Adverse Drug Events Among Hospitalized Medicare Patients: Epidemiology and National Estimates from a New Approach to Surveillance

The Joint Commission Journal on Quality and Patient Safety Medication Safety

Adverse Drug Events Among Hospitalized Medicare Patients: Epidemiology and National Estimates from a New Approach to Surveillance David C. Classen, M.D., M.S.; Lisa Jaser, R.Ph., Pharm.D.; Daniel S. Budnitz, M.D., M.P.H.

A

dverse drug events (ADEs) continue to harm hospitalized patients and may represent the single greatest safety risk for patients in hospitals. Although there are a number of definitions of ADEs in the medical literature, in general, an ADE is considered to be an injury resulting from medical intervention related to a drug. ADEs were reported as one key area for improvement in the Institute of Medicine (IOM)’s Quality Chasm report.1 These findings ignited interest among the health care community, as well as public awareness of this safety problem. In 2007, the IOM appraisal of the health care chasm continued to report dismal findings, with no significant improvement in overall outcomes.2 The IOM has estimated that at least 1.5 million preventable ADEs occur annually in the United States.3 National attempts to address ADEs have been hampered by lack of efficient surveillance systems.4 Most ADE monitoring systems rely on voluntary reporting and thus underestimate the true health burden of ADEs. Attempts to improve the identification and reporting of ADEs have used enhanced reporting by health care providers, submission via electronic databases, patient surveys, and implicit review of medical records.3,5–13 These approaches all require considerable resources and have not been generalized across various hospital systems for sustained periods.3,5 Despite the challenges in quantifying the incidence of ADEs among hospitalized patients, the economic costs of ADEs are likely to be considerable. In one study of hospitalized patients, Bates et al. found an additional length of stay of 2.2 days and additional costs of $3,244 per ADE.14 In another study, which examined the use of antithrombolytics in heart disease, Eckman et al. estimated an additional cost of $3,000–$12,000 per major anticoagulation-related bleed.15 On the basis of a projection of 400,000 ADEs per year in hospitalized patients, the IOM estimated that ADEs accounted for $3.5 billion dollars of additional costs incurred by hospitals.3 Several studies have also demonstrated an increased risk for ADEs among older hospitalized patients.16–18 Factors that can be 12

January 2010

Article-at-a-Glance Background: Although adverse drug events (ADEs) are a

well-recognized problem among hospitalized patients, there is no system for monitoring them. Six high-alert medications and associated adverse events were selected for inclusion in the Medicare Patient Safety Monitoring System (MPSMS), a national surveillance system designed to identify and track over time inpatient adverse events within the hospitalized fee-for-service Medicare population. Methods: Explicit chart review algorithms were used to identify medication exposures and associated adverse events from the the 2004 MPSMS sample’s medical records. The associations of ADEs with patient characteristics, length of stay, mortality, and 30-day readmission were assessed with bivariate analyses and hierarchical linear regression modeling (HGLM) approaches. National ADE rates and numbers of adverse events were estimated using weighted HGLM. Results: On the basis of 25,145 hospital visits in the 2004 MPSMS sample, an estimated 8.2% of patients exposed to warfarin experienced associated ADEs, as did 13.6% exposed to heparin, 10.7% exposed to insulin/hypoglycemic agents, and 0.5% exposed to digoxin. Some 0.6% of patients exposed to antibiotics experienced antibiotic-associated Clostridium difficile infection (CDI). Patients with ADEs had increased length of stay and in-hospital and 30-day mortality, except that patients with antibiotic-associated CDI did not have increased in-hospital mortality, and patients with ADEs associated with heparin did not have increased 30-day mortality. An estimated 888,000 ADEs occurred in hospitalized Medicare patients from these medications alone. Discussion: This new approach to detecting ADEs and estimating the national burden of ADEs from selected medications may be adapted for other types of ADEs in the Medicare population and may offer guidance to policymakers on appropriate areas of focus for patient safety.

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety correlated with the increased risk in elderly patients include increased use of medications and comorbid conditions such as renal and hepatic impairment. More than 80% of elderly patients have one or more chronic illnesses, resulting in complicated daily medication regimens of multiple medications.19 An increase in the number of medications increases the likelihood of drug-drug interactions, drug-food interactions, as well as drug-disease interactions.19 Not all of the thousands of medications in clinical use have equal risks in older patients. Among ambulatory older adults, most acute, serious adverse events are caused by a relatively small number of medications.20 The Institute of Safe Medication Practice has identified a number of medications considered “high-alert” medications,21 defined by The Joint Commission as those “that have the highest risk of causing injury when misused.”22 Because of the complexity of attempting to identify and prevent all ADEs, focusing surveillance and prevention on the high-alert medications may be a prudent approach. For example, the Institute for Healthcare Improvement (IHI)’s 5 Million Lives Campaign has focused on four groups of high-alert medications: anticoagulants, narcotics and opiates, insulin, and sedatives.23 On the basis of the evidence of health burden, costs, and preventability, six high-alert medications and associated adverse events were selected for inclusion in the Medicare Patient Safety Monitoring System (MPSMS), a national surveillance system designed to identify and track over time inpatient adverse events within the hospitalized fee-for-service Medicare population. The six high-alert medications are warfarin, heparin, low molecular weight heparin (LMWH), insulin/hypoglycemic agents, digoxin, and antibiotics. In this article we describe the methodology and findings from the first year of MPSMS ADE surveillance.

Methods THE MPSMS The MPSMS is a national surveillance system to determine and track over time the rates of specific inpatient adverse events within the hospitalized, fee-for-service Medicare population. The MPSMS was created under the auspices of the Department of Health and Human Services Patient Safety Task Force (HHSPST).24 Development of the MPSMS was announced in April 2001 by Stephen Jencks, M.D., M.P.H., Director of the Quality Improvement Group at the Centers for Medicare & Medicaid Services (CMS), because of the increasing need to better measure the safety of care in the inpatient Medicare population.

January 2010

This surveillance system identifies exposures and adverse events in medical records and Medicare administrative data from randomly selected inpatient Medicare discharges in the United States, Puerto Rico, and the Virgin Islands. The MPSMS operationally defines an adverse event as unintended, measurable harm, injury, or loss more likely associated with the patient’s exposure and interaction with the health care delivery system than from any attendant disease process. In the MPSMS exposures are the events and processes of health care delivery that occur when providing inpatient care.24

