Adverse Drug Event Nonrecognition in Emergency Departments
Adverse Drug Event Nonrecognition in Emergency Departments
This exploratory study was conducted in the medical ED of a French 3,000-bed tertiary care hospital with an annual ED census of 64,000 visits. The trauma, gynecologic, and psychiatric EDs are physically separated from the medical ED and were not included in this study.
At the time of the study, emergency medicine in France was a supra-specialty that was not recognized as a stand-alone specialty. At any time of day or week, every patient admitted to our medical ED was managed by a fellow physician who was supervised by a senior physician with a qualification in emergency medicine. Emergency physicians that intervened in our medical ED during the study were unaware of its specific objective, even if they were informed that a research project on ADEs in ED patients was being conducted there. The investigators reported any ADE that they could identify at the time of the care to the emergency physicians.
Institutional Review Board approval for noninterventional studies was obtained.
The selection process was designed as described previously. All adult patients presenting to the medical ED of the study hospital between January 2009 and December 2009 were eligible for enrollment. Of 261 weekdays during the study period, 85 were randomly selected, which allowed us to balance the number of time slots per weekday and per yearly quarter.
All patients who were physically present in the ED at the beginning of each time slot were screened for eligibility, regardless of entry date or illness severity. Patients were included if they (or their support person) agreed to participate, they did not visit the ED due to intentional drug poisoning, and it was their first visit to the ED during the study period. Readmissions were analyzed separately so as not to miss any ADE.
The investigators in this study were an emergency physician with special experience in internal medicine and a trained clinical pharmacist, neither intervened in the care of included patients. They are subsequently referred as "the investigator pair."
Data were collected by 12 pregraduate pharmacists (5th-year graduate students) completing a training course in the ED on weekday mornings during their university hospital internship. The students became familiar with the data-collection process during a standardized 1-week pilot period. They were trained by the clinical pharmacist to review all available ED charts (eg, clinician records, nursing notes, emergency medical services logs, and discharge instructions) and interview the patients or their relatives when possible. Information was prospectively collected in real time after patient inclusion, at the time of the care, under daily supervision.
Data were collected in a standardized abstraction form (Sphinx 5 software, Sphinx Développement, Chavanod, France). The data collected included sociodemographic characteristics, medical history, current clinical status, and final diagnosis. Special attention was focused on drug exposure during the 2 weeks preceding the ED visit. If data collected during the ED visit were insufficient to identify an ADE with certainty, additional information was obtained from other medical contact, but not from the patients themselves.
Drugs were classified by pharmacy students on the basis of the Anatomical Therapeutic Chemical Classification Index (World Health Organization Collaborating Centre for Drug Statistics Methodology). ED diagnosis and injuries associated with ADEs were coded by the investigator pair according to the International Classification of Diseases, 10th revision. The Charlson Comorbidity Index, one of the most extensively studied comorbidity index for predicting mortality, was used to assess the burden of concomitant disease for each patient.
ADEs were classified according to the definition used by the National Electronic Injury Surveillance System: Cooperative Adverse Drug Events Surveillance System (NEISS-CADES). The NEISS-CADES defines an ADE as an injury related to the patient's use of a drug and resulting from an allergic reaction (immunologically mediated effect), an adverse effect (undesirable pharmacologic effect at a recommended dose), an unintentional overdose (toxic effect linked to excess dose or impaired excretion), or a secondary effect (eg, falls and secondary infections). This definition excludes drug therapeutic failures, drug withdrawal, drug abuse, intentional drug poisonings, and ADEs that occurred as a result of ED medical treatment.
The investigator pair reviewed all cases together to identify ADEs. ADEs were identified on the basis of collected data using clinical knowledge and validated databases of known adverse drug reactions that were also accessible to the ED physicians. Contentious issues were resolved by consensus and, when required, by the expertise of a pharmacologist who was not involved in the research project. We studied ADEs that were either the cause for the patient's visit to the ED, or were unrelated to the patient's chief complaint.
