Raporlanmış önlenebilir advers ilaç reaksiyonlarına eşlik eden faktörler…

…

Authors and Disclosures

Robert D Beckett PharmD BCPS, Clinical Assistant Professor of Pharmacy Practice, School of Pharmacy, Manchester College, Fort Wayne, INAmy Heck Sheehan PharmD, Associate Professor of Pharmacy Practice, College of Pharmacy, Purdue University, West Lafayette, INJennifer G Reddan PharmD FASHP, Director, Center for Medication Management, Indiana University Health, Indianapolis, IN; and, during the study period, Interim Coordinator for Medication Safety, Indiana University HealthConflict of interestAuthors reported noneCorrespondenceDr. Beckett, [email protected]

From The Annals of Pharmacotherapy

Factors Associated With Reported Preventable Adverse Drug Events

A Retrospective, Case-control Study

Robert D Beckett PharmD BCPS; Amy Heck Sheehan PharmD; Jennifer G Reddan PharmD FASHP

Posted: 05/28/2012; The Annals of Pharmacotherapy. 2012;46(5):634-641. © 2012 Harvey Whitney Books Company

Abstract and Introduction

Abstract

Background: It has been reported that error occurs at some point during the medication use process in approximately 6% of medication doses administered in the inpatient setting. An estimated 1–10% of medication errors lead to patient harm; however, factors affecting the risk of harm from a medication error are undefined in the literature.Objective: To identify independent factors affecting the risk of reported preventable adverse drug events (ADEs) (ie, medication errors contributing to patient harm) compared to medication errors that did not contribute to patient harm in a diverse patient population.Methods: This was a retrospective, case-control study conducted at 3 hospitals within a large health system. Medication error reports from July 1, 2009, through June 30, 2010, were assessed. All reported medication errors determined to have contributed to patient harm were matched 1:1 with a medication error that did not contribute to harm. Data collected through review of the incident report and medical record included patient, provider, medication, and other related factors. Multivariable logistic regression was used to determine the relationship of potential factors to patient harm.Results: Of 4321 medication errors reported at study sites, 182 (4%) contributed to patient harm. Factors associated with increased independent risk of harm were 30-day readmission, time of day 0300–0659, and Institute for Safe Medication Practices (ISMP) high-alert medications. Factors associated with decreased independent risk of harm were multiple medication errors, occurrence during February or April, dispensing errors, and pharmacist review of medication order.Conclusions: Health systems should develop programs to promote safe, conscientious use of ISMP high-alert medications, promote pharmacist review, control the use of cabinet overrides, and direct provider attention toward recently admitted patients. Efforts should be made to determine factors associated with risk of harm at local levels.

Introduction

The prevalence of medication errors in inpatient settings is a significant international concern.^[1] Because of a general reliance on voluntary incident reporting to quantify medication safety metrics, precisely how many medication errors occur in hospitals is unknown. However, it has been reported that an error has occurred at some point during the medication use process in approximately 6% of medication doses (from a median of 5% in academic hospitals up to 14% in community hospitals).^[1–4] Considering the number of medication orders written during an inpatient hospital stay, the potential for error is considerable.

Adverse drug events (ADEs) may be defined as “an injury from a medication (or lack of intended medication).”^[5] In the US alone, ADEs are implicated in 7000 deaths, at an estimated cost of $2 billion annually.^[1] When an ADE is related to a medication error on the part of any health care professional, it is considered preventable.^[6] The percentage of medication errors that have been found to contribute to patient harm has been estimated to be from less than 1% to 10%.^{[1–4,7–10]} The effects of preventable ADEs can range from transient morbidity to mortality.

A number of factors are expected to increase the risk of a medication error becoming an ADE, including the phase of the medication use process (eg, prescribing errors) in which the error occurs.^[7,8] Despite the potential adverse outcomes associated with preventable ADEs, independent risk factors for these harmful medication errors, as opposed to those not associated with harm, are fairly undefined in the literature.^[1–10] The objective of this study was to review medication error data from a large health system to identify independent factors affecting the risk of reported preventable ADEs (ie, medication errors contributing to patient harm) compared to medication errors that did not contribute to patient harm.

