Computerized detection of adverse drug reactions in the medical intensive care unit

Interventions, Barriers, and Proposed Solutions Associated With the Implementation of a Protocol That Uses Clinical Decision Support and a Stress Biomarker Test to Identify ICU Patients at High-Risk for Drug Associated Acute Kidney Injury

BACKGROUND

Damage biomarkers are helpful in early identification of patients who are at risk of developing acute kidney injury (AKI). Investigations are ongoing to identify the optimal role of stress/damage biomarkers in clinical practice regarding AKI risk prediction, surveillance, diagnosis, and prognosis.

OBJECTIVE

To determine the impact of utilizing a clinical decision support system (CDSS) to guide stress biomarker testing in intensive care unit (ICU) patients at risk for drug-induced acute kidney injury (D-AKI).

METHODS

A protocol was designed utilizing a clinical decision support system (CDSS) alert to identify patients that were ordered 3 or more potentially nephrotoxic medications, suggesting risk for progressing to AKI from nephrotoxic burden. Once alerted to these high-risk patients, the pharmacist determined if action was needed by ordering a stress biomarker test, tissue inhibitor of metalloproteinase-2-insulin-like growth factor-binding protein 7 (TIMP-2•IGFBP7). If the biomarker test result was elevated, the pharmacist provided nephrotoxin stewardship recommendations to the team. Pharmacists recorded the response to the clinical decision support alert, ordering, and interpreting the TIMP-2•IGFBP7, and information regarding clinical interventions. An alert in conjunction with TIMP-2•IGFBP7 as a strategy for AKI risk prediction and stimulant for patient care management was assessed. In addition, barriers and solutions to protocol implementation were evaluated.

RESULTS

There were 394 total activities recorded by pharmacists for 345 unique patients. Ninety-three (93/394; 23.6%) actionable alerts resulted in a TIMP-2•IGFBP7 test being ordered. Thirty-one TIMP-2•IGFBP7 results were >0.3 (31/81; 38.3%), suggesting a high-risk of progression to AKI, which prompted 191 pharmacist/team interventions. On average, there were 1.64 interventions per patient in the low-risk patients, 3.43 in high-risk patients, and 3.75 in the highest-risk patients.

CONCLUSION AND RELEVANCE

Stress biomarkers can be used in conjunction with CDSS alerts to affect therapeutic decisions in ICU patients at high-risk for D-AKI.

Comprehensive analysis of rule formalisms to represent clinical guidelines: Selection criteria and case study on antibiotic clinical guidelines

BACKGROUND

The over-use of antibiotics in clinical domains is causing an alarming increase in bacterial resistance, thus endangering their effectiveness as regards the treatment of highly recurring severe infectious diseases. Whilst Clinical Guidelines (CGs) focus on the correct prescription of antibiotics in a narrative form, Clinical Decision Support Systems (CDSS) operationalize the knowledge contained in CGs in the form of rules at the point of care. Despite the efforts made to computerize CGs, there is still a gap between CGs and the myriad of rule technologies (based on different logic formalisms) that are available to implement CDSSs in real clinical settings.

OBJECTIVE

To helpCDSS designers to determine the most suitable rule-based technology (medical-oriented rules, production rules and semantic web rules) with which to model knowledge from CGs for the prescription of antibiotics. We propose a framework of criteria for this purpose that is extensible to more generic CGs.

MATERIALS AND METHODS

Our proposal is based on the identification of core technical requirements extracted from both literature and the analysis of CGs for antibiotics, establishing three dimensions for analysis: language expressivity, interoperability and industrial aspects. We present a case study regarding the John Hopkins Hospital (JHH) Antibiotic Guidelines for Urinary Tract Infection (UTI), a highly recurring hospital acquired infection. We have adopted our framework of criteria in order to analyse and implement these CGs using various rule technologies: HL7 Arden Syntax, general-purpose Production Rules System (Drools), HL7 standard Rule Interchange Format (RIF), Semantic Web Rule Language (SWRL) and SParql Inference Notation (SPIN) rule extensions (implementing our own ontology for UTI).

RESULTS

We have identified the main criteria required to attain a maintainable and cost-affordable computable knowledge representation for CGs. We have represented the JHH UTI CGs knowledge in a total of 12 Arden Syntax MLMs, 81 Drools rules and 154 ontology classes, properties and individuals. Our experiments confirm the relevance of the proposed set of criteria and show the level of compliance of the different rule technologies with the JHH UTI CGs knowledge representation.

CONCLUSIONS

The proposed framework of criteria may help clinical institutions to select the most suitable rule technology for the representation of CGs in general, and for the antibiotic prescription domain in particular, depicting the main aspects that lead to Computer Interpretable Guidelines (CIGs), such as Logic expressivity (Open/Closed World Assumption, Negation-As-Failure), Temporal Reasoning and Interoperability with existing HIS and clinical workflow. Future work will focus on providing clinicians with suggestions regarding new potential steps for CGs, considering process mining approaches and CGs Process Workflows, the use of HL7 FHIR for HIS interoperability and the representation of Knowledge-as- a-Service (KaaS).

Outcomes and Adverse Effects With Peramivir for the Treatment of Influenza H1N1 in Critically Ill Pediatric Patients

OBJECTIVES

Influenza is an environmental pathogen and infection presents as a range from asymptomatic to fulminant illness. Though treatment is supportive, antiviral agents have a role in the management of infection. Pediatric use of peramivir is largely based on reports and extrapolations of pharmacokinetic data. We seek to describe efficacy and safety of peramivir in critically ill pediatric patients.