ADE severity was assessed according to the five-stage Common Terminology Criteria for Adverse Events: A) spontaneous regression, B) regression after symptomatic treatment, C) hospitalization with no life threat, D) life-threatening risk, and E) death. Drug causality was assessed according to the four-stage Naranjo Probability Scale: "doubtful" (score 0), "possible" (score 1–4), "probable" (score 5–8), or "definite" (score ≥9). When two or more drugs were assessed in a clinical scenario suspicious for an ADE, the drug with the highest probability score was considered for the attribution of a global causality level to the ADE case.
Finally, the investigator pair examined whether or not the identified ADEs were recognized by the ED physicians.
The primary outcome was the frequency of ADEs that were attributed to a medication-related problem by emergency physicians. An ADE was considered "attributed" if any evidence of ADE suspicion, ADE diagnosis, or ADE management (eg, drug regimen optimization, drug discontinuation, prescription of a symptomatic treatment, hospitalization for further investigation) was documented on ED charts. An ADE was labeled as "unattributed" when no evidence was found that the emergency physician suspected, recognized, or provided management of the ADE; conversely, an ADE could be labeled as attributed even though it was not explicitly documented in the ED charts.
A Student's t-test was used to compare differences between the means for normally distributed variables, and a Wilcoxon rank sum test was used for the other variables. A comparison of groups for categorical variables was tested by a χ test or a Fisher's exact test. The level of significance was set at 5%.
To explore the factors associated with ADE nonrecognition, the following variables of interest were studied after verifying that no data were missing: age, sex, Charlson Comorbidity score, priority at the ED visit (triage acuity score), number of daily medications, ADE diagnosis, drug categories causing ADEs, relation between ADE and the patient's chief complaint, ADE causality assessment, and ADE severity. The link between ADE nonrecognition and covariables was first tested by univariate logistic models. Only the variables with p < 0.2 were retained for multivariate analysis. After assessing for collinearity, variables were selected in multivariate logistic models with a manual backward procedure including clinical judgment. Only clinically relevant interactions were tested. The model fit was tested (Hosmer-Lemeshow test). All analyses were performed using SAS statistical software (SAS 9.1, SAS Institute, Cary, NC).
Methods
Study Design and Setting
This exploratory study was conducted in the medical ED of a French 3,000-bed tertiary care hospital with an annual ED census of 64,000 visits. The trauma, gynecologic, and psychiatric EDs are physically separated from the medical ED and were not included in this study.
At the time of the study, emergency medicine in France was a supra-specialty that was not recognized as a stand-alone specialty. At any time of day or week, every patient admitted to our medical ED was managed by a fellow physician who was supervised by a senior physician with a qualification in emergency medicine. Emergency physicians that intervened in our medical ED during the study were unaware of its specific objective, even if they were informed that a research project on ADEs in ED patients was being conducted there. The investigators reported any ADE that they could identify at the time of the care to the emergency physicians.
Institutional Review Board approval for noninterventional studies was obtained.
Selection of Participants
The selection process was designed as described previously. All adult patients presenting to the medical ED of the study hospital between January 2009 and December 2009 were eligible for enrollment. Of 261 weekdays during the study period, 85 were randomly selected, which allowed us to balance the number of time slots per weekday and per yearly quarter.
All patients who were physically present in the ED at the beginning of each time slot were screened for eligibility, regardless of entry date or illness severity. Patients were included if they (or their support person) agreed to participate, they did not visit the ED due to intentional drug poisoning, and it was their first visit to the ED during the study period. Readmissions were analyzed separately so as not to miss any ADE.
Data Collection and Processing
The investigators in this study were an emergency physician with special experience in internal medicine and a trained clinical pharmacist, neither intervened in the care of included patients. They are subsequently referred as "the investigator pair."
Data were collected by 12 pregraduate pharmacists (5th-year graduate students) completing a training course in the ED on weekday mornings during their university hospital internship. The students became familiar with the data-collection process during a standardized 1-week pilot period. They were trained by the clinical pharmacist to review all available ED charts (eg, clinician records, nursing notes, emergency medical services logs, and discharge instructions) and interview the patients or their relatives when possible. Information was prospectively collected in real time after patient inclusion, at the time of the care, under daily supervision.