Methods

This was a retrospective, case-control study approved by the local institutional review board (IRB) in October 2010. It was exempted from full IRB review, and informed consent was not required.

Research was conducted at Indiana University Health, Indianapolis, IN. Three hospitals from this 15-hospital health system were selected for this study. Hospitals were chosen for the diversity and acuity of the patient population, consistency of medication error review procedures during the reporting time, and data availability. The first facility was a large community teaching hospital (approximately 23,000 monthly patient days and 750 beds). The second was a medium-sized academic teaching hospital (approximately 14,000 monthly patient days and 300 beds). Lastly, we included our academic children’s hospital (approximately 9500 monthly patient days and 250 beds).

A file detailing all medication errors reported through a local voluntary electronic incident-reporting system at the 3 hospitals from July 1, 2009, through June 30, 2010, was obtained. The institution categorizes medication error severity using the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) Index (Table 1).^[11] All NCC MERP Severity E through I medication errors (ie, preventable ADEs) were included in the study as cases. Cases were matched 1:1 with an NCC MERP Severity B through D medication error control that occurred during the study time period. There were no additional matching criteria. Controls were selected using a random number generator. NCC MERP Severity A medication errors were excluded from the study because of the possibility that such events would not necessarily be related to a specific patient. There were no additional exclusion criteria.

[ CLOSE WINDOW ]

Table 1. Cases and Controls Included by the NCC MERP Index¹¹

NCC MERP = National Coordinating Council for Medication Error Reporting and Prevention.

Data collected were determined through review of similar studies and peer input^[4–6] and were divided into 4 types of factors: medication, patient, provider, and other (Table 2). All data entries were considered to potentially affect risk of patient harm. Data were collected through review of the original incident report completed by nursing, pharmacy, and risk management staff, as well as the electronic medical record for each patient. Root cause was defined as the major event or situation contributory to the error (eg, medications not appropriately reconciled, incorrect order entry in pharmacy). Laboratory values collected were from the time of admission. If this information was unavailable, the most recent value within 1 month prior to admission was recorded. For the purposes of this study, a patient was defined as experiencing multiple errors if at least 1 additional error was reported for the patient during the reporting period. Information included in the original incident report (eg, severity, root cause) was not altered when collected. All review was performed by a pharmacist (RDB) with experience in reviewing medication error incident reports; approximately 10–20 minutes were spent reviewing each medication error.

[ CLOSE WINDOW ]

Table 2. Data Collected

ALT = alanine aminotransferase; AST = aspartate aminotransferase; CrCl = creatinine clearance: ISMP = Institute for Safe Medication Practices.

Baseline characteristics were described using median with interquartile range for continuous variables and number with percentages for categorical variables. Univariate logistic regression was used to identify which risk factors were potentially associated with patient harm using individual incident reports as the unit of analysis (ie, multiple errors associated with the same medication order were analyzed separately). All potential factors were entered into the univariate analysis using categorical data forms. All factors suggesting potential for interaction (p < 0.1) based on the univariate analysis were evaluated for inclusion in the multivariate logistic regression model; those that were found to have sufficient entries in the data categories were included. Odds ratios and 95% confidence intervals were then calculated to determine which of the factors entered into multivariable analysis were independently associated with patient harm.

Results

A total of 4321 medication errors were reported during the study period. Of these, 553 (13%) were rated as Severity A errors and were excluded. There were 182 (4%) Severity E through I errors, all of which were included as cases (ie, preventable ADEs). Of the remaining 3586 (83%) medication errors rated as Severity B through D, 182 were randomly selected to be included as controls. The 364 errors reviewed affected a total of 340 patients. Table 1 provides a breakdown of the cases and controls included in the study by severity level.

[ CLOSE WINDOW ]

Table 1. Cases and Controls Included by the NCC MERP Index¹¹

NCC MERP = National Coordinating Council for Medication Error Reporting and Prevention.