STUDY SAMPLE The MPSMS, described in detail elsewhere,24 is a retrospective cohort study with a sample randomly selected from the Medicare National Claims History (NCH) database by CMS. In brief, the medical records for the MPSMS are identified for an annual random sample drawn from the NCH database, a pool of approximately 1 million Medicare fee-for-service (FFS) beneficiaries’ hospital discharges. Every month a set number of discharges are selected from each state for review by abstractors. For selected discharges, copies of the medical records are sent to two Clinical Data Abstraction Centers (CDACs), where they are used for multiple Medicare review programs, including the MPSMS. For this specific MPSMS ADE study, from the database as described, a sample of 25,145 hospital discharge records between January 1, 2004, and December 31, 2004, from complete medical charts were selected for each of 49 states; Washington, DC; and Puerto Rico. Hospitals forwarded randomly selected medical records to the two CDACs for data abstraction monthly.24 Complete hospital charts were reviewed by abstractors at the CDACs monthly in 2005, and all data collected were forwarded to the MPSMS project team, which reports the results in the following year and which also conducted further analysis for the 2005–2008 period. CDAC abstractors collect data directly from copies of the medical records using an electronic data collection software program, MedQuest (MedQuest Software, Inc., Middlebury, CT). This program has a series of modules that guide the abstractor through a structured review of the complete chart to identify adverse events that occurred during the hospitalization. Trained CDAC abstractors process each medical record by identifying and electronically recording the predefined data elements from the medical record. Information on patient demographics, ICD-9-CM (International Classification of Diseases, 9th Revision, Clinical Modification), postdischarge events, and readmissions after the index hospital discharge were obtained

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

13

The Joint Commission Journal on Quality and Patient Safety through the NCH database for Part A (hospital) Information. In-hospital mortality data were obtained from the NCH database, and 30-day mortality data were retrieved from the Medicare Enrollment database.24

MEASURES AND OUTCOMES The MPSMS program leadership convened a technical expert panel, which included the authors, on the basis of specific patient safety expertise. The panel selected six high-risk medications for ADE surveillance: warfarin, heparin, LMWH/Factor Xa inhibitor, insulin/oral hypoglycemic agents, digoxin, and systemic antibiotics. The CDAC abstractors reviewed all MPSMS medical records for exposure to these high-risk medications during the hospitalization and identified adverse events linked to these medications by using medication-specific algorithms (Figures 1–6, as shown in Appendix 1, available in online article) determined by the panel through a consensus process. Digoxin Exposure. Digoxin exposure was defined as patients who received digoxin and who had a digoxin serum level documented during their hospital stay. Adverse events were defined as the occurrence of any of the following conditions from the second day of admission through the day of discharge on the same hospital day in a patient with a measured digoxin level > 2.5 mg/mL or a diogoxin level > 1.5 mg/mL but < 2.5 mg/mL and a potassium level < 3.5 mEq/L (Figure 1): death, cardiac arrest/emergency measure to sustain life, ventricular fibrillation, ventricular tachycardia, premature ventricular contractions, bradycardia, atrioventricular (AV) block, PR interval prolongation, or administration of digoxin immune FAB. Antibiotic Exposure. Antibiotic exposure was defined as patients who had documentation that they received an antibiotic during the hospital stay. Hospital-acquired Clostridium difficile infection (CDI) was the only antibiotic-associated adverse event identified. Hospital-acquired CDI was defined as a positive C. difficile toxin assay more than two days after arrival to the hospital in a patient who had received an antibiotic during the hospital stay at least one day before the C. difficile toxin assay was ordered. Patients with a positive C. difficile toxin assay more than two days after arrival to the hospital but who did not have documentation of receiving an antibiotic during the hospital stay were excluded (Figure 2). Heparin Exposure. Heparin exposure was defined as patients who received heparin during hospitalization and who had a documented partial thromboplastin time (PTT) result during the hospital stay. Adverse events were defined as the occurrence at least of one of the following conditions within two days of a 14

January 2010

documented PTT > 100 or PTT > 45 accompanied by abrupt cessation/hold of heparin, administration of fresh frozen plasma (FFP), or blood transfusion (absent a surgical procedure): death, cardiac arrest/emergency measure to sustain life, intracranial bleeding, gastrointestinal bleeding, genitourinary bleeding, pulmonary bleeding, hematocrit drop of 3 points (at least 48 hours after admission), new hematoma, or other types of bleeding (Figure 3). Warfarin Exposure. Warfarin exposure was defined as patients who received wafarin before and during hospitalization and who had a documented international normalized ratio (INR) result during the hospital stay. Adverse events were defined as the occurrence of at least one of the following conditions within two days of a documented INR > 4.0 or abrupt cessation/hold of warfarin, administration of vitamin K, administration of FFP, or blood transfusion (absent a surgical procedure): death, cardiac arrest/emergency measure to sustain life, intracranial bleeding, gastrointestinal bleeding, genitourinary bleeding, pulmonary bleeding, hematocrit drop of 3 points (at least 48 hours after admission), new hematoma, or other types of bleeding (Figure 4). LMWH/Factor Xa Inhibitor Exposure. LMWH/Factor Xa inhibitor exposure was defined as patients who received either of these drugs during hospitalization. Adverse events were defined as the occurrence of at least one of the following conditions within one day of abrupt cessation/hold of LMWH/Factor Xa inhibitor, administration of protamine, administration of FFP, or blood transfusion (absent a surgical procedure): death, cardiac arrest/emergency measure to sustain life, intracranial bleeding, gastrointestinal bleeding, genitourinary bleeding, pulmonary bleeding, hematocrit drop of 3 points (at least 48 hours after admission), new hematoma, other types of bleeding (Figure 5). Exposure to Hypoglycemic Agents. Exposure to hypoglycemic agents was defined as patients who received insulins, oral hypoglycemics, or both, and who had a serum glucose result during the hospital stay. Adverse events were defined as the occurrence of a serum glucose level < 70 with one or more of the following conditions from the second day of admission through the day of discharge: death; cardiac arrest/emergency measure to sustain life; coma, loss of consciousness, or seizure; drowsiness, confusion, anxiety, or irritability; sweating, weakness, increased heart rate, or trembling; intravenous administration of 50% dextrose (D50) or glucagon; or oral administration of juice or sugar (Figure 6). For example, abstractors would look at the medication administration record to determine if there was documentation of D50, oral sugar, or juice administration.

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Determination of interrater accuracy is a key component of MPSMS internal quality control. Medical record abstraction accuracy was determined as the raw agreement rates of both the original abstractor and the reabstractor of an adjudicated gold standard datum and was found to be 96%–99% for identification of exposure and 94%–99% for identification of adverse events.

STATISTICAL ANALYSIS To compare patient demographics, clinical characteristics, and outcomes (mortality, readmission, and length of stay) of patients with ADEs and without ADEs we performed descriptive and bivariate analyses.16,24 The chi-square test was used for comparing dichotomous and categorical variables, and the t-test was used for comparing continuous variables. To assess the independent association of patient and clinical characteristics with ADEs, we used the hierarchical generalized linear modeling (HGLM) approach, modeling the log-odds of adverse events as a function of patient demographic and clinical variables. The HGLM approach was also used to assess the association between ADEs and outcomes with and without adjustment for patient characteristics. To account for withinstate correlation of the observed adverse events and outcomes and to separate within-state variation from between-state variation for calculation of 95% confidence intervals (CIs), all HGLMs were fitted with a random state-specific effect.24 We assigned the greater-than-85-year age category as referent for statistical analysis. We did not extend the analysis down to the hospital level because the number of cases in each hospital was too small to allow such an analysis. We estimated national ADE rates from sample ADE rates on the basis of weighted HGLM of the MPSMS sample data that accounted for the number of discharges from each state and the different probabilities of sample selection across states. To estimate the total number of Medicare discharges in which an ADE occurred, we first estimated the national exposure rate to each drug on the basis of the weighted HGLM of the MPSMS sample data.24 We then multiplied this rate by the total number of Medicare discharges to estimate the national number of drug exposures. Finally, we multiplied the estimated number of drug exposures by the estimated national ADE rates. We repeated this process for each ADE measure. All statistical analyses were conducted using STATA version 8.0 (STATA Corporation, College Station, TX) and SAS version 8.12 (SAS Institute Inc., Cary, NC). Hierarchical models were estimated using the GLIMMIX macro in SAS.24

