Vincent Liu - Academia.edu (original) (raw)
Papers by Vincent Liu
BMC health services research, Jan 10, 2014
Randomized controlled trial evidence supports a restrictive strategy of red blood cell (RBC) tran... more Randomized controlled trial evidence supports a restrictive strategy of red blood cell (RBC) transfusion, but significant variation in clinical transfusion practice persists. Patient characteristics other than hemoglobin levels may influence the decision to transfuse RBCs and explain some of this variation. Our objective was to evaluate the role of patient comorbidities and severity of illness in predicting inpatient red blood cell transfusion events. We developed a predictive model of inpatient RBC transfusion using comprehensive electronic medical record (EMR) data from 21 hospitals over a four year period (2008-2011). Using a retrospective cohort study design, we modeled predictors of transfusion events within 24 hours of hospital admission and throughout the entire hospitalization. Model predictors included administrative data (age, sex, comorbid conditions, admission type, and admission diagnosis), admission hemoglobin, severity of illness, prior inpatient RBC transfusion, admi...
Transfusion, 2014
Blood conservation strategies have been shown to be effective in decreasing red blood cell (RBC) ... more Blood conservation strategies have been shown to be effective in decreasing red blood cell (RBC) utilization in specific patient groups. However, few data exist describing the extent of RBC transfusion reduction or their impact on transfusion practice and mortality in a diverse inpatient population. We conducted a retrospective cohort study using comprehensive electronic medical record data from 21 medical facilities in Kaiser Permanente Northern California. We examined unadjusted and risk-adjusted RBC transfusion and 30-day mortality coincident with implementation of RBC conservation strategies. The inpatient study cohort included 391,958 patients who experienced 685,753 hospitalizations. From 2009 to 2013, the incidence of RBC transfusion decreased from 14.0% to 10.8% of hospitalizations; this change coincided with a decline in pretransfusion hemoglobin (Hb) levels from 8.1 to 7.6 g/dL. Decreased RBC utilization affected broad groups of admission diagnoses and was most pronounced ...
B24. PROCESS IMPROVEMENT: IMPACT ON OUTCOMES OF PATIENTS WITH PNEUMONIA, 2012
JAMA Internal Medicine, 2014
Medical Care, 2010
Length of stay (LOS) is a common measure of hospital resource utilization. Most methods for risk-... more Length of stay (LOS) is a common measure of hospital resource utilization. Most methods for risk-adjusting LOS are limited by the use of only administrative data. Recent studies suggest that adding automated clinical data to these models improves performance. To evaluate the utility of adding "point of admission" automated laboratory and comorbidity measures-the Laboratory Acute Physiology Score (LAPS) and Comorbidity Point Score (COPS)-to risk adjustment models that are based on administrative data. We performed a retrospective analysis of 155,474 hospitalizations between 2002 and 2005 at 17 Northern California Kaiser Permanente hospitals. We evaluated the benefit of adding LAPS and COPS in linear regression models using full, trimmed, truncated, and log-transformed LOS, as well as in logistic and generalized linear models. Mean age was 61 +/- 19 years; females represented 55.2% of subjects. The mean LOS was 4.5 +/- 7.7 days; median LOS was 2.8 days (interquartile range, 1.3-5.1 days). Adding LAPS and COPS to the linear regression model improved R by 29% from 0.113 to 0.146. Similar improvements with the inclusion of LAPS and COPS were observed in other regression models. Together, these variables were responsible for >50% of the predictive ability of logistic regression models that identified outliers with longer LOS. The inclusion of automated laboratory and comorbidity data improved LOS predictions in all models, underscoring the need for more widespread adoption of comprehensive electronic medical records.