METHODS

This is a retrospective, institutional review board-approved chart review of all patients under 21 years of age, admitted to the PICU, and treated with peramivir for influenza H1N1 infection between January 1, 2016, and March 31, 2016, at a single-center, 12-bed PICU. The primary outcome was time to sustained resolution of fever; secondary outcomes included dose, duration, and adverse effects of peramivir therapy.

RESULTS

Seven patients were included with median age of 3.7 years. Median time to sustained resolution of fever was 49.3 hours, median duration of mechanical ventilation was 14.2 days, median ICU LOS was 18.7 days, and hospital LOS was 24.7 days. No patients suffered mortality. Three patients experienced leukopenia, one of which experienced a concurrent neutropenia. Three patients experienced hyperglycemia, 2 experienced hypertension, 1 experienced increased aspartate aminotransferase and increased alanine aminotransferase, and 1 experienced diarrhea. All adverse events assessed were classified as possible using published adverse event causality assessments.

CONCLUSIONS

Peramivir has been shown to be an effective therapy for the treatment of influenza H1N1 in critically ill pediatric patients. In our experience with 7 pediatric patients, peramivir was well tolerated at typical durations of therapy; however, increased vigilance is warranted during prolonged courses or in patients with reasons for altered pharmacokinetics and pharmacodynamics.

Authenticity and credibility aware detection of adverse drug events from social media

OBJECTIVES

Adverse drug events (ADEs) are among the top causes of hospitalization and death. Social media is a promising open data source for the timely detection of potential ADEs. In this paper, we study the problem of detecting signals of ADEs from social media.

METHODS

Detecting ADEs whose drug and AE may be reported in different posts of a user leads to major concerns regarding the content authenticity and user credibility, which have not been addressed in previous studies. Content authenticity concerns whether a post mentions drugs or adverse events that are actually consumed or experienced by the writer. User credibility indicates the degree to which chronological evidence from a user's sequence of posts should be trusted in the ADE detection. We propose AC-SPASM, a Bayesian model for the authenticity and credibility aware detection of ADEs from social media. The model exploits the interaction between content authenticity, user credibility and ADE signal quality. In particular, we argue that the credibility of a user correlates with the user's consistency in reporting authentic content.

RESULTS

We conduct experiments on a real-world Twitter dataset containing 1.2 million posts from 13,178 users. Our benchmark set contains 22 drugs and 8089 AEs. AC-SPASM recognizes authentic posts with F₁ - the harmonic mean of precision and recall of 80%, and estimates user credibility with precision@10 = 90% and NDCG@10 - a measure for top-10 ranking quality of 96%. Upon validation against known ADEs, AC-SPASM achieves F₁ = 91%, outperforming state-of-the-art baseline models by 32% (p < 0.05). Also, AC-SPASM obtains precision@456 = 73% and NDCG@456 = 94% in detecting and prioritizing unknown potential ADE signals for further investigation. Furthermore, the results show that AC-SPASM is scalable to large datasets.

CONCLUSIONS

Our study demonstrates that taking into account the content authenticity and user credibility improves the detection of ADEs from social media. Our work generates hypotheses to reduce experts' guesswork in identifying unknown potential ADEs.

Authenticity and credibility aware detection of adverse drug events from social media

OBJECTIVES

Adverse drug events (ADEs) are among the top causes of hospitalization and death. Social media is a promising open data source for the timely detection of potential ADEs. In this paper, we study the problem of detecting signals of ADEs from social media.

METHODS

Detecting ADEs whose drug and AE may be reported in different posts of a user leads to major concerns regarding the content authenticity and user credibility, which have not been addressed in previous studies. Content authenticity concerns whether a post mentions drugs or adverse events that are actually consumed or experienced by the writer. User credibility indicates the degree to which chronological evidence from a user's sequence of posts should be trusted in the ADE detection. We propose AC-SPASM, a Bayesian model for the authenticity and credibility aware detection of ADEs from social media. The model exploits the interaction between content authenticity, user credibility and ADE signal quality. In particular, we argue that the credibility of a user correlates with the user's consistency in reporting authentic content.

RESULTS

We conduct experiments on a real-world Twitter dataset containing 1.2 million posts from 13,178 users. Our benchmark set contains 22 drugs and 8089 AEs. AC-SPASM recognizes authentic posts with F₁ - the harmonic mean of precision and recall of 80%, and estimates user credibility with precision@10 = 90% and NDCG@10 - a measure for top-10 ranking quality of 96%. Upon validation against known ADEs, AC-SPASM achieves F₁ = 91%, outperforming state-of-the-art baseline models by 32% (p < 0.05). Also, AC-SPASM obtains precision@456 = 73% and NDCG@456 = 94% in detecting and prioritizing unknown potential ADE signals for further investigation. Furthermore, the results show that AC-SPASM is scalable to large datasets.

CONCLUSIONS

Our study demonstrates that taking into account the content authenticity and user credibility improves the detection of ADEs from social media. Our work generates hypotheses to reduce experts' guesswork in identifying unknown potential ADEs.

Trigger alerts associated with laboratory abnormalities on identifying potentially preventable adverse drug events in the intensive care unit and general ward

Background

Medication safety strategies involving trigger alerts have demonstrated potential in identifying drug-related hazardous conditions (DRHCs) and preventing adverse drug events in hospitalized patients. However, trigger alert effectiveness between intensive care unit (ICU) and general ward patients remains unknown. The objective was to investigate trigger alert performance in accurately identifying DRHCs associated with laboratory abnormalities in ICU and non-ICU settings.