January 2010

Results PATIENTS WITH MEDICATION EXPOSURE Among the 25,145 sample hospitalizations, 3,629 (14.4%) patients received warfarin and had INR testing; 2,244 (8.9%) patients received heparin and had PTT testing; 4,649 (18.5%) patients received LMWH/Factor Xa inhibitor; 7,065 (28.1%) patients received hypoglycemic agents and had blood glucose testing; 1,618 (6.4%) patients received digoxin and had serum digoxin level testing; and 15,543 (61.8%) patients received antibiotics.

ADVERSE EVENTS Among these patient groups, we identified 318 (8.8%) adverse events associated with warfarin; 327 (14.6%) adverse events associated with heparin; 449 (9.7%) adverse events associated with LMWH/Factor Xa inhibitor; 758 (10.7%) adverse events associated with hypoglycemic agents; 8 (0.5%) adverse events associated with digoxin; and 93 (0.6%) cases of CDI associated with antibiotics. Because so few adverse events associated with digoxin were identified, subgroup analyses and modeling of these events were not performed.

PATIENT CHARACTERISTICS With few exceptions, patients with adverse events were similar to those without adverse events for most age, sex, and race characteristics (Table 1a, page 16, and Tables 1b–1f, shown in Appendix 2, available in online article). Simultaneously controlling for the variables in Tables 1a–1f, patients receiving warfarin who were 65 to 74 years old and patients receiving heparin who were 65 years old or younger were less likely than other age groups to have had adverse events. Patients with CDI were less likely to be women, and patients with ADEs associated with hypoglycemic agents were less likely to be white.

COMORBID CONDITIONS Certain comorbid conditions were more common among patients with ADEs. Simultaneously controlling for other variables, renal disease was more common among patients with adverse events associated with anticoagulants, hypoglycemic agents, and antibiotics. Among patient receiving hypoglycemic agents, more than half of the patients with ADEs had documented renal disease, compared with fewer than one third of patients who did not have ADEs (risk difference, 18.5%). Among patients receiving antibiotics, more than half of patients with ADEs had documented renal disease, compared with just over one fifth of patients who did not have ADEs (risk difference, 28.8%). In addition to renal disease, coronary artery

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

15

The Joint Commission Journal on Quality and Patient Safety Table 1a. Characteristics of Patients Exposed to Warfarin, Medicare Patient Safety Monitoring System, 2004*

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroid use Coronary artery disease Renal failure

Patients With Adverse Events (n = 318) No. % 32 87 135 64 183 289 113 88 153 118 35 41 191 108

10.1 27.4 42.5 20.1 57.5 90.9 35.5 27.7 48.1 37.1 11.0 12.9 60.1 34.0

Patients Without Adverse Events (n = 3,311) No. % 313 1,116 1,364 518 1,870 2,995 1,062 924 1,370 998 366 248 1,607 710

9.5 33.7 41.2 15.6 56.5 90.5 32.1 27.9 41.4 30.1 11.1 7.5 48.5 21.4

Adjusted Odds Ratio† 0.8 0.6 0.8 1.0 1.1 1.1 1.0 1.2 0.9 1.0 1.7 1.0 1.5 1.7

95% CI

P Value

0.48–1.27 0.45–0.92 0.58–1.11

.321 .015 .188

0.87–1.41 0.71–1.63 0.75–1.24 0.94–1.56 0.68–1.16 0.79–1.31 1.19–2.46 0.66–1.44 1.18–1.95 1.34–2.26

.386 .727 .772 .142 .379 .897 .004 .896 .001 < .001

* CI, confidence interval. † Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstruc-

tive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

disease and smoking were more common in patients with ADEs associated with warfarin; diabetes and smoking were more common in patients with ADEs associated with heparin; diabetes, chronic obstructive pulmonary disease, and corticosteroids were more common in patients with ADEs associated with hypoglycemic agents; and diabetes was more common in patients with CDI who were receiving antibiotics.

TYPE, FREQUENCY, AND SEVERITY OF ADVERSE EVENTS The type, frequency, and severity of adverse events varied according to the class of medication that patients received. There were 18 (5.7%) patients who died, experienced cardiac arrest, or required emergency life-sustaining measures from ADEs associated with warfarin; 24 (7.3%) patients who died, experienced cardiac arrest, or required emergency life-sustaining measures from ADEs associated with heparin; 14 (3.1%) patients who died, experienced cardiac arrest, or required emergency life-sustaining measures from ADEs associated with LMWH/Factor Xa inhibitor; and 15 (2.0%) patients who died, experienced cardiac arrest, or required emergency life-sustaining measures from ADEs associated with hypoglycemic agents (Table 2a, page 17; Table 2b, page 17). One of the eight patients with an ADE associated with digoxin died, and no patients with antibiotic-associated CDI died, had a cardiac arrest, or required emergency life-sustaining measures. Two fifths of patients with ADEs associated with anticoagu16

January 2010

lants had documented bleeding (39.0% for warfarin, 40.1% for heparin, and 43.0% for LMWH/Factor Xa inhibitor; Table 2a). One-fifth (19.8%) of patients with ADEs associated with hypoglycemic agents had symptoms of hypoglycemia (Table 2b). More than half of patients with ADEs associated with anticoagulants had hematocrit drops of 3 points or more but did not have documented clinical bleeding. Nearly four fifths of patients with ADEs associated with hypoglycemic agents were treated with juice, D50, sugar, or glucagon but did not have documented clinical symptoms.

LENGTH OF STAY Patients with ADEs had longer lengths of stay than patients without ADEs, ranging from an additional 3.2 days for patients with ADEs associated with hypoglycemic agents to an additional 8.7 days for patients with CDI (Table 3, page 18). Simultaneously controlling for demographic characteristics and comorbid conditions and excluding patients with length of stay of greater than 60 days did not change the statistical significance of these findings, nor did log-transformation of hospital days (Table 3).