Data were collected in a standardized abstraction form (Sphinx 5 software, Sphinx Développement, Chavanod, France). The data collected included sociodemographic characteristics, medical history, current clinical status, and final diagnosis. Special attention was focused on drug exposure during the 2 weeks preceding the ED visit. If data collected during the ED visit were insufficient to identify an ADE with certainty, additional information was obtained from other medical contact, but not from the patients themselves.
Drugs were classified by pharmacy students on the basis of the Anatomical Therapeutic Chemical Classification Index (World Health Organization Collaborating Centre for Drug Statistics Methodology). ED diagnosis and injuries associated with ADEs were coded by the investigator pair according to the International Classification of Diseases, 10th revision. The Charlson Comorbidity Index, one of the most extensively studied comorbidity index for predicting mortality, was used to assess the burden of concomitant disease for each patient.
Methods of Measurement
ADEs were classified according to the definition used by the National Electronic Injury Surveillance System: Cooperative Adverse Drug Events Surveillance System (NEISS-CADES). The NEISS-CADES defines an ADE as an injury related to the patient's use of a drug and resulting from an allergic reaction (immunologically mediated effect), an adverse effect (undesirable pharmacologic effect at a recommended dose), an unintentional overdose (toxic effect linked to excess dose or impaired excretion), or a secondary effect (eg, falls and secondary infections). This definition excludes drug therapeutic failures, drug withdrawal, drug abuse, intentional drug poisonings, and ADEs that occurred as a result of ED medical treatment.
The investigator pair reviewed all cases together to identify ADEs. ADEs were identified on the basis of collected data using clinical knowledge and validated databases of known adverse drug reactions that were also accessible to the ED physicians. Contentious issues were resolved by consensus and, when required, by the expertise of a pharmacologist who was not involved in the research project. We studied ADEs that were either the cause for the patient's visit to the ED, or were unrelated to the patient's chief complaint.
ADE severity was assessed according to the five-stage Common Terminology Criteria for Adverse Events: A) spontaneous regression, B) regression after symptomatic treatment, C) hospitalization with no life threat, D) life-threatening risk, and E) death. Drug causality was assessed according to the four-stage Naranjo Probability Scale: "doubtful" (score 0), "possible" (score 1–4), "probable" (score 5–8), or "definite" (score ≥9). When two or more drugs were assessed in a clinical scenario suspicious for an ADE, the drug with the highest probability score was considered for the attribution of a global causality level to the ADE case.
Finally, the investigator pair examined whether or not the identified ADEs were recognized by the ED physicians.
Outcome Measures
The primary outcome was the frequency of ADEs that were attributed to a medication-related problem by emergency physicians. An ADE was considered "attributed" if any evidence of ADE suspicion, ADE diagnosis, or ADE management (eg, drug regimen optimization, drug discontinuation, prescription of a symptomatic treatment, hospitalization for further investigation) was documented on ED charts. An ADE was labeled as "unattributed" when no evidence was found that the emergency physician suspected, recognized, or provided management of the ADE; conversely, an ADE could be labeled as attributed even though it was not explicitly documented in the ED charts.
Primary Data Analysis
A Student's t-test was used to compare differences between the means for normally distributed variables, and a Wilcoxon rank sum test was used for the other variables. A comparison of groups for categorical variables was tested by a χ test or a Fisher's exact test. The level of significance was set at 5%.
To explore the factors associated with ADE nonrecognition, the following variables of interest were studied after verifying that no data were missing: age, sex, Charlson Comorbidity score, priority at the ED visit (triage acuity score), number of daily medications, ADE diagnosis, drug categories causing ADEs, relation between ADE and the patient's chief complaint, ADE causality assessment, and ADE severity. The link between ADE nonrecognition and covariables was first tested by univariate logistic models. Only the variables with p < 0.2 were retained for multivariate analysis. After assessing for collinearity, variables were selected in multivariate logistic models with a manual backward procedure including clinical judgment. Only clinically relevant interactions were tested. The model fit was tested (Hosmer-Lemeshow test). All analyses were performed using SAS statistical software (SAS 9.1, SAS Institute, Cary, NC).
Source...