Methods

This retrospective, observational study was conducted at a university hospital over a 1-year period involving 20 unique trigger alerts aimed at identifying possible drug-induced laboratory abnormalities. The primary outcome was to determine the positive predictive value (PPV) in distinguishing drug-induced abnormal laboratory values using trigger alerts in critically ill and general ward patients. Aberrant lab values attributed to medications without resulting in an actual adverse event ensuing were categorized as a DRHC.

Results

A total of 634 patients involving 870 trigger alerts were included. The distribution of trigger alerts generated occurred more commonly in general ward patients (59.8%) than those in the ICU (40.2%). The overall PPV in detecting a DRHC in all hospitalized patients was 0.29, while the PPV in non-ICU patients (0.31) was significantly higher than the critically ill (0.25) (p = 0.03). However, the rate of DRHCs was significantly higher in the ICU than the general ward (7.49 versus 0.87 events per 1000 patient days, respectively, p < 0.0001). Although most DRHCs were considered mild or moderate in severity, more serious and life-threatening DRHCs occurred in the ICU compared with the general ward (39.8% versus 12.4%, respectively, p < 0.001).

Conclusions

Overall, most trigger alerts performed poorly in detecting DRHCs irrespective of patient care setting. Continuous process improvement practices should be applied to trigger alert performance to improve clinician time efficiency and minimize alert fatigue.

Technologic Distractions (Part 1): Summary of Approaches to Manage Alert Quantity With Intent to Reduce Alert Fatigue and Suggestions for Alert Fatigue Metrics

OBJECTIVE

To provide ICU clinicians with evidence-based guidance on tested interventions that reduce or prevent alert fatigue within clinical decision support systems.

DESIGN

Systematic review of PubMed, Embase, SCOPUS, and CINAHL for relevant literature from 1966 to February 2017.

PATIENTS

Focus on critically ill patients and included evaluations in other patient care settings, as well.

INTERVENTIONS

Identified interventions designed to reduce or prevent alert fatigue within clinical decision support systems.

MEASUREMENTS AND MAIN RESULTS

Study selection was based on one primary key question to identify effective interventions that attempted to reduce alert fatigue and three secondary key questions that covered the negative effects of alert fatigue, potential unintended consequences of efforts to reduce alert fatigue, and ideal alert quantity. Data were abstracted by two reviewers independently using a standardized abstraction tool. Surveys, meeting abstracts, "gray" literature, studies not available in English, and studies with non-original data were excluded. For the primary key question, articles were excluded if they did not provide a comparator as key question 1 was designed as a problem, intervention, comparison, and outcome question. We anticipated that reduction in alert fatigue, including the concept of desensitization may not be directly measured and thus considered interventions that reduced alert quantity as a surrogate marker for alert fatigue. Twenty-six articles met the inclusion criteria.

CONCLUSION

Approaches for managing alert fatigue in the ICU are provided as a result of reviewing tested interventions that reduced alert quantity with the anticipated effect of reducing fatigue. Suggested alert management strategies include prioritizing alerts, developing sophisticated alerts, customizing commercially available alerts, and including end user opinion in alert selection. Alert fatigue itself is studied less frequently, as an outcome, and there is a need for more precise evaluation. Standardized metrics for alert fatigue is needed to advance the field. Suggestions for standardized metrics are provided in this document.

Clinical Practice Guideline: Safe Medication Use in the ICU

OBJECTIVE

To provide ICU clinicians with evidence-based guidance on safe medication use practices for the critically ill.

DATA SOURCES

PubMed, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, CINAHL, Scopus, and ISI Web of Science for relevant material to December 2015.

STUDY SELECTION

Based on three key components: 1) environment and patients, 2) the medication use process, and 3) the patient safety surveillance system. The committee collectively developed Population, Intervention, Comparator, Outcome questions and quality of evidence statements pertaining to medication errors and adverse drug events addressing the key components. A total of 34 Population, Intervention, Comparator, Outcome questions, five quality of evidence statements, and one commentary on disclosure was developed.

DATA EXTRACTION

Subcommittee members were assigned selected Population, Intervention, Comparator, Outcome questions or quality of evidence statements. Subcommittee members completed their Grading of Recommendations Assessment, Development, and Evaluation of the question with his/her quality of evidence assessment and proposed strength of recommendation, then the draft was reviewed by the relevant subcommittee. The subcommittee collectively reviewed the evidence profiles for each question they developed. After the draft was discussed and approved by the entire committee, then the document was circulated among all members for voting on the quality of evidence and strength of recommendation.

DATA SYNTHESIS

The committee followed the principles of the Grading of Recommendations Assessment, Development, and Evaluation system to determine quality of evidence and strength of recommendations.

CONCLUSIONS

This guideline evaluates the ICU environment as a risk for medication-related events and the environmental changes that are possible to improve safe medication use. Prevention strategies for medication-related events are reviewed by medication use process node (prescribing, distribution, administration, monitoring). Detailed considerations to an active surveillance system that includes reporting, identification, and evaluation are discussed. Also, highlighted is the need for future research for safe medication practices that is specific to critically ill patients.

Prospective Assessment of Inpatient Boxed Warning Prescriber Adherence

OBJECTIVE

To evaluate medication boxed warning nonadherence in the inpatient setting.

METHODS

This was a prospective cohort quality improvement project approved by our institution's Total Quality Council. General medicine and ICU patients 18 years and older were included if they were cared for by a prescriber-led multidisciplinary team that included a pharmacist. Patients were evaluated for medication orders with an actionable boxed warning; if boxed warning nonadherence occurred, the physician's reason was determined. Patients with boxed warning nonadherence were monitored for adverse drug reactions until discharge.

