BACKGROUND: Despite advances in healthcare technology and resources, patient survival after cardiac arrest is still very low. Patients who do initially survive cardiac arrest are still at high risk from potentially fatal neurological or other organ injuries (Laver et al., 2004). After initial treatment, clinical care of unresponsive cardiac arrest patients relies heavily on neuroprognostication to assess the extent of brain injury and help form outcome predictions. If neuroprognosis evaluates a patient outcome to be very poor, withdrawal of life-sustaining therapy is likely to occur. These factors make the improvement of neuroprognostication an important challenge in the healthcare industry.
METHODS: Our retrospective study constructed and analyzed qualitative and quantitative models of patient neurological outcomes based on diffusion-weighted images (DWI). The patient cohorts included 204 patient post-cardiac arrest comatose targeted temperature management (TTM) patients, as well as a 48-patient non-cardiac arrest control group. Magnetic resonance imaging (MRI) reports were qualitatively assessed for the presence of anoxic injury, while MRI images were quantitatively assessed via their apparent diffusion coefficients (ADC). Models were then constructed based on this data using logistic regression.
RESULTS: Our results indicated that ADC values differed most in cortical regions and that cortical regions quantitatively predicted good patient neurological outcomes superior to all other regions, with only the occipital lobe model having significantly higher predictive value than the qualitative models. We also saw that qualitative models predicted good neurological outcomes with a higher specificity but lower sensitivity than quantitative models. The qualitative and quantitative models showed substantial agreement in most regions, with the qualitative model having a higher positive and lower negative predictive value for good outcomes compared to the quantitative models.
DISCUSSION: We saw a large concentration of anoxic injury in cortical regions, with these regions predicting good patient outcomes better than all others. Generally, the qualitative and quantitative models had the same predictive value across all regions except the occipital lobe. The difference between qualitative and quantitative models seems to provide evidence of a self-fulfilling prophecy in post-cardiac arrest care. Further directions for this project include modeling with the addition of other prognostic tests to form a more valuable predictive model.
| Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/48396 |
| Date | 14 March 2024 |
| Creators | Van Roy, Sam |
| Contributors | McKnight, C. James |
| Source Sets | Boston University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
Page generated in 0.0024 seconds