IN-HOSPITAL MORTALITY Patients with ADEs had higher in-hospital mortality than patients without ADEs, ranging from a risk difference of 3.3% for patients with ADEs associated with hypoglycemic agents to

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Table 2a. Type of Adverse Events Among Patients Exposed to Warfarin, Heparin, and Low Molecular Weight Heparin/Factor Xa Inhibitor, Medicare Patient Safety Monitoring System, 2004

Adverse Events* Deaths Documented bleeding or hematocrit drop† No documented bleeding or hematocrit drop Cardiac arrests/emergency measures Documented bleeding or hematocrit drop‡ No documented bleeding or hematocrit drop Documented bleeding Intracranial bleeding Pulmonary bleeding Gastrointestinal bleeding Genitourinary bleeding New hematoma only Other bleeding Bleeding from two sites Hematocrit drop of 3 points or more without documented bleeding Total patients with adverse events

Patients Receiving Warfarin No. %

Patients Receiving Receiving Heparin No. %

Patients Receiving LMWH/Factor Xa Inhibitor No. %

4 10

1.3 3.1

10 9

3.1 2.8

6 5

1.3 1.1

3 1

0.9 0.3

3 2

0.9 0.6

2 1

0.4 0.2

0 2 24 55 6 25 12 176 318

0.0 0.6 7.5 17.3 1.9 7.9 3.8 55.3 100.0

1 4 28 43 17 24 14 172 327

0.3 1.2 8.6 13.1 5.2 7.3 4.3 52.6 100.0

2 5 47 55 12 51 21 242 449

0.4 1.1 10.5 12.2 2.7 11.4 4.7 53.9 100.0

* Adverse event categories are hierarchical and are mutually exclusive. For example, if a patient experienced gastrointestinal bleeding with hematocrit drop of 3 points or more and then had a cardiac arrest and died, that patient would be categorized as a death. † No deaths with documented intracranial bleeding for warfarin and heparin but one death for LMWH/Factor Xa inhibitor. ‡ No cardiac arrests/emergency measures associated with documented intracranial bleeding.

Table 2b. Type of Adverse Events Among Patients Exposed to Insulin/Hypoglycemic Agents, Medicare Patient Safety Monitoring System, 2004

Adverse Events* Deaths with documented hypoglycemia Cardiac arrests/emergency measures after documented hypoglycemia Coma, loss of consciousness, or seizure Drowsiness, confusion, anxiety, or irritability Sweating, weakness, increased heart rate, or trembling without death/arrest (no deaths, arrests, or coma/LOC/Sz) Admin of D50, juice, glucagon, or sugar (without any of above symptoms or admin of glucagon) Total patients with adverse events

Patients Receiving Hypoglycemic Agents No. % 9 1.2 6 0.8 2 0.3 90 11.9 58 7.7 593 78.2 758 100.0

* Adverse event categories are arranged hierarchically and are mutually exclusive. For example, if a patient experienced sweating and confusion, was administered juice but lost consciousness, and had a cardiac arrest and died, that patient would be categorized as a death. LOC, loss of consciousness; Sz, seizure; D50, 50% dextrose.

9.8% for patients with ADEs associated with heparin (Table 4, page 19). After controlling for demographic characteristics and comorbid conditions, higher mortality rates among patients with ADEs persisted except for patients with CDI. Patients with ADEs had higher 30-day mortality than patients without ADEs, ranging from a risk difference of 14.7% for patients with CDI associated with antibiotics to 1.7% for patients with ADEs associated with hypoglycemic agents. Among patients

January 2010

receiving heparin, 30-day mortality was not significantly different. Controlling for demographic characteristics and comorbid conditions did not change the statistical significance of the 30day mortality findings. Patients with ADEs associated with LMWH/Factor Xa inhibitor and hypoglycemic agents had significantly higher 30-day readmission rates than patients without ADEs, but after controlling for demographic characteristics and comorbid conditions, only readmission rates among

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

17

The Joint Commission Journal on Quality and Patient Safety Table 3. Length of Stay for Patients With and Without Adverse Drug Events by Drug Exposure, Medicare Patient Safety Monitoring System, 2004* Patients With Adverse Events† Length of Stay (Days) 95% CI Warfarin Unadjusted Adjusted* Heparin Unadjusted Adjusted* LMWH/Factor Xa inhibitor Unadjusted Adjusted* Hypoglycemic agents Unadjusted Adjusted* Antibiotics Unadjusted Adjusted*

Patients Without Adverse Events Length of Stay (Days) 95% CI

Difference in Length of Stay (Days)

P Value‡

10.5 10.2

9.9-11.0 9.6-10.8

5.6 5.7

5.5-5.8 5.5-5.9

4.8 4.5

< .001 < .001

11.3 11.0

10.6-12.0 10.3-11.7

6.3 6.3

6.0-6.6 6.1-6.6

5.0 4.7

< .001 < .001

10.3 9.9

9.7-10.9 9.4-10.5

6.0 6.0

5.8-6.2 5.8-6.2

4.3 3.9

< .001 < .001

9.2 9.2

8.7-9.6 8.8-9.6

6.0 6.0

5.8-6.2 5.9-6.2

3.2 3.2

< .001 < .001

14.6 13.6

13.4-15.7 12.5-14.7

5.9 5.9

5.8-6.0 5.7-6.1

8.7 7.7

<.001 <.001

* CI, confidence interval; LMWH, low molecular weight heparin. † Adverse events to antibiotics only include Clostridium difficile infection (CDI). ‡ Difference in length of stay between patients with adverse events and patients without adverse events. § Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

patients receiving LMWH/Factor Xa inhibitor remained significantly higher.

NUMBER OF ADES In total, an estimated 888,000 ADEs associated with these high-risk medications occurred in hospitalized Medicare patients (Table 5, page 20). National estimates of adverse event rates were similar to the sample rates with the highest rate of adverse events from heparin (14.6%) and the lowest rate from digoxin (0.5%). Systemic anticoagulants were associated with an estimated 487,638 adverse events, and insulin/hypoglycemic agents were associated with a total of 356,163 adverse events.

DISCUSSION ADEs continue to be a major concern in ensuring the safety of hospitalized older patients. Chart review of a national sample of Medicare records which used explicit criteria to identify adverse events demonstrated that even though risks of adverse events from certain agents, such as anticoagulants and hypoglycemic agents, are well known, adverse events associated with these medications remain a common problem in older hospitalized patients. The risk factor and outcomes data collected from explicit review can be used to help direct safety inter18

January 2010

ventions. As interventions are made, surveillance using explicit review can monitor improvement. We found that anticoagulants were commonly used and were associated with a high rate of clinically significant adverse events in hospitalized older patients. More than 40% of older patients were treated with anticoagulation during their hospitalization. Nearly 1 in 11 patients treated with LMWH/Factor Xa inhibitor, nearly 1 in 10 patients treated with warfarin, and more than 1 in 7 patients treated with heparin experienced an associated adverse event during their hospitalization. The adverse events were clinically significant, with more than 45% involving both a process of care “trigger” or signal such as a very prolonged PTT and an event such as clinical bleeding, cardiac arrest, or death. Use of hypoglycemic agents was also common. Nearly 1 in 9 patients treated with hypoglycemic agents experienced an adverse event. However, far fewer of these events resulted in clinically significant injury. Nearly 80% of the patients with adverse events associated with hypoglycemic agents were asymptomatic and were treated only for low glucose readings. The rates of CDI among patients prescribed antibiotics (0.6%) and of digoxin-associated adverse events (0.5%) were less than one tenth the rate of other adverse events measured.