RESULTS

A total of 393 patients (224 general medicine and 169 ICU) were evaluated for nonadherence to 149 actionable boxed warnings. There were 293 drugs (175 general medicine and 118 ICU) with boxed warnings prescribed, and more than 50% of these were medications restarted from home. A total of 23 boxed warning nonadherences occurred in general medicine patients, and NSAIDs accounted for 81% of these events. ICU patients experienced 11 boxed warning nonadherences, with nearly 54% from anti-infectives and immunosuppressants. Antipsychotics were the most commonly ordered boxed warning medication class in ICU patients. Reasons for nonadherence included knowledge deficit and an acceptable risk-to-benefit ratio. Two adverse drug reactions occurred from boxed warning nonadherences, both because of a drug-drug interaction.

CONCLUSIONS

Boxed warning nonadherence is a concern in the inpatient setting, specifically with NSAID use in general medicine patients and antipsychotic use in ICU patients. More than half of boxed warning nonadherence occurred in medications restarted from home, which emphasizes the need for medication evaluation during transitions of care.

Adverse drug reactions in therapeutic hypothermia after cardiac arrest

BACKGROUND

Therapeutic hypothermia (TH) improves survival and neurologic function in comatose survivors of cardiac arrest. Many medications used to support TH have altered pharmacokinetics and pharmacodynamics during this treatment. It is unknown if or at what frequency the medications used during TH cause adverse drug reactions (ADRs).

METHODS

A retrospective chart review was conducted for patients admitted to an intensive care unit (ICU) after cardiac arrest and treated with TH from January 2009 to June 2012 at two urban, university-affiliated, tertiary-care medical centres. Medications commonly used during TH were screened for association with significant ADRs (grade 3 or greater per Common Terminology Criteria for Adverse Events) using three published ADR detection instruments.

RESULTS

A total of 229 patients were included, the majority being males with median age of 62 presenting with an out-of-hospital cardiac arrest in pulseless electrical activity or asystole. The most common comorbidities were hypertension, coronary artery disease, and diabetes mellitus. There were 670 possible ADRs and 69 probable ADRs identified. Of the 670 possible ADRs, propofol, fentanyl, and acetaminophen were the most common drugs associated with ADRs. Whereas fentanyl, insulin, and propofol were the most common drugs associated with a probable ADR. Patients were managed with TH for a median of 22 hours, with 38% of patients surviving to hospital discharge.

CONCLUSIONS

Patients undergoing TH after cardiac arrest frequently experience possible adverse reactions associated with medications and the corresponding laboratory abnormalities are significant. There is a need for judicious use and close monitoring of drugs in the setting of TH until recommendations for dose adjustments are available to help prevent ADRs.

An Evaluation of Adverse Drug Reactions Associated With Antipsychotic Use for the Treatment of Delirium in the Intensive Care Unit

Purpose:

This investigation evaluated the incidence, severity, and harm of adverse drug reactions (ADRs) associated with antipsychotic use for intensive care unit (ICU) delirium.

Methods:

In this prospective, observational study patients were screened for development of delirium with the Intensive Care Delirium Screening Checklist (ICDSC). An ICDSC score of ≥4 was considered delirious. Patients with delirium were screened daily for ADRs. Suspected ADRs were evaluated for drug causality using 3 published, objective assessment tools. Suspected ADRs were considered positive when 2 of 3 instruments had an agreement rating of “possible” or greater. ADR severity was defined as “mild/moderate” or “severe” using the National Cancer Institute’s Common Terminology Criteria for Adverse Events scale. A modified National Coordinating Council Medication Error Reporting Index for Categorizing Errors categorized ADRs into “no harm” or “harmful.”

Results:

Of 90 patients with delirium, 56 received antipsychotics. Ten suspected ADRs occurred attributed to antipsychotic use. QTc prolongation was the most observed ADR (50%). Patients with ADRs had higher mean Acute Physiology and Chronic Health Evaluation II (APACHE II) scores ( P = .038). Patients who received haloperidol experienced more severe ( P = .048) ADRs.

Conclusions:

ADRs were observed in 18% of patients having delirium treated with antipsychotics with about half considered severe or harmful. A risk versus benefit assessment is needed before initiating antipsychotic therapy in ICU patients.

Evaluation of an automated surveillance system using trigger alerts to prevent adverse drug events in the intensive care unit and general ward

INTRODUCTION

Adverse events in the intensive care unit (ICU) may be associated with several possible causes, so determining a drug-related causal assessment is more challenging than in general ward patients. Therefore, the hypothesis was that automated trigger alerts may perform differently in various patient care settings. The purpose of this study was to compare the frequency and type of clinically significant automated trigger alerts in critically ill and general ward patients as well as evaluate the performance of alerts for drug-related hazardous conditions (DRHCs).

METHODS

A retrospective cohort study was conducted in adult ICU and general ward patients at three institutions (academic, community, and rural hospital) in a health system. Automated trigger alerts generated during two nonconsecutive months were obtained from a centralized database. Pharmacist responses to alerts and prescriber response to recommendations were evaluated for all alerts. A clinical significant event was defined as an actionable intervention requiring drug therapy changes that the pharmacist determined to be appropriate for patient safety and where the physician accepted the pharmacist's recommendation. The positive predictive value (PPV) was calculated for each trigger alert considered a DRHC (i.e., abnormal laboratory values and suspected drug causes).

