography represent reasonable differences in opinion among clinicians? Or do some of us have unrealistic expectations about the benefits and harms of the procedure?

Categories: Cardiac Imaging
Tags: , ,


You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

5 Responses to “Left Ventriculography: Procedural Variation Writ Large”

  1. Gervasio Antonio Lamas, MD says:
    December 2, 2013 at 7:12 pm

    When I saw your paper, the first thought I had was to look at the age of the catheterizer. Those of us who trained in pre-good quality echo days put a lot of trust in the V-gram, and learned to properly opacify the ventricle. Younger catheterizers seem to do fewer V-grams.

    Did you have data on the age of the cath docs?

    GL

  2. Alain Efstratiou, MD says:
    December 3, 2013 at 9:20 am

    Echo often fails to reveal regional wall motion abnormalities. It notoriously misses the true apex. If renal function is not an issue, I believe an LV gram offers useful information.

  3. Paul Allen Heidenreich, MD, MS says:
    December 4, 2013 at 1:08 am

    Unfortunatley we don’t have age of the cath docs, though we are surveying them to see if time since completion of training is associated with LV Gram use. We are also asking about their views on the accuracy of echo as an alternative. If local echos are felt to be poor quality then an LV gram would make sense.

  4. John Carroll, MD says:
    December 5, 2013 at 5:36 pm

    Were there differences in the performance of LV gram based on diagnosis? In those with STEMI there is the frequent need to know systolic function (and certainly also LVEDP) and quality echoes may be hard to come by at night and weekends. One common scenario is in making the diagnosis of Takotsubo cardiomyopathy in the middle of the night.
    It would appear that no one factor will account for this variation in use. Contrary to the use of registry data alone, it might be most efficient to use operator interviews in a randomly selected cohort to determine factors that influence the decision to perform or not perform LV gram.

  5. Paul Allen Heidenreich, MD, MS says:
    December 5, 2013 at 6:58 pm

    Dr. Carroll is correct that no one factor explained the large variation in use. We are conducting electronic surveys of the cardiologists though interviews would likely provide additional insight.

    One thing I would ask is the following: Would you have interest in seeing how your use of the left ventriculogram compares to others in your community and acrosss the country?