Portal de Periódicos da CAPES

Out‐of‐hospital Endotracheal Intubation: Are Observational Data Useful?

2010; Wiley; Volume: 17; Issue: 9 Linguagem: Inglês

10.1111/j.1553-2712.2010.00864.x

1553-2712

Henry E. Wang, Sandy Bogucki,

Emergency and Acute Care Studies

In this issue of Academic Emergency Medicine, Studnek et al.1 and Hanif et al.2 evaluate the association between paramedic endotracheal intubation (ETI) and patient outcomes after out-of-hospital cardiac arrest (OOHCAs). In the former piece, the authors reviewed 1,142 OOHCA treated in Mecklenberg County, North Carolina, during 2006–2008 and found improved adjusted outcomes for the patients in whom ETI was not attempted. In the latter work, the authors evaluated 1,294 OOHCAs treated in Los Angeles County, California, from 1994 to 2008 and found improved adjusted survival among patients only receiving bag-valve-mask ventilation, compared with those who were intubated by paramedics. These studies employed observational designs using administrative data sets. To gauge the importance of the reports, readers must recognize the strengths and weaknesses of observational studies. Observational studies using existing (or administrative) data are appealing because they provide efficient, immediate access to large numbers of subjects—often far larger than would be feasible for a prospective, clinical trial. These data sets are often large and heterogeneous, reflecting the natural setting of “real” popula-tions or communities and current clinical practice, thereby enhancing the potential generalizability of the findings. The utility of observational data may be limited by the available data or data elements. For example, the analysis by Studnek et al. associating ETI attempts with return of spontaneous circulation or hospital survival may not be feasible with data sets lacking airway process variables or patient outcomes. Administrative data sets often lack formal definitions of study variables. Even with standard definitions, inconsistent coding of the variables within or between studies could lead to biased or unexpected observations. For example, the difference in OOHCA survival between the two studies (10% vs. 4%) may reflect inherent population or response system variations or differences in identification and coding techniques. Confounders may also influence the inferences that can be drawn from observational data.3 Following prior analyses of paramedic ETI and traumatic brain injury, Studnek et al. and Hanif et al. used multivariable logistic regression to adjust for the influence of confounders such as age, sex, presenting rhythm, and whether the arrest was witnessed.4 However, unmeasured or unmeasurable confounders may limit the inferences of a regression analysis. In these two studies, uncaptured and uncoded clinical, environmental, operational, and educational factors may have led paramedics to selectively attempt ETI. Paramedics may have chosen not to intubate patients with rapid return of pulses or with intact gag reflexes. Is attempted ETI in these instances a surrogate marker for cardiac arrest prognosis? The study by Hanif et al. draws upon over 14 years of data; unmeasured secular trends in OOHCA survival and ETI timing and technique may have also confounded the results. Missing data often pose challenges. While excluding cases with missing covariates or outcomes is an option, this approach can lead to a reduction in sample size. While Studnek et al. handled missing outcomes data by assuming that these cases were nonsurvivors, a more comprehensive sensitivity analysis would have examined associations under a range of assumptions, including classifying missing cases as survivors. Multiple imputation is another viable, albeit challenging, strategy for handling missing data.5 A prospective trial is often the best way to circumvent the limitations of observational data. However, prospective clinical trials of paramedic ETI are methodologically and logistically difficult. The evaluation of pediatric ETI versus bag-valve-mask ventilation by Gausche et al.6 remains the only completed large prospective trial of paramedic ETI. Baseline adult OOHCA survival is on the order of 6% to 8%, indicating the need for over 10,000 subjects to detect a 1% absolute ETI survival benefit.7 The Resuscitation Outcomes Consortium is one of the only clinical out-of-hospital consortia currently able to carry out a trial of this magnitude.8,9 Often designed to address specific scientific questions in select subject populations, the characteristics of prospective trial populations may differ significantly from the general population. Furthermore, each incremental question may require large numbers of additional subjects—or a separate study. How should we view the Studek et al. and Hanif et al. papers? Even with their inherent limitations, they are well-designed and well-executed analyses that add to our current understanding of paramedic ETI. The papers originate from different settings, but arrived at similar inferences, supporting the robustness of their findings. These studies can and should be used as part of the evidentiary base that guides clinical practice. Critics may demand prospective controlled data, but prospective trials have formidable logistical and scientific barriers and may lead to inferences of limited utility. The findings of the Studnek et al. and Hanif et al. studies do deserve additional study. Experts must replicate the analyses using data from different care settings. Scientists and medical directors must work together to promulgate existing national airway management consensus definitions.10 The National EMS Information System (NEMSIS) is striving to achieve this type of EMS data standardization nationally.11 EMS clinicians must also work closely with scientists to improve the characterization of the paramedic ETI process, formulating variables and risk adjustment strategies that better account for the unmeasurable confounders that challenge current analyses. In addition, we may look more broadly beyond patient outcomes. Paramedic ETI is associated with adverse events such as endotracheal tube misplacement and dislodgement, airway trauma due to multiple ETI attempts, and iatrogenic oxygen desaturation and bradycardia.12–16 Paramedic ETI may interfere with other key resuscitation goals such as CPR continuity.17 Current methods for paramedic ETI training and educating may be inadequate, and recent studies hint that ETI experience may in fact bear associations with improved outcomes.18 Even without prospective controlled data, our understanding of paramedic ETI process and organization continues to grow and may help to inform evolving out-of-hospital airway management strategies. Supervising Editor: David C. Cone, MD.