Electronic health record surveillance algorithms facilitate the detection of transfusion‐related pulmonary complications

Abstract

BACKGROUND Transfusion-related acute lung injury (TRALI) and transfusion-associated circulatory overload (TACO) are leading causes of transfusion-related mortality. Notably, poor syndrome recognition and underreporting likely result in an underestimate of their true attributable burden. We aimed to develop accurate electronic health record–based screening algorithms for improved detection of TRALI/transfused acute lung injury (ALI) and TACO. STUDY DESIGN AND METHODS This was a retrospective observational study. The study cohort, identified from a previous National Institutes of Health–sponsored prospective investigation, included 223 transfused patients with TRALI, transfused ALI, TACO, or complication-free controls. Optimal case detection algorithms were identified using classification and regression tree (CART) analyses. Algorithm performance was evaluated with sensitivities, specificities, likelihood ratios, and overall misclassification rates. RESULTS For TRALI/transfused ALI detection, CART analysis achieved a sensitivity and specificity of 83.9% (95% confidence interval [CI], 74.4%–90.4%) and 89.7% (95% CI, 80.3%–95.2%), respectively. For TACO, the sensitivity and specificity were 86.5% (95% CI, 73.6%–94.0%) and 92.3% (95% CI, 83.4%–96.8%), respectively. Reduced PaO2/FiO2 ratios and the acquisition of posttransfusion chest radiographs were the primary determinants of case versus control status for both syndromes. Of true-positive cases identified using the screening algorithms (TRALI/transfused ALI, n = 78; TACO, n = 45), only 11 (14.1%) and five (11.1%) were reported to the blood bank by physicians, respectively. CONCLUSIONS Electronic screening algorithms have shown good sensitivity and specificity for identifying patients with TRALI/transfused ALI and TACO at our institution. This supports the notion that active electronic surveillance may improve case identification, thereby providing a more accurate understanding of TRALI/transfused ALI and TACO epidemiology.

Cytokines and clinical predictors in distinguishing pulmonary transfusion reactions

“…10,28,29 Given the advent of electronic medical record surveillance and focus on adverse clinical outcomes related to transfusion, we can expect increased identification of complex cases of post-transfusion pulmonary edema. 30–32 While providing some guidance in their identification, definitions of pulmonary transfusion reactions require more specific criteria to help differentiate complicated clinical cases.…”

Characterizing the Epidemiology of Postoperative Transfusion-related Acute Lung Injury

“…1 Perhaps more concerning is the fact that passive reporting has repeatedly been shown to significantly underestimate the true burden of this syndrome. 2,3 For example, Kopko et al . 2 noted that only 13.3% of cases with symptoms suggestive of TRALI were reported to the blood bank by the clinical service as a possible transfusion reaction.…”

Characterizing the Epidemiology of Perioperative Transfusion-Associated Circulatory Overload

“…Our primary reasons for selecting this population are that a large proportion of blood product transfusions take place in the operating room environment 18 and that a substantial proportion of those who experience TACO will do so after intraoperative transfusion (70% of all TACO episodes). 7 Notwithstanding the findings of a few studies, 2,16 most investigations to date have focused on the epidemiology of TACO in the critically ill. 3,5 How well these results apply to most patients encountered in the operating room is unclear. Historically, incidence rates have ranged from less than 1 to 11%.…”

“…Structured data points that were collected included partial pressure of arterial oxygen (Pa o 2 ), Pa o 2 to fraction of inspired oxygen ratio (Pa o 2 :F io 2 ), oxygen saturations (Sp o 2 ), and respiratory rate (Supplemental Digital Content 1, http://links.lww.com/ALN/B112: TACO screening algorithm). 7,14 To ensure completeness, we collected these data from three databases: (1) the perioperative data mart (described above), (2) the ICU data mart 9 -a near real-time database that captures data directly from the EMR for patients in the ICU, 9 and (3) the Mayo Clinic Life Sciences System 11,12 -a repository of replicated data from the EMR of patients admitted to the general surgical floor postoperatively. The sensitivity of our algorithm was maximized by identifying the worst or most extreme values for the patients up to 6 h after blood product transfusion.…”