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Table 4. Mortality and Readmissions for Patients With and Without Adverse Drug Events by Drug Received, Medicare Patient Safety Monitoring System, 2004*

Outcomes Warfarin Mortality in-hospital Mortality within 30 days Readmission within 30 days Heparin Mortality in-hospital Mortality within 30 days Readmission within 30 days LMWH/Factor Xa inhibitor Mortality in-hospital Mortality within 30 days Readmission within 30 days Hypoglycemic agents Mortality in-hospital Mortality within 30 days Readmission within 30 days Antibiotics Mortality in-hospital Mortality within 30 days Readmission within 30 days

Patients With Adverse Events† No. %

Patients Without Adverse Events No. %

Risk Difference %

P Value

Adjusted Odds Ratio‡

95% CI

P Value

27 28 63

8.5 8.8 19.8

52 159 596

1.6 4.8 18.0

6.9 4.0 1.8

< .001 .005 .158

5.8 1.8 1.1

3.6–9.4 1.14–2.75 0.78–1.43

<.001 .011 .717

46 18 59

14.1 5.5 18.0

81 101 351

4.2 5.3 18.3

9.8 0.2 –0.3

<.001 .540 .458

2.9 1.0 1.1

1.93–4.32 0.6–1.69 0.77–1.46

< .001 .986 .711

40 38 100

8.9 8.5 22.3

134 224 730

3.2 5.3 17.4

5.7 3.1 4.9

< .001 .002 .001

2.4 1.5 1.4

1.63–3.42 1.03–2.16 1.07–1.75

< .001 .034 .011

59 54 185

7.8 7.1 24.4

284 343 1,278

4.5 5.4 20.3

3.3 1.7 4.1

< .001 .031 .002

1.8 1.4 1.2

1.34–2.47 1–1.85 0.99–1.43

< .001 .048 .067

8 19 21

8.6 20.4 22.6

656 881 2,579

4.3 5.7 16.8

4.3 14.7 5.8

.040 < .001 .085

1.4 3.2 1.2

0.65–2.81 1.91–5.5 0.72–1.97

.415 < .001 .492

* CI, confidence interval; LMWH, low molecular weight heparin. † Adverse events to antibiotics only include Clostridium difficile infection (CDI). ‡ Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

These measures may underestimate the true impact of antibiotic-associated adverse events and digoxin-associated adverse events, however. First, antibiotics were prescribed for hospitalized patients 50% more often than anticoagulants and twice as often as hypoglycemic agents. Second, although bleeding and hypoglycemia are the most common adverse events from anticoagulants and hypoglycemic agents, the most common adverse events from antibiotics are allergic reactions, which were not included in this study. Similarly, the measure of digoxin-associated adverse events included only cardiac manifestations, therefore excluding patients who only experienced neurological and gastrointestinal manifestations—which are more common. These data provide evidence supporting the extensive health burden associated with ADEs from high-risk medications in hospitalized patients. Patients with ADEs associated with all six medications had increased length of stay, in-hospital mortality, and 30-day mortality, except for patients with hospitalacquired antibiotic-associated CDI, who did not have increased in-hospital mortality, and patients with ADEs associated with heparin, who did not have increased 30-day mortality.

January 2010

Adjusting for age, sex, and preexisting conditions did not change the statistical significance of the associations with increased length of stay or mortality rates. However, these findings should be interpreted with caution. We did not control for admitting diagnoses or the severity of the acute illness. In addition, the association between ADEs and these outcomes does not prove ADEs caused these outcomes, and in some cases the reverse may be true. For example, among patients treated with antibiotics, the length of stay for patients with CDI was eight to nine days longer than for patients without CDI. From these data, we cannot determine if a longer length of hospitalization contributes to CDI or if CDI contributes to a longer length of stay, or if both types of situations occur. Our results supported earlier data that an increased number of comorbidities is associated with adverse events, although increased age did not appear to be associated with an increased risk of having an ADE. We did not identify medication errors or assess the preventability of ADEs, nor did we assign causality of the ADEs, so that these data do not suggest specific targets in the medication use process for interventions. On the other hand, these data may provide information to help target

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

19

The Joint Commission Journal on Quality and Patient Safety Table 5. National Estimates of Rates and Numbers of Adverse Drug Events by Drug, Medicare Patient Safety Monitoring System, 2004*

Drug Warfarin Heparin LMWH/Factor Xa inhibitor Hypoglycemic agents Systemic antibiotics Digoxin

Number of Discharges 3,629 2,244 4,649 7,065 15,437 1,618

Sample Cases Number of Adverse Events 318 327 449 758 93‡ 8

National Estimates† Adverse Event Rate (%) 8.2 13.6 9.6 10.8 0.6 0.1

95% CI 7.2-9.4 11.6-15.8 8.6-10.6 9.9-11.9 0.5-0.8 0.04-0.3

Number of Adverse Events 138,898 141,615 207,125 356,163 43,075 828

* CI, confidence internal; LMWH, low molecular weight heparin. † National adverse drug event (ADE) rates were calculated from sample cases on the basis of weighted hierarchical generalized linear modeling (HGLM) that accounted for the number of discharges from each state and the different probabilities of sample selection across states. To estimate the total number of Medicare discharges in which an ADE occurred, we first estimated the national exposure rate to each drug on the basis of the weighted HGLM of the sample data. We then multiplied this rate by the total number of Medicare discharges to estimate the national number of drug exposures. Finally, we multiplied the estimated number of drug exposures by the estimated national ADE rates. ‡

Adverse event includes only hospital-acquired, antibiotic-associated Clostridium difficile infection (CDI).

patients who might benefit from interventions. For example, renal disease was found to be an associated risk factor for all ADEs. Perhaps targeting prevention interventions among patients with impaired renal function would provide a significant reduction in adverse events, all the while using scarce resources as efficiently as possible. In addition, anticoagulation appears to be an appropriate area on which to focus safety improvement efforts because of the high rate and high severity of ADEs associated with anticoagulation. Insulin/hypoglycemic agents may also be an important area of focus for improvement, and adverse effects from these medications were found to be both common and severe in the outpatient setting as well.25 The IHI 5 Million Lives Campaign includes a focus on high-risk medications, and an approach similar to that used here could be applied at the hospital level to evaluate the impact of highrisk medication safety programs.23 We do not believe that repeating this type of surveillance for ADEs at the hospital level will likely be as productive or impactful as targeted intervention programs for high-risk medications at the hospital level. However, with the deployment of electronic health records, the ability to automate this type of surveillance may provide a potential sustainable approach at the hospital level.

LIMITATIONS A number of factors may contribute to underreporting of ADEs in the MPSMS. First, inclusion of ADEs is based solely on the medical record of the index hospitalization and Part A claims data. Because past and future hospital and outpatient 20

January 2010

records are unavailable, ADEs leading to hospitalization or ADEs occurring after hospital discharge are not identified. Second, identification of the event is based solely on medical record documentation. Clinical documentation by health care providers may fail to record data needed to identify adverse events by the MPSMS algorithms. Third, because data collected is based solely on abstraction according to the MPSMS algorithms and not review by drug safety experts, some events that might be considered ADEs by “implicit review” or clinical judgment will not be identified. Fourth, although the MPSMS algorithms were carefully considered to maximize positive predictive value, because clinical diagnoses were not used to identify ADEs, the MPSMS methods identify adverse events associated with medications but do not specifically identify causation for adverse events, mortality, readmission, or length of stay. Fifth, we did not screen for all ADEs, just those related to the medications of interest, and thus we were not able to estimate the overall rate of all ADEs in the Medicare population or the percentage of all ADEs that these specific drugs account for. Sixth, some patients younger than 65 years of age were included in the study, although it was a small part of the sample and the Medicare population. Finally, the data from this study were not available until a year after the study period, perhaps limiting their relevance, although most studies of drug safety are often published with greater latency than the MPSMS approach.