RESULTS

A total of 751 alerts were generated in 623 patients during the study period. Pharmacists intervened on 39.8 and 44.8 % alerts generated in the ICU and general ward, respectively. Overall, the physician acceptance rate of approximately 90 % was comparable irrespective of patient care setting. Therefore, the number of clinically significant alerts was 88.9 and 83.4 % for the ICU and non-ICU, respectively. The types of drug therapy changes were similar between settings. The PPV of alerts identifying a DRHC was 0.66 in the ICU and 0.76 in general ward patients.

CONCLUSIONS

The number and type of clinically significant alerts were similar irrespective of patient population, suggesting that the alerts may be equally as beneficial in the ICU population, despite the challenges in drug-related event adjudication. An opportunity exists to improve the performance of alerts in both settings, so quality improvement programs for measuring alert performance and making refinements is needed.

Using a Clinical Surveillance System to Detect Drug-Associated Hypoglycemia in Nursing Home Residents

OBJECTIVES

To determine whether a clinical surveillance system could be used to detect drug-associated hypoglycemia events and determine their incidence in nursing home (NH) residents.

DESIGN

Retrospective cohort.

SETTING

Four NHs in western Pennsylvania.

PARTICIPANTS

Any resident of the four NHs who had a computer-generated alert detecting potential drug-associated hypoglycemia over a 6-month period were included.

MEASUREMENTS

Descriptive statistics were used to summarize all variables, including the frequency and distribution of alert type according to glucose threshold. Analyses were conducted according to numbers of alerts and residents. The medications associated with the drug-associated hypoglycemia alerts were identified, and frequency of their inclusion in alerts was calculated. Additional calculations included time to drug-associated hypoglycemic event alert from date of admission and frequency of events associated with postacute, short-stay (≤35 days) admissions.

RESULTS

Seven hundred seventy-two alerts involving 141 residents were detected. Ninety (63.8%) residents had a glucose level of 55 mg/dL or less, and 42 (29.8%) had a glucose level of 40 mg/dL or less. Insulin orders were associated with 762 (98.7%) alerts. Overall incidence of drug-associated hypoglycemia events was 9.5 per 1,000 resident-days.

CONCLUSION

Hypoglycemia can be detected using a clinical surveillance system. This evaluation found a high incidence of drug-associated hypoglycemia in a general NH population. Future studies are needed to determine the potential benefits of use of a surveillance system in real-time detection and management of hypoglycemia in NHs.

A Multicenter Evaluation of Off-Label Medication Use and Associated Adverse Drug Reactions in Adult Medical ICUs

OBJECTIVE

Prior research indicates that off-label use is common in the ICU; however, the safety of off-label use has not been assessed. The study objective was to determine the prevalence of adverse drug reactions associated with off-label use and evaluate off-label use as a risk factor for the development of adverse drug reactions in an adult ICU population.

DESIGN

Multicenter, observational study

SETTING

: Medical ICUs at three academic medical centers.

PATIENTS

Adult patients (age ≥ 18 yr old) receiving medication therapy.

INTERVENTIONS

All administered medications were evaluated for Food and Drug Administration-approved or off-label use. Patients were assessed daily for the development of an adverse drug reaction through active surveillance. Three adverse drug reaction assessment instruments were used to determine the probability of an adverse drug reaction resulting from drug therapy. Severity and harm of the adverse drug reaction were also assessed. Cox proportional hazard regression was used to identify a set of covariates that influenced the rate of adverse drug reactions.

MEASUREMENTS AND MAIN RESULTS

Overall, 1,654 patient-days (327 patients) and 16,391 medications were evaluated, with 43% of medications being used off-label. One hundred and sixteen adverse drug reactions were categorized dichotomously (Food and Drug Administration or off-label), with 56% and 44% being associated with Food and Drug Administration-approved and off-label use, respectively. The number of adverse drug reactions for medications administered and the number of harmful and severe adverse drug reactions did not differ for medications used for Food and Drug Administration-approved or off-label use (0.74% vs 0.67%; p = 0.336; 33 vs 31 events, p = 0.567; 24 vs 24 events, p = 0.276). Age, sex, number of high-risk medications, number of off-label medications, and severity of illness score were included in the Cox proportional hazard regression. It was found that the rate of adverse drug reactions increases by 8% for every one additional off-label medication (hazard ratio = 1.08; 95% CI, 1.018-1.154).

CONCLUSION

Although adverse drug reactions do not occur more frequently with off-label use, adverse drug reaction risk increases with each additional off-label medication used.

Risk factors for acute kidney injury in older adults with critical illness: a retrospective cohort study

BACKGROUND

Risk for acute kidney injury (AKI) in older adults has not been evaluated systematically. We sought to delineate the determinants of risk for AKI in older compared with younger adults.

STUDY DESIGN

Retrospective analysis of patients hospitalized in July 2000 to September 2008.

SETTING & PARTICIPANTS

We identified all adult patients admitted to an intensive care unit (n=45,655) in a large tertiary-care university hospital system. We excluded patients receiving dialysis or a kidney transplant prior to hospital admission and patients with baseline creatinine levels ≥ 4mg/dL, liver transplantation, indeterminate AKI status, or unknown age, leaving 39,938 patients.

PREDICTOR

We collected data for multiple susceptibilities and exposures, including age, sex, race, body mass, comorbid conditions, severity of illness, baseline kidney function, sepsis, and shock.

OUTCOMES

We defined AKI according to KDIGO (Kidney Disease: Improving Global Outcomes) criteria. We examined susceptibilities and exposures across age strata for impact on the development of AKI.

