November 26th, 2014

Resolving the Questions About Door-to-Balloon Times

CardioExchange’s John Ryan interviews Brahmajee K. Nallamothu about his research group’s retrospective study of changes, over time, in the relation between door-to-balloon (D2B) times and mortality after primary PCI. The study is published in The Lancet.

Ryan: Why is this study important? Haven’t we figured out D2B time already?

Nallamothu: This study was very important. D2B time is a highly visible performance metric for hospitals that conduct primary PCI. The D2B Alliance and Mission:Lifeline, two large national quality initiatives, have both targeted D2B time in their protocols (or its equivalent “first medical contact-to-balloon”), and the CMS considers D2B one of its core measures for acute MI. However, recent reports using population-level data, including a highly visible NEJM article from last year (Menees et al.), showed that contemporary decreases in D2B time do not correlate with improvements in mortality. According to some observers, this finding suggests that D2B time may not matter for individual patients. If true, that has big implications, as health systems have devoted enormous resources to tackling this problem.

However, Harlan Krumholz and I have been involved in this space for more than a decade now, and we found this interpretation very difficult to understand. It conflicted with prior literature and was inconsistent even with data presented within these reports themselves, which noted consistent individual-level relationships between D2B time and mortality. We felt that resolving this apparent paradox of “shorter D2B times but no change in mortality at the population-level” was important because misinterpretations could lead health systems to pull back from some of the dramatic gains in STEMI care made during the past decade. 

Ryan: Did you resolve the contradiction? Could you summarize your findings?

Nallamothu: We did resolve the contradiction at least partially. The key to understanding it was the ecological fallacy that population-level or “aggregate” relationships do not necessarily represent individual-level relationships. For example, the Menees et al. article examined the median D2B time for all the patients who underwent primary PCI in that year and then assigned that value to individuals, regardless of their own D2B time. In such a model, the aggregated median D2B time variable represents many secular factors that were simultaneously changing with year — not just D2B time. What we did was to tease out the individual-level relationship of D2B time with mortality after accounting for the population-level relationship. Our key insight was that the individual-level relationship between D2B time and mortality remained consistent over years, whereas the population-level relationship suggested increased mortality with the population undergoing primary PCI over years. This last finding explains why an overall “null” effect existed at the population-level despite shorter D2B times. The ecological fallacy is something that epidemiologists have long understood and accounted for in their evaluation of population-level data and aggregate relationships. We have not done as good a job in health services research, and this can be dangerous by leading to errors in statistical inference (see Finney et al.).

Ryan: Why is there a trend toward increased mortality with primary percutaneous PCI despite the decrease in D2B time? Why are the patients sicker and more complicated?

Nallamothu: We think it is important to put our findings in the context of what’s been happening in STEMI care during the past decade. Our best estimates suggest that the proportion of STEMI patients who are now treated with primary PCI rather than fibrinolytic therapy has doubled. In addition, the growth of STEMI systems of care are now bringing more STEMI patients to the cath lab who might have died in prior years, as treatment has sped up. Terkelsen and colleagues have called this the “survivor cohort” effect. Both of these facts make it challenging to evaluate the impact of changes at the population-level in D2B time, independent of other unrelated factors in STEMI care that were occurring. This is apparent in our data, where we noted that in a stable cohort of hospitals performing primary PCI, the number of patients treated with primary PCI grew by 55%, and several procedural factors changed over time. Bottom line: The patients who are going to the cath lab with STEMI in 2011 are not the same as those treated in 2006 — and this needs to be appreciated.

Ryan: There appear to be variable rates of usage of drug-eluting stents (DES) vs. bare-metal stents (BMS) in these patients (Table 1 of your article). Please explain this variability.

Nallamothu: This is a great example of what I discuss above: so many moving parts were happening simultaneously in STEMI care, even within this short time period, as both the patients and treatments changed. The variability in DES vs. BMS usage has much to do with the concerns raised about subacute stent thrombosis with DES in 2006, leading to an FDA warning and lots of press about their use relative to BMS. Much of this variability, I believe, is a shift in interventionalists’ views toward those technologies. We see the same patterns in use of antithrombotic medications and devices such as thrombectomy catheters.

Ryan: On a population level, what are the next targets worth pursuing to improve outcomes from STEMI and primary PCI?

Nallamothu: The Menees et al. article is an important study that started a valuable conversation among providers of STEMI care. I agree with them and others that we need to start looking toward new targets for STEMI care but that we also need to safeguard the gains we’ve made in D2B times. We have now established remarkable STEMI systems of care in Europe and North America that can bring more patients, more rapidly, to the right providers. We are also seeing tremendous advances in medical therapy and risk-factor management. I think the major goals of STEMI care continue to be (1) reducing errors in false-activation of STEMI systems to reduce the burden on providers; (2) reducing time from symptom onset to seeking treatment; (3) improving STEMI systems of care in underserved, rural and isolated communities; (4) finding the best mix of pharmacologic and mechanical reperfusion strategies; and (5) developing better options for late presenters who are at risk for reperfusion injury and shock that can lead to early mortality. Pedersen and colleagues recently published data from Denmark showing that patients who survive the first month after STEMI do remarkably well in follow-up (less than a 1.5% annual risk for death from cardiovascular causes). This is highly treatable disease!


How does Dr. Nallamothu’s study affect your understanding of the literature about door-to-balloon times?

One Response to “Resolving the Questions About Door-to-Balloon Times”

  1. What if stenting of acute MI on the net makes no difference? Is it possible that the current level of post MI care obviates the historic benefit seen or assumed with primary PCI?

    Also, is it possible that stenting small infarcts increases future risk even though stenting of large area infarcts may reduce future risk.

    In any event, it seems that our MEGA investment in providing rapid door to balloon times for all may be a mega waste. At the very least, it should be taken off of the list of quality measures for which hospitals are rewarded. The only thing worse than no quality metrics are useless quality metrics.