BMC Medical Informatics and Decision Making, 2013
Background: Prior studies demonstrate the suitability of natural language processing (NLP) for id... more Background: Prior studies demonstrate the suitability of natural language processing (NLP) for identifying pneumonia in chest radiograph (CXR) reports, however, few evaluate this approach in intensive care unit (ICU) patients. Methods: From a total of 194,615 ICU reports, we empirically developed a lexicon to categorize pneumonia-relevant terms and uncertainty profiles. We encoded lexicon items into unique queries within an NLP software application and designed an algorithm to assign automated interpretations ('positive', 'possible', or 'negative') based on each report's query profile. We evaluated algorithm performance in a sample of 2,466 CXR reports interpreted by physician consensus and in two ICU patient subgroups including those admitted for pneumonia and for rheumatologic/ endocrine diagnoses. Results: Most reports were deemed 'negative' (51.8%) by physician consensus. Many were 'possible' (41.7%); only 6.5% were 'positive' for pneumonia. The lexicon included 105 terms and uncertainty profiles that were encoded into 31 NLP queries. Queries identified 534,322 'hits' in the full sample, with 2.7 ± 2.6 'hits' per report. An algorithm, comprised of twenty rules and probability steps, assigned interpretations to reports based on query profiles. In the validation set, the algorithm had 92.7% sensitivity, 91.1% specificity, 93.3% positive predictive value, and 90.3% negative predictive value for differentiating 'negative' from 'positive'/'possible' reports. In the ICU subgroups, the algorithm also demonstrated good performance, misclassifying few reports (5.8%). Conclusions: Many CXR reports in ICU patients demonstrate frank uncertainty regarding a pneumonia diagnosis. This electronic tool demonstrates promise for assigning automated interpretations to CXR reports by leveraging both terms and uncertainty profiles.
Annals of the American Thoracic Society, Jan 23, 2015
Sepsis contributes to 1 in every 2 to 3 inpatient hospital deaths. Early recognition and treatmen... more Sepsis contributes to 1 in every 2 to 3 inpatient hospital deaths. Early recognition and treatment are instrumental in reducing mortality; yet there are substantial quality gaps. Sepsis bundles containing quality metrics are often used in efforts to improve outcomes. Several prominent organizations have published their own bundles, but there are few head-to-head comparisons of content. We reviewed the components of and level of agreement among several sepsis resuscitation and management bundles. We compared the individual bundle elements, along with their associated goals and metrics. We performed a systematic review (PubMed 2008 - 2015) and searched publically available online content, supplemented by interviews with key informants, to identify 8 distinct bundles. Bundles are presented as current as of April 2015. Broadly, elements of care covered early resuscitation and short-term management. Bundles varied from 6 to 10 elements, and there were 12 distinct elements listed across a...
BMC health services research, Jan 10, 2014
Randomized controlled trial evidence supports a restrictive strategy of red blood cell (RBC) tran... more Randomized controlled trial evidence supports a restrictive strategy of red blood cell (RBC) transfusion, but significant variation in clinical transfusion practice persists. Patient characteristics other than hemoglobin levels may influence the decision to transfuse RBCs and explain some of this variation. Our objective was to evaluate the role of patient comorbidities and severity of illness in predicting inpatient red blood cell transfusion events. We developed a predictive model of inpatient RBC transfusion using comprehensive electronic medical record (EMR) data from 21 hospitals over a four year period (2008-2011). Using a retrospective cohort study design, we modeled predictors of transfusion events within 24 hours of hospital admission and throughout the entire hospitalization. Model predictors included administrative data (age, sex, comorbid conditions, admission type, and admission diagnosis), admission hemoglobin, severity of illness, prior inpatient RBC transfusion, admi...
Transfusion, 2014
Blood conservation strategies have been shown to be effective in decreasing red blood cell (RBC) ... more Blood conservation strategies have been shown to be effective in decreasing red blood cell (RBC) utilization in specific patient groups. However, few data exist describing the extent of RBC transfusion reduction or their impact on transfusion practice and mortality in a diverse inpatient population. We conducted a retrospective cohort study using comprehensive electronic medical record data from 21 medical facilities in Kaiser Permanente Northern California. We examined unadjusted and risk-adjusted RBC transfusion and 30-day mortality coincident with implementation of RBC conservation strategies. The inpatient study cohort included 391,958 patients who experienced 685,753 hospitalizations. From 2009 to 2013, the incidence of RBC transfusion decreased from 14.0% to 10.8% of hospitalizations; this change coincided with a decline in pretransfusion hemoglobin (Hb) levels from 8.1 to 7.6 g/dL. Decreased RBC utilization affected broad groups of admission diagnoses and was most pronounced ...