MEASUREMENTS

We calculated area under the receiver operating characteristic curve (AUC) for prediction of AKI across age groups.

RESULTS

25,230 (63.2%) patients were 55 years or older. Overall, 25,120 (62.9%) patients developed AKI (69.2% aged ≥55 years). Examples of risk factors for AKI in the oldest age category (≥75 years) were drugs (vancomycin, aminoglycosides, and nonsteroidal anti-inflammatories), history of hypertension (OR, 1.13; 95% CI, 1.02-1.25), and sepsis (OR, 2.12; 95% CI, 1.68-2.67). Fewer variables remained predictive of AKI as age increased and the model for older patients was less predictive (P<0.001). For the age categories 18 to 54, 55 to 64, 65 to 74, and 75 years or older, AUCs were 0.744 (95% CI, 0.735-0.752), 0.714 (95% CI, 0.702-0.726), 0.706 (95% CI, 0.693-0.718), and 0.673 (95% CI, 0.661-0.685), respectively.

LIMITATIONS

Analysis may not apply to non-intensive care unit patients.

CONCLUSIONS

The likelihood of developing AKI increases with age; however, the same variables are less predictive for AKI as age increases. Efforts to quantify risk for AKI may be more difficult in older adults.

AKI in low-risk versus high-risk patients in intensive care

BACKGROUND AND OBJECTIVES

AKI in critically ill patients is usually part of multiorgan failure. However, nonrenal organ failure may not always precede AKI and patients without evidence of these organ failures may not be at low risk for AKI. This study examined the risk and outcomes associated with AKI in critically ill patients with and without cardiovascular or respiratory organ failures at presentation to the intensive care unit (ICU).

DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS

A large, academic medical center database, with records from July 2000 through October 2008, was used and the authors identified a low-risk cohort as patients without cardiovascular and respiratory organ failures defined as not receiving vasopressor support or mechanical ventilation within the first 24 hours of ICU admission. AKI was defined using Kidney Disease Improving Global Outcomes criteria. The primary end points were moderate to severe AKI (stages 2-3) and risk-adjusted hospital mortality.

RESULTS

Of 40,152 critically ill patients, 44.9% received neither vasopressors nor mechanical ventilation on ICU day 1. Stages 2-3 AKI occurred less frequently in the low-risk patients versus high-risk patients within 24 hours (14.3% versus 29.1%) and within 1 week (25.7% versus 51.7%) of ICU admission. Patients developing AKI in both risk groups had higher risk of death before hospital discharge. However, the adjusted odds of hospital mortality were greater (odds ratio, 2.99; 95% confidence interval, 2.62 to 3.41) when AKI occurred in low-risk patients compared with those with respiratory or cardiovascular failures (odds ratio, 1.19; 95% confidence interval, 1.09 to 1.3); interaction P<0.001.

CONCLUSIONS

Patients admitted to ICU without respiratory or cardiovascular failure have a substantial likelihood of developing AKI. Although survival for low-risk patients is better than for high-risk patients, the relative increase in mortality associated with AKI is actually greater for low-risk patients. Strategies aimed at preventing AKI should not exclude ICU patients without cardiovascular or respiratory organ failures.

A multicenter study of the point prevalence of drug-induced hypotension in the ICU

OBJECTIVE

To determine the point prevalence of drug-induced hypotension episodes in critically ill patients, to assess the episodes resulting from error, and to describe how episodes are treated.

DESIGN

Multicenter observational, 24-hour snapshot study.

SETTING

Forty-seven ICUs in 27 institutions located in the United States, Canada, and Singapore.

PATIENTS

A total of 688 ICU patients were evaluated.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Patients were included in the study if they had an episode of hypotension in the 24 hours prior to the clinical pharmacists' evaluation. The definition for a hypotensive episode is either a systolic blood pressure less than 90 mm Hg or a decrease in systolic blood pressure of 30 mm Hg over a 2-hour period. Each episode of unintentional hypotension was assessed for suspected drug-related causes. When a drug-related cause was suspected, an objective assessment tool, the modified Kramer, was used to determine causality. A score of at least "possible" was considered drug induced, referred to as a "drug-related hazardous condition." A drug-related hazardous condition is the temporal gap (intermediate stage) between the identification of an adverse drug reaction and the subsequent onset of drug-induced injury, known as an "adverse drug event." Drug-induced episodes were evaluated for medication errors and treatment. One hundred fifty-eight patients experienced 204 hypotensive episodes that were considered unintentional and drug related. Common drugs implicated included propofol, fentanyl, metoprolol, lorazepam, hydralazine, and furosemide. A total of 54 episodes (26.5%) resulted from medication errors. Common error types were improper dose/quantity (46%) and prescribing (25%). A total of 56.9% episodes were treated.

CONCLUSIONS

Many hypotensive episodes in the ICU are drug related and require treatment. A substantial portion of these episodes result from errors and are therefore preventable. This presents opportunities to improve prescribing including optimizing drug dosing to avoid possible patient harm from drug-induced hypotension.

A study of agreement between the Naranjo algorithm and WHO-UMC criteria for causality assessment of adverse drug reactions

Objectives:

Reliability and usefulness of various adverse drug reaction (ADR) causality assessment scales have not been fully explored. There is no universally accepted method for causality grading of ADRs. In the present study we assessed agreement between the two widely used causality assessment scales, that is, the World Health Organization-Uppsala Monitoring Center (WHO-UMC) criteria and the Naranjo algorithm.