Conclusion Despite numerous national reports calling attention to the problem of ADEs among hospitalized patients, a significant

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety burden of injury remains. There continues to be a need to demonstrate an efficient and effective mechanism to identify the high-risk adverse events in an effort to improve medication safety. This study describes the application of a new method to detect ADEs in hospitalized Medicare patients. Not only can the findings can not only contribute national data to our understanding of the epidemiology of ADEs, but the surveillance methodology can also be used to create a national monitoring systems for trends in adverse events related to these drugs. This study focused on a limited number of high-risk drugs and adverse outcomes but can be expanded to include other drugs or other outcomes as determined by future drug safety studies. In addition, this method could be applied at the hospital or health system level to perform benchmarking analysis as one method to track the effectiveness of interventions to improve the safety of high-risk medications. J

David C. Classen, M.D., M.S., is Associate Professor, Division of Infectious Diseases, Department of Internal Medicine, University of Utah School of Medicine, Salt Lake City, and Senior Partner, CSCHealthcare, Salt Lake City. Lisa Jaser, R.Ph., Pharm.D., is Clinical Advisor, Qualidigm, Salt Lake City, and Pharmacist, Department of Pharmacy, Griffin Hospital, Derby, Connecticut. Daniel S. Budnitz, M.D., M.P.H., is Epidemiologist, Division of Healthcare Quality Promotion, National Center for Detection, Preparedness, and Control of Infectious Diseases, Coordinating Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta. Please address correspondence to David C. Classen, [email protected].

Online-Only Content

8

See the online version of this article for

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, Antibiotic-Associated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents Appendix 2. Characteristics of Patients Exposed to Heparin, Low Molecular Weight Heparin/Factor Xa Inhibitor, Insulin/Hypoglycemic Agents, Systemic Antibiotics, and Digoxin, Medicare Patient Safety Monitoring System

References 1. Institute of Medicine: Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press, 2001. 2. Institute of Medicine: Advancing Quality Improvement Research: Challenges and Opportunities: Workshop Summary. Chao S., Rapporteur, Forum on the Science of Health Care Quality Improvement and Implementation, Washington, DC: National Academy Press, Oct. 2007. 3. Institute of Medicine: Preventing Medication Errors. Washington, DC:

January 2010

National Academies Press, 2006. 4. U.S. General Accountability Office: Adverse Drug Events: The Magnitude of Health Risk Is Uncertain Because of Limited Incidence Data. Report to Congressional Requesters. Jan. 2000. http://www.ascp.com/advocacy/ briefing/upload/adegaoreport.pdf (last accessed Nov. 18, 2009). 5. Classen D.: Medication safety: Moving from illusion to reality. JAMA 289:1154–1156, Mar. 5, 2003. 6. Classen D.C., et al.: Computerized surveillance of adverse drug events in hospital patients. JAMA 266:2847–2851, Nov. 27, 1991. Erratum in: JAMA 267:1922, Apr. 8, 1992. 7. Rozich J.D., Haraden C.R., Resar R.K.: Adverse drug event trigger tool: A practical methodology for measuring medication related harm. Qual Saf Health Care 12:194–200, Jun. 2003. 8. Kilbridge P., Classen D.C.: Surveillance for Adverse Drug Events: History, Methods, and Current Issues. Research Series, Issue 3, pp. 1–35. Irving, TX: VHA, Inc., Apr. 2001. 9. Morimoto T., et al.: Adverse drug events and medication errors: Detection and classification methods. Qual Saf Health Care 13:306–314, Aug. 2004. 10. Seger A.C., Jha A.K., Bates D.W.: Adverse drug event detection in a community hospital utilizing computerized medication and laboratory data. Drug Saf 30:817–824, Sep. 2007. 11. Kopp B.J., et al.: Medication errors and adverse drug events in an intensive care unit: Direct observation approach for detection. Crit Care Med 34:415–425, Feb. 2006. 12. Field T.S., et al.: Strategies for detecting adverse drug events among older persons in the ambulatory setting. J Am Med Inform Assoc 11:492–498, Nov.–Dec. 2004. 13. Kilbridge P.M., Classen D.C.: Automated surveillance for adverse events in hospitalized patients: Back to the future. Qual Saf Health Care 15:148–149, Jun. 2006. 14. Bates, D.W., et al.: The costs of adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA 277:307–311, Jan. 22–29, 1997. 15. Eckman M.H., et al.: Making decisions about antithrombolytic therapy in heart disease: Decision analytic and cost-effectiveness issues. Chest 114:699–714, Nov. 1998. 16. Classen D.C., et al.: Adverse drug events in hospitalized patients: Excess length of stay, extra costs, and attributable mortality. JAMA 277:301–306, Jan. 22–29, 1997. 17. Lazarou J., Pomeranz B.H., Corey P.N.: Incidence of adverse drug reactions in hospitalized patients: A meta-analysis of prospective studies. JAMA 279:1200–1205, Apr. 15, 1998. 18. Leape L.L., et al., for the ADE Prevention Group: Systems analysis of adverse drug events. JAMA 274:35–43, Jul. 5, 1995. 19. Thomas E.J., Brennan T.A.: Incidience and types of preventable adverse events in elderly population-based review of medical records. BMJ 320:741–744, Mar. 18, 2000. 20. Budnitz D.S., et al.: Medication use leading to emergency department visits for adverse drug events in older adults. Ann Intern Med 147:755–765, Dec. 4, 2007. 21. Smetzer J.L., et al.: Findings from the ISMP Medication Safety SelfAssessment for hospitals. Jt Comm J Qual Saf 29:586–597, Nov. 2003. 22. The Joint Commission: The Joint Commission announces the 2008 National Patient Safety Goals and Requirements. Jt Comm Perspect 27:1, 9–22, Jul. 2007. 23. Hackbarth A.D., et al.: The hard count: Calculating lives saved in the 100,000 Lives Campaign. ACP Guide for Hospitalists, pp. 1–5, Apr. 2006. http://www.ihi.org/IHI/Topics/Improvement/SpreadingChanges/Literature/ ThehardcountCalculatinglivessavedinthe100000LivesCampaign.htm (last accessed Nov. 19, 2009). 24. Hunt D.R., et al.: Fundamentals of Medicare patient safety surveillance: Intent, relevance, and transparency. Advances in Patient Safety: From Research to Implementation, vol. 2, Feb. 2005. http://www.ahrq.gov/downloads/pub/ advances/vol2/Hunt.pdf (last accessed Nov. 18, 2009). 25. Budnitz D.S., et al.: National surveillance of emergency department visits for outpatient adverse drug events. JAMA 296:1858–1866, Oct. 18, 2006.

