Sanghavi P, Jena AB, Newhouse JP, et al. Outcomes after out-of-hospital cardiac arrest treated by basic vs. advanced life support. JAMA Intern Med. Nov. 24, 2014. [Epub ahead of print.]
Nothing is more controversial and polarizing in EMS than arbitrarily separating what’s considered a “basic” skill and an “advanced” skill. Historically, our profession has been bogged down in divisive battles to categorize “invasive” or “specific” skills associated with different scopes of practice.
The research we’re reviewing this month has gained a lot of publicity for its title: “BLS vs. ALS,” as well as the prestige of the journal and authors’ institutional affiliations. Let’s be careful to look beyond the headline, and consider the science.
Methods: In this study, Harvard University researchers retrospectively reviewed a sample of U.S. cardiac arrests among Medicare beneficiaries from 2009–2011. Using ICD-9 diagnosis codes, the authors tried to verify these patients arrived at the hospital in cardiac arrest. Patients from certain states, rural counties and any traumatic arrests were omitted to prevent confounding data. Therefore, urban and suburban patients were grouped into categories based on an ALS or BLS billing charge by EMS.
Primary outcomes were measured as survival to hospital discharge, survival to 30 days, and survival to 90 days. The secondary outcome measured was neurological performance.
Results: After making statistical adjustments for the difference in the size of the two comparison groups (31,292 billed ALS vs. 1,643 billed BLS), the authors report a 4% higher rate of survival to hospital discharge in the BLS group (95% confidence interval (CI) 1.9–4.8), as well as a 3.4% higher rate of survival to 30 days (95% CI 1.2–4.0) and a 2.6% higher rate of survival at 90 days (95% CI 1.2–4.0), when compared to the ALS group. They also reported a 3.5% decrease in patients determined to have poor neurological functioning in the BLS group compared to the ALS group (95% CI 2.2–4.8).
Lastly, they found that mean medical spending was higher for the BLS group than the ALS group, which they attributed to longer survival of those in the BLS group. The authors went on to claim their findings “suggest that 1,479 (95% CI 683–2,276) additional Medicare beneficiaries who experience cardiac arrest would survive to 90 days if provided BLS instead of ALS.”
Limitations: First and foremost, this was a retrospective study. While the researchers went to great lengths to utilize robust analytical techniques to control for the influence of extraneous variables, no statistical analysis can establish a cause and effect relationship using retrospective data. Any conclusions drawn from these results must be tempered with regard to the lack of internal validity inherent to this type of study design.
Next, there’s a notable inconsistency in the description of the sampling technique used. The authors state in their methods section that they “analyzed a 20% simple random sample of fee-for-service Medicare beneficiaries from non-rural counties who experienced out-of-hospital cardiac arrest.” However, a figure depicting their sample construction demonstrates that they actually began with 4,081,432 records of Medicare claims involving ambulance transport.
These claims don’t appear specific to those involving out-of-hospital cardiac arrest. Rather, the authors subsequently applied a series of inclusion and exclusion criteria to reduce their original sample to a final sample of 32,935 ambulance rides, meeting their desired criteria (e.g. non-traumatic, non-rural cases). This doesn’t represent a random sampling technique as claimed by the authors. Rather, it’s more akin to a data-mining approach, thus further weakening their results.
While the authors contend that the patients included in their final sample experienced an out-of-hospital cardiac arrest, their use of the ICD-9 code for cardiac arrest, combined with a “present on admission” indicator, presents a considerable methodological flaw.
According to the ICD-9-CM Official Guidelines for Coding and Reporting, “conditions that develop during an outpatient encounter … are considered as being present on admission.” The “time the admission order was written” is the cutoff point for determining if a condition is present on admission. As such, an unknown number of the cases included in this analysis may have been alive for their entire encounter with EMS and then experienced cardiac arrest later in the ED.
These issues aside, there are several additional limitations to consider. This study looks only at Medicare patients. While they attempted to control for the effects of patient age within their sample, it doesn’t account for the fact that the sample contains an age-related bias. Thiscompromises the external validity and the ability to generalize their findings to other subsets of the population that experience out-of-hospital cardiac arrest.
The researchers also made a stretch in their use of billing records to represent the patient care that was provided. Because they relied on generic billing categories (ALS vs. BLS), there’s no way to distinguish between the exact interventions performed. There’s also no guarantee the level of service billed for accurately represents the level of care or specific interventions provided. Calls billed as ALS may have been treated by BLS first responders for an unknown period of time prior to the arrival of ALS. They further rely on the “likelihood” that ALS providers would bill for ALS, even if they only delivered BLS interventions, based on a financial motive. Unless they can support such a claim with evidence, then it’s little more than an ethical allegation against the practices of our profession and doesn’t contribute to an examination of the issue being studied.
It’s worth noting that patients billed for ALS were more often picked up at a residence, whereas patients billed for BLS were most commonly picked up at skilled nursing facilities. In the event these were cases of out-of-hospital cardiac arrest, BLS patients may have had a quicker recognition and AED response time due to their location in a healthcare facility.
Conclusions: No matter what patch you wear, today’s state-of-the-art evidence repeatedly points to survival being tied to coordination of care, such as “pit-crew CPR.” This is a complex skill, requiring advanced training and repeated practice, regardless of the bundle of care being used or the diagnosis code used for billing.
Visit www.pcrfpodcast.org for audio commentary.
What we already know: Previous evidence shows well-coordinated advanced care leads to neurologically intact survival of out-of-hospital cardiac arrest.
What this study adds: This study confirms that retrospective studies with flawed methodology can capture publicity based on catchy headlines. Biased scientific publication creates confusing debate and non-actionable results.