Materials and Methods:

The same observer assessed all ADRs (n = 913) collected between January 2010 and December 2012 using the WHO-UMC criteria and Naranjo algorithm at a tertiary care hospital in India. We found that the most frequently assigned causality category was “possible” with both the scales.

Results:

A disagreement in the causality assessment was found in 45 (4.9%) cases reflecting “poor” agreement between the two scales (Kappa statistic with 95% confidence interval = 0.143 [0.018, 0.268]). The mean time taken to assess causality of the ADR using the WHO-UMC criteria was shorter than that by the Naranjo algorithm.

Conclusion:

This study showed that there is a poor agreement between the WHO-UMC criteria and Naranjo algorithm with the former being less time-consuming.

Medication safety and knowledge-based functions: a stepwise approach against information overload

AIMS

The aim was to improve medication safety in an emergency department (ED) by enhancing the integration and presentation of safety information for drug therapy.

METHODS

Based on an evaluation of safety of drug therapy issues in the ED and a review of computer-assisted intervention technologies we redesigned an electronic case sheet and implemented computer-assisted interventions into the routine work flow. We devised a four step system of alerts, and facilitated access to different levels of drug information. System use was analyzed over a period of 6 months. In addition, physicians answered a survey based on the technology acceptance model TAM2.

RESULTS

The new application was implemented in an informal manner to avoid work flow disruption. Log files demonstrated that step I, 'valid indication' was utilized for 3% of the recorded drugs and step II 'tooltip for well-known drug risks' for 48% of the drugs. In the questionnaire, the computer-assisted interventions were rated better than previous paper based measures (checklists, posters) with regard to usefulness, support of work and information quality.

CONCLUSION

A stepwise assisting intervention received positive user acceptance. Some intervention steps have been seldom used, others quite often. We think that we were able to avoid over-alerting and work flow intrusion in a critical ED environment.

A real-world, multicenter assessment of drugs requiring weight-based calculations in overweight, adult critically ill patients

Prescribing appropriate doses of drugs requiring weight-based dosing is challenging in overweight patients due to a lack of data. With 68% of the US population considered overweight and these patients being at an increased risk for hospitalization, clinicians need guidance on dosing weight-based drugs. The purpose of this study was to identify “real-world” dose ranges of high-risk medications administered via continuous infusion requiring weight-based dosing and determine the reasons for dosing changes (ineffectiveness or adverse drug reactions). A prospective, multicenter, observational study was conducted in four intensive care units at three institutions. A total of 857 medication orders representing 11 different high-risk medications in 173 patients were reviewed. It was noted that dosing did not increase in proportion to weight classification. Overall, 14 adverse drug reactions occurred in nine patients with more in overweight patients (9 of 14). A total of 75% of orders were discontinued due to ineffectiveness in groups with higher body mass indexes. Ineffectiveness leads to dosing adjustments resulting in the opportunity for medication errors. Also, the frequent dosing changes further demonstrate our lack of knowledge of appropriate dosing for this population. Given the medications' increased propensity to cause harm, institutions should aggressively monitor these medications in overweight patients.

Clinical evaluation of the ADE scorecards as a decision support tool for adverse drug event analysis and medication safety management

AIMS

The prevention of adverse drug events (ADEs) demands co-ordination of different health care professionals. ADE scorecards are a novel approach to raise the team awareness regarding ADE risks and causes. It makes information on numbers and on possible causes of possible ADE cases available to the clinical team. The aim of the study was to investigate the usage and acceptance of ADE scorecards by healthcare professionals and their impact on rates of possible ADEs.

METHODS

ADE scorecards were introduced in three departments of a French hospital. A controlled time series analysis of ADE data was conducted to assess the impact of the ADE scorecards. In addition, qualitative interviews and a standardized survey with all participating staff members were performed.

RESULTS

Physicians, nurses and pharmacists found ADE scorecards effective to increase medication safety and recommended future usage. The time-series analysis did not show changes in rates of possible ADEs.

CONCLUSION

ADE scorecards appear to be useful to raise awareness of ADE-related issues among professionals. Although the evaluation did not show significant reductions of ADE rates, the participating physicians, nurses and pharmacists believed that the ADE scorecards could contribute to increased patient safety and to a reduction in ADE rates. Strategies need to be designed to integrate ADE scorecards better into the clinical routine and to increase the precision of ADE detection.

Evaluation of medication dose alerts in pediatric inpatients

OBJECTIVE

This study evaluates the impact of 12,093 consecutive dose alerts generated by a computerized provider order entry system on pediatric medication ordering.

PATIENTS AND METHODS

All medication orders entered and all resulting medication dose alerts at the Johns Hopkins Children's Medical and Surgical Center in 2010, were retrospectively evaluated. Inclusion criteria were hospitalized patients less than 21 years old. There were no exclusion criteria.

RESULTS

During 2010, there were 7738 admissions for 5553 unique patients. A total of 182,308 medication orders for 1092 unique medications were submitted by providers. Six percent (11,155) of orders or order attempts generated alerts for 2046 patients and 524 medications. Two categories of alerts were analyzed: dose range alerts and informational alerts. 73.4% (8187) of all alerts were dose range alerts, with a compliance rate of 8.5% (694); 26.6% (2968) were informational alerts, with a compliance rate of 5.5% (163).

CONCLUSIONS

We found that underdosing alerts provide less value to providers than overdosing alerts. However, the low compliance with the alerts should trigger the evaluation of clinical practice behavior and the existing alert thresholds. Informational alerts noting the absence of established dosing guidelines had little effect on provider behavior and should be avoided when building a dose range alert system.