B24. PROCESS IMPROVEMENT: IMPACT ON OUTCOMES OF PATIENTS WITH PNEUMONIA, 2012
JAMA Internal Medicine, 2014
Medical Care, 2010
Length of stay (LOS) is a common measure of hospital resource utilization. Most methods for risk-... more Length of stay (LOS) is a common measure of hospital resource utilization. Most methods for risk-adjusting LOS are limited by the use of only administrative data. Recent studies suggest that adding automated clinical data to these models improves performance. To evaluate the utility of adding "point of admission" automated laboratory and comorbidity measures-the Laboratory Acute Physiology Score (LAPS) and Comorbidity Point Score (COPS)-to risk adjustment models that are based on administrative data. We performed a retrospective analysis of 155,474 hospitalizations between 2002 and 2005 at 17 Northern California Kaiser Permanente hospitals. We evaluated the benefit of adding LAPS and COPS in linear regression models using full, trimmed, truncated, and log-transformed LOS, as well as in logistic and generalized linear models. Mean age was 61 +/- 19 years; females represented 55.2% of subjects. The mean LOS was 4.5 +/- 7.7 days; median LOS was 2.8 days (interquartile range, 1.3-5.1 days). Adding LAPS and COPS to the linear regression model improved R by 29% from 0.113 to 0.146. Similar improvements with the inclusion of LAPS and COPS were observed in other regression models. Together, these variables were responsible for >50% of the predictive ability of logistic regression models that identified outliers with longer LOS. The inclusion of automated laboratory and comorbidity data improved LOS predictions in all models, underscoring the need for more widespread adoption of comprehensive electronic medical records.
BMC Medical Informatics and Decision Making, 2013
Background: Prior studies demonstrate the suitability of natural language processing (NLP) for id... more Background: Prior studies demonstrate the suitability of natural language processing (NLP) for identifying pneumonia in chest radiograph (CXR) reports, however, few evaluate this approach in intensive care unit (ICU) patients. Methods: From a total of 194,615 ICU reports, we empirically developed a lexicon to categorize pneumonia-relevant terms and uncertainty profiles. We encoded lexicon items into unique queries within an NLP software application and designed an algorithm to assign automated interpretations ('positive', 'possible', or 'negative') based on each report's query profile. We evaluated algorithm performance in a sample of 2,466 CXR reports interpreted by physician consensus and in two ICU patient subgroups including those admitted for pneumonia and for rheumatologic/ endocrine diagnoses. Results: Most reports were deemed 'negative' (51.8%) by physician consensus. Many were 'possible' (41.7%); only 6.5% were 'positive' for pneumonia. The lexicon included 105 terms and uncertainty profiles that were encoded into 31 NLP queries. Queries identified 534,322 'hits' in the full sample, with 2.7 ± 2.6 'hits' per report. An algorithm, comprised of twenty rules and probability steps, assigned interpretations to reports based on query profiles. In the validation set, the algorithm had 92.7% sensitivity, 91.1% specificity, 93.3% positive predictive value, and 90.3% negative predictive value for differentiating 'negative' from 'positive'/'possible' reports. In the ICU subgroups, the algorithm also demonstrated good performance, misclassifying few reports (5.8%). Conclusions: Many CXR reports in ICU patients demonstrate frank uncertainty regarding a pneumonia diagnosis. This electronic tool demonstrates promise for assigning automated interpretations to CXR reports by leveraging both terms and uncertainty profiles.
Annals of the American Thoracic Society, Jan 23, 2015
Sepsis contributes to 1 in every 2 to 3 inpatient hospital deaths. Early recognition and treatmen... more Sepsis contributes to 1 in every 2 to 3 inpatient hospital deaths. Early recognition and treatment are instrumental in reducing mortality; yet there are substantial quality gaps. Sepsis bundles containing quality metrics are often used in efforts to improve outcomes. Several prominent organizations have published their own bundles, but there are few head-to-head comparisons of content. We reviewed the components of and level of agreement among several sepsis resuscitation and management bundles. We compared the individual bundle elements, along with their associated goals and metrics. We performed a systematic review (PubMed 2008 - 2015) and searched publically available online content, supplemented by interviews with key informants, to identify 8 distinct bundles. Bundles are presented as current as of April 2015. Broadly, elements of care covered early resuscitation and short-term management. Bundles varied from 6 to 10 elements, and there were 12 distinct elements listed across a...