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

21

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents

Figure 1. Algorithm for Identifying Adverse Events Associated with Digoxin*

No

Digoxin administered during hospital stay

Patient excluded

Yes

Digoxin level obtained during hospital stay Denominator

No

Patient excluded

Yes

Digoxin level > 2.5 mg/ml

No

Digoxin level No > 1.5 mg/ml but < 2.5 mg/mL

Patient excluded

Yes Yes

Potassium < 3.5 mEq/L

No

Patient excluded

Yes

The patient experienced at least one of the following events the same date of the above digoxin/potassium level (incidences of adverse events that occurred on the date of arrival were not counted): Death Cardiac arrest/emergency measure to sustain life Ventricular tachycardia AV block Bradycardia Premature ventricular contraction PR prolongation Ventricular fibrillation Administration of Digibind (Digoxin immune FAB)

No

Yes

At least one adverse drug event associated with digoxin during hospital stay. * AV, atrioventricular.

AP1

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

Patient excluded

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents (continued)

Figure 2. Algorithm for Identifying Hospital-Acquired, Antibiotic-Associated Clostridium difficile Infections (CDIs)

All hospital discharges

No

C. difficile toxin assay was ordered during hospital stay.

Patient excluded

Yes

C. difficile toxin assay was positive.

No

Patient excluded

Yes

The assay was positive for C. difficile toxin > 2 days after arrival.

No

Patient excluded

Yes

Antibiotic administered during hospital stay prior to C. difficile toxin assay.

No

Patient excluded

Yes

Hospital-acquired, antibiotic-associated CDI

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

AP2

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents (continued)

Figure 3. Algorithm for Identifying Adverse Events Associated with Heparin* IV heparin administered during hospital stay.

Patient excluded

Yes No

PTT obtained during hospital stay. Denominator

Patient excluded

Yes No

PTT > 45 sec. during the hospital stay

Patient excluded

Yes

Patient experienced at least one of the following trigger events: PTT ≥ 100 sec Abrupt cessation/hold of heparin, Protamine, Fresh Frozen Plasma, or Transfusion (Absent a surgical procedure)

No

Patient excluded

Yes

The patient experienced at least one of the following events within 1 day before and 1 day after a “trigger” event. (Incidences of adverse events that occurred on the date of arrival were not counted). Death Cardiac arrest/emergency measure to sustain life Intracranial bleed (subdural hematoma) Gastrointestinal bleeding Genitourinary bleeding Pulmonary bleeding Hematocrit drop of 3 points (at least 48 hours after admission) New hematoma Other types of bleeding

No

Patient excluded

Yes

At least one adverse drug event associated with heparin during hospital stay. * IV, intravenous; PTT, partial thromboplastin time.

AP3

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents (continued)

Figure 4. Algorithm for Identifying Adverse Events Associated with Warfarin* Warfarin administered during hospital stay.

No

Patient excluded

Yes No

INR obtained during hospital stay. Denominator

Patient excluded

Yes No

INR > 1.5 during the hospital stay.

Patient excluded

Yes

Patient experienced at least one of the following trigger events: INR > 4.0 Abrupt cessation/hold of warfarin, Vitamin K, Fresh Frozen Plasma, or Transfusion (Absent a surgical procedure)

No

Patient excluded

Yes

The patient experienced at least one of the following events within 2 days before and 2 days after a “trigger” event. (Incidences of adverse events that occurred on the date of arrival were not counted.) Death Cardiac arrest/emergency measure to sustain life Intracranial bleed (subdural hematoma) Gastrointestinal bleeding Genitourinary bleeding Pulmonary bleeding Hematocrit drop of 3 points (at least 48 hours after admission) New hematoma Other types of bleeding

No

Patient excluded

Yes

At least one adverse drug event associated with warfarin during hospital stay. * INR, international normalized ratio.

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

AP4

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents (continued)

Figure 5. Algorithm for Identifying Adverse Events Associated with Low Molecular Weight Heparin (LMWH)/Factor Xa Inhibitors

LMWH/Factor Xa inhibitor administered during hospital stay. Patient excluded Denominator

Yes

Patient experienced at least one of the following trigger events: Abrupt cessation/hold of LMWH/Factor Xa Inhibitor, Protamine, Fresh Frozen Plasma, or Transfusion (Absent a surgical procedure)

No

Patient excluded

Yes

The patient experienced at least one of the following events within 1 day before and 1 day after a “trigger” event. (Incidences of adverse events that occurred on the date of arrival were not counted.) Death Cardiac arrest/emergency measure to sustain life Intracranial bleed (subdural hematoma) Gastrointestinal bleeding Genitourinary bleeding Pulmonary bleeding Hematocrit drop of 3 points (at least 48 hours after admission) New hematoma Other types of bleeding

No

Patient excluded

Yes

At least one adverse drug event associated with LMWH/Factor Xa inhibitor during hospital stay.

AP5

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 1. Algorithms for Identifying Adverse Events Associated with Digoxin; Hospital-Acquired, AntibioticAssociated Clostridium difficile Infections (CDIs); Heparin; Warfarin; Low Molecular Weight Heparin/Factor Xa Inhibitor; and Insulin/Hypoglycemic Agents (continued)

Figure 6. Algorithm for Identifying Adverse Drug Events Associated with Insulin/ Hypoglycemic Agents* Insulin, oral hypoglycemic agents, or a combination of both administered during hospital stay.

No

Patient excluded

Yes

Blood glucose level obtained during hospital stay. Denominator

No

Patient excluded

Yes

Blood glucose was ≤ 70 mg/dL

No

Patient excluded

Yes

The day the blood glucose was ≤ 70 mg/dL the patient experienced at least one of the following events (incidences of adverse events that occurred on the date of arrival were not counted): Death Cardiac arrest/emergency measure to sustain life Coma or loss of consciousness Myocardial infarction Stroke TIA Seizure Anxiety Confusion Drowsiness Increased heart rate Irritability Sweating Trembling Weakness Administration of D50 Administration of glucagon Administration of juice and/or sugar

No

Patient excluded

Yes

At least one adverse drug event associated with insulin/oral hypoglycemic agents or a combination of both during the hospital stay. * TIA, transient ischemic attack; D50, 50% dextrose.