Improving adverse drug event detection in critically ill patients through screening intensive care unit transfer summaries

PURPOSE

This study aimed to determine the frequency and type of adverse drug events (ADEs) identified in intensive care unit (ICU) transfer summaries and in the hospital discharge summaries to demonstrate the effectiveness of ICU transfer summary surveillance in the identification of ADEs.

METHODS

A retrospective electronic medical record review was conducted for medical ICU patients admitted between January 2009 and April 2009 to a large, academic medical center. The Harvard Practice Scale and the modified Leonard Assessment Scale were used to evaluate the presence of an ADE from the ICU transfer and hospital discharge summaries.

RESULTS

Two hundred and fifty-four patients were identified for inclusion with a median medical ICU length of stay of 4.5 days and hospital length of stay of 13 days. The ICU transfer summary review revealed 173 ADEs among 124 unique patients with a rate of 33.9 ADEs per 1000 hospital patient days. Sixty-nine ADEs among 63 unique patients were identified through the hospital discharge summary with a rate of 13.5 ADEs per 1000 hospital patient days. Only 23.1% of ADEs discussed in the ICU transfer summary were also discussed in the hospital discharge summary.

CONCLUSIONS

The use of ICU transfer summaries is an effective tool to increase ADE detection. The use of an ICU transfer summary should be considered as an adjunct method to an existing ADE surveillance system for heightened pharmacovigilance.

Drug-drug interactions contributing to QT prolongation in cardiac intensive care units

PURPOSE

To determine the most common drug-drug interaction (DDI) pairs contributing to QTc prolongation in cardiac intensive care units (ICUs).

MATERIALS AND METHODS

This retrospective evaluation included patients who were admitted to the cardiac ICUs between January 2009 and July 2009 aged ≥ 18 years with electrocardiographic evidence of a QTc ≥ 500 ms. Patients receiving at least two concomitant drugs known to prolong the QT interval were considered to experience a pharmacodynamic DDI. Drugs causing CYP450 inhibition of the metabolism of QT prolonging medications were considered to cause pharmacokinetic DDIs. The causality between drug and QTc prolongation was evaluated with an objective scale.

RESULTS

One hundred eighty-seven patients experienced QT prolongation out of a total of 501 patients (37%) admitted during the study period. Forty-three percent and 47% of patients experienced 133 and 179 temporally-related pharmacodynamic and pharmacokinetic interactions, respectively. The most common medications related to these DDIs were ondansetron, amiodarone, metronidazole, and haloperidol.

CONCLUSION

DDIs may be a significant cause of QT prolongation in cardiac ICUs. These data can be used to educate clinicians on safe medication use. Computerized clinical decision support could be applied to aid in the detection of these events.

Real-time pharmacy surveillance and clinical decision support to reduce adverse drug events in acute kidney injury: a randomized, controlled trial

OBJECTIVES: Clinical decision support (CDS), such as computerized alerts, improves prescribing in the setting of acute kidney injury (AKI), but considerable opportunity remains to improve patient safety. The authors sought to determine whether pharmacy surveillance of AKI patients could detect and prevent medication errors that are not corrected by automated interventions. METHODS: The authors conducted a randomized clinical trial among 396 patients admitted to an academic, tertiary care hospital between June 1, 2010 and August 31, 2010 with an acute 0.5 mg/dl change in serum creatinine over 48 hours and a nephrotoxic or renally cleared medication order. Patients randomly assigned to the intervention group received surveillance from a clinical pharmacist using a web-based surveillance tool to monitor drug prescribing and kidney function trends. CDS alerting and standard pharmacy services were active in both study arms. Outcome measures included blinded adjudication of potential adverse drug events (pADEs), adverse drug events (ADEs) and time to provider modification or discontinuation of targeted nephrotoxic or renally cleared medications. RESULTS: Potential ADEs or ADEs occurred for 104 (8.0%) of control and 99 (7.1%) of intervention patient-medication pairs (p=0.4). Additionally, the time to provider modification or discontinuation of targeted nephrotoxic or renally cleared medications did not differ between control and intervention patients (33.4 hrs vs. 30.3 hrs, p=0.3). CONCLUSIONS: Pharmacy surveillance had no incremental benefit over previously implemented CDS alerts.

Use of an abnormal laboratory value-drug combination alert to detect drug-induced thrombocytopenia in critically Ill patients

PURPOSE

The aim of this study was to assess the performance of a commercially available clinical decision support system (CDSS) drug-laboratory result alert in detecting drug-induced thrombocytopenia in critically ill patients.

MATERIALS AND METHODS

Adult patients admitted to the medical and cardiac intensive care unit during an 8-week period and identified by 1 of 3 signals in the CDSS, TheraDoc, were eligible. Alerts were generated when the patient had a low platelet count and was ordered a potentially causal drug. Patients were evaluated in real time for the occurrence of an adverse drug reaction using 3 causality instruments. Positive predictive values were calculated for the alert.

RESULTS

Sixty-four patients with a mean age of 54 years met the inclusion criteria, generating 350 alerts. Positive predictive values were 0.36, 0.83, and 0.40 for signals 1, 2, and 3, respectively. Overall, there were 137 adverse drug reactions identified in the 350 alerts, with heparin, vancomycin, and famotidine as the 3 most common potential causes.

CONCLUSIONS

A commercial CDSS drug-laboratory alert is effective at identifying drug-induced thrombocytopenia in the intensive care unit and may improve patient safety. Compared with previous studies, the combination alert performs better than alerts based exclusively on laboratory values and should be considered to reduce alert fatigue.