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

AP6

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 2. Characteristics of Patients Exposed to Heparin, Low Molecular Weight Heparin/Factor Xa Inhibitor, Insulin/Hypoglycemic Agents, Systemic Antibiotics, and Digoxin, Medicare Patient Safety Monitoring System

Table 1b. Characteristics of Patients Exposed to Heparin, Medicare Patient Safety Monitoring System, 2004*

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroid use Coronary artery disease Renal failure

Patients With Adverse Events (n = 327) No. % 35 117 134 41 156 281 118 102 151 111 64 33 251 122

10.7 35.8 41.0 12.5 47.7 85.9 36.1 31.2 46.2 33.9 19.6 10.1 76.8 37.3

Patients Without Adverse Events (n = 1,917) No. % 318 738 664 197 897 1,623 701 567 686 549 347 129 1,343 580

16.6 38.5 34.6 10.3 46.8 84.7 36.6 29.6 35.8 28.6 18.1 6.7 70.1 30.3

Adjusted Odds Ratio†

95% CI

P Value

0.5 0.8 1.0 1.0 1.0 1.0 1.3 1.1 1.0 0.9 1.6 1.3 1.3 1.4

0.29–0.81 0.51–1.14 0.66–1.44 — 0.8–1.30 0.7–1.44 1.03–1.71 0.86–1.45 0.78–1.30 0.72–1.21 1.05–2.38 0.95–1.81 0.99–1.75 1.05–1.78

.01 .19 .91 — .84 .98 .03 .42 .96 .59 .03 .10 .06 .02

* CI, confidence interval. † Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

Table 1c. Characteristics of Patients Exposed to Low Molecular Weight Heparin/Factor Xa Inhibitor, Medicare Patient Safety Monitoring System, 2004*

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroid use Coronary artery disease Renal failure

Patients With Adverse Events (n = 327) No. % 51 123 189 86 252 385 153 118 198 169 67 40 262 156

11.4 27.4 42.1 19.2 56.1 85.7 34.1 26.3 44.1 37.6 14.9 8.9 58.4 34.7

Patients Without Adverse Events (n = 1,917) No. % 569 1,374 1,562 695 2,455 3,592 1,413 984 1,457 1,403 614 319 2,185 873

13.5 32.7 37.2 16.5 58.5 85.5 33.6 23.4 34.7 33.4 14.6 7.6 52.0 20.8

Adjusted Odds Ratio†

95% CI

P Value

0.8 0.8 1.0 1.0 1.0 1.0 1.2 1.1 1.1 0.9 1.1 1.1 1.1 1.9

0.51–1.11 0.56–1.01 0.75–1.30 — 0.78–1.17 0.75–1.34 0.97–1.48 0.87–1.34 0.84–1.32 0.71–1.10 0.8–1.60 0.83–1.49 0.92–1.40 1.53–2.38

.156 .062 .942 .659 .987 .099 .469 .632 .271 .492 .469 .223 < .001

* CI, confidence interval. † Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

AP7

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 2. Characteristics of Patients Exposed to Heparin, Low Molecular Weight Heparin/Factor Xa Inhibitor, Insulin/Hypoglycemic Agents, Systemic Antibiotics, and Digoxin, Medicare Patient Safety Monitoring System (continued)

Table 1d. Characteristics of Patients Exposed to Insulin/Hypoglycemic Agents, Medicare Patient Safety Monitoring System, 2004*

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroid use Coronary artery disease Renal failure

Patients With Adverse Events (n = 758) No. % 164 243 265 86 405 570 720 233 380 250 508 82 448 383

21.6 32.1 35.0 11.3 53.4 75.2 95.0 30.7 50.1 33.0 67.0 10.8 59.1 50.5

Patients Without Adverse Events (n = 6,307) No. % 1,149 2,156 2,221 781 3,438 5,071 5,435 1,739 2,453 2,100 4,207 580 3,460 2,020

18.2 34.2 35.2 12.4 54.5 80.4 86.2 27.6 38.9 33.3 66.7 9.2 54.9 32.0

Adjusted Odds Ratio†

95% CI

P Value

1.2 1.0 1.1 1.0 1.0 0.8 1.3 0.9 1.1 2.9 1.2 1.3 1.0 1.9

0.89–1.60 0.79–1.34 0.85–1.43 — 0.82–1.12 0.68–0.99 1.14–1.58 0.79–1.11 0.91–1.27 2.04–4.00 0.97–1.60 1.03–1.60 0.86–1.19 1.65–2.27

.226 .857 .450 .599 .041 .001 .427 .388 < .001 .086 .027 .871 < .001

* CI, confidence interval. † Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

Table 1e. Characteristics of Patients Exposed to Systemic Antibiotics, Medicare Patient Safety Monitoring System, 2004*

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroid use Coronary artery disease Renal failure

Patients With Hospital- Patients Without HospitalAcquired AntibioticAcquired AntibioticAssociated Clostridium Associated Clostridium Adjusted difficile (n = 93) difficile (n = 15,344) Odds No. % No. % Ratio† 12 22 40 19 42 80 43 30 47 34 15 12 47 47

12.9 23.7 43.0 20.4 45.2 86.0 46.2 32.3 50.5 36.6 16.1 12.9 50.5 50.5

2,379 4,870 5,451 2,644 8,712 13,125 4,914 3,433 4,512 5,174 2,461 1,343 6,222 3,334

15.5 31.7 35.5 17.2 56.8 85.5 32.0 22.4 29.4 33.7 16.0 8.8 40.6 21.7

0.7 0.7 1.0 1.0 0.7 1.3 1.7 0.9 1.3 1.4 1.5 1.2 0.9 2.8

95% CI

P Value

0.33–1.27 0.39–1.16 0.63–1.64 — 0.46–0.95 0.77–2.27 1.12–2.45 0.61–1.32 0.89–1.91 0.96–2.02 0.87–2.57 0.74–2.06 0.61–1.30 1.93– 4.13

.206 .157 .931 .024 .305 .012 .575 .181 .079 .141 .416 .540 < .001

* CI, confidence interval. † Adjusted for age (continuous), race (white or non-white), gender, diabetes, cerebrovascular disease, congestive heart failure/pulmonary edema, chronic obstructive pulmonary disease, smoking status, corticosteroid use, coronary artery disease, and renal failure. Cases with length of stay > 60 days excluded.

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

AP8

The Joint Commission Journal on Quality and Patient Safety Online-Only Content

8

Appendix 2. Characteristics of Patients Exposed to Heparin, Low Molecular Weight Heparin/Factor Xa Inhibitor, Insulin/Hypoglycemic Agents, Systemic Antibiotics, and Digoxin, Medicare Patient Safety Monitoring System (continued)

Table 1f. Characteristics of Patients Exposed to Digoxin, Medicare Patient Safety Monitoring System, 2004* Patients With Adverse Events (n = 8) No. %

Characteristics Age: < 65 65–74 75–84 85 and over Female White Diabetes Cerebrovascular disease Congestive heart failure/pulmonary edema Chronic obstructive pulmonary disease Smoking Corticosteroids Coronary artery disease Renal disease

0 4 2 2 4 8 3 0 5 4 1 1 4 3

0.0 50.0 25.0 25.0 50.0 100.0 37.5 0.0 62.5 50.0 12.5 12.5 50.0 37.5

Patients Without Adverse Events (n = 1,610) No. % 103 375 695 437 888 1,412 601 489 1,100 692 201 166 1,023 480

* Adjusted Odds Ratio not calculated because of small numbers.

AP9

January 2010

Volume 36 Number 1

Copyright 2010 Joint Commission on Accreditation of Healthcare Organizations

6.4 23.3 43.2 27.1 55.2 87.7 37.3 30.4 68.3 43.0 12.5 10.3 63.5 29.8