Cases and control patients differed in some key characteristics: control group patients were younger than cases (median age 48 vs 54 years), with a higher percentage of adults (55% vs 44%) and a lower percentage of geriatric patients (18% vs 37%). Additionally, there was a lower percentage of female patients in the control group (44% vs 53%). Control patients were less likely to have been readmitted within 30 days of a previous hospitalization (21% vs 31%) but were more likely (54% vs 37%) to have experienced more than 1 medication error within the study time frame. No major differences were noted in number of medications (median 16), patient length of stay at the time of error (median 3 days), or primary problem treated. The most common problems were infectious (17%), cardiovascular (15%), and hematologic/oncologic (13%). Table 3 provides a further breakdown of patient-related factors.

[ CLOSE WINDOW ]

Table 3. Patient Factors

ALT = alanine aminotransferase; AST = aspartate aminotransferase; CrCl = creatinine clearance.^a Median (interquartile range).

There were several key differences between the cases and controls in terms of provider-related factors. The time of occurrence of the medication error was fairly evenly distributed in the cases but not in the controls; there were no significant differences between controls and cases in terms of day of the week. The overall percentage of errors reported at each site was roughly proportional to the monthly patient days and bed size ratio among the 3 hospitals. The mix of individual nursing unit types was generally similar between groups. A greater percentage of dispensing errors (38% vs 14%) and a smaller percentage of administration errors (36% vs 54%) were reported in the control group relative to the cases. Similarly, errors attributed to nursing staff were less common in the control group (34% vs 60%), while errors attributed to pharmacy staff were more common (35% vs 15%). The proportion of wrong-time errors was higher in the control group (14% vs 3%), while the proportion of wrong-drug errors was lower (5% vs 11%). Table 4 describes the complete results for provider-related factors.

[ CLOSE WINDOW ]

Table 4. Provider Factors

The most common medication classes for which errors were reported were cardiovascular, opioid, endocrine, and antiinfective; endocrine class errors primarily involved insulin. All of these medications, except antiinfectives, were prescribed more commonly in the cases. Overall, errors due to Institute for Safe Medication Practices (ISMP) high-alert medications were more common in the cases than in the controls (72% vs 42%). Errors in injectable routes of administration (ie, intravenous infusion and injection, sub-cutaneous injection) were more common in the cases, while oral medications were more commonly implicated in controls. Table 5 further breaks down the results for medication-related factors. Additionally, the root causes of medication error were highly variable overall but similar between groups; the most common were technology failures (11%), nursing staff inexperience (11%), medication reconciliation errors (9%), technology overrides (9%), and pharmacist order entry mistakes (9%).

[ CLOSE WINDOW ]

Table 5. Medication Factors

ISMP = Institute for Safe Medication Practices.

When the data were entered into univariate analysis to assess for a relationship to patient harm, the following factors had a p value of less than 0.1, indicating the potential for interaction, and were entered into the multivariate analysis: adults, geriatric patients, females, creatinine clearance less than 30 mL/min, readmission within 30 days, more than 1 medication error, time of day 2300–0259 and 0300–0659, error during February or April, dispensing errors, ISMP high-alert medications, and pharmacist review of the medication order. Medication class, route, and frequency also had univariate results suggesting interaction but were unevaluable in the multivariate model because of the low number of entries in some subcategories. Table 6 describes the results of the multivariate analysis. Factors associated with an increased independent risk of patient harm were 30-day readmission (OR 1.95; 95% CI 1.08 to 3.49), time of day 0300–0659 (OR 4.30; 95% CI 1.89 to 9.81), and ISMP high-alert medications (OR 4.00; 95% CI 2.38 to 6.75). Factors associated with a decreased independent risk of patient harm were experience with more than 1 medication error (OR 0.37; 95% CI 0.22 to 0.63), errors occurring during February (OR 0.42; 95% CI 0.18 to 0.96) or April (OR 0.16; 95% CI 0.05 to 0.44), dispensing errors (OR 0.25; 95% CI 0.14 to 0.46), and pharmacist review of the medication order (OR 0.54; 95% CI 0.29 to 0.99).

[ CLOSE WINDOW ]

Table 6. Multivariate Analysis

CrCl = creatinine clearance; ISMP = Institute for Safe Medication Practices.

Discussion

Our study suggests a number of potential factors that could affect the risk for reported preventable ADEs and that several of these independently increase or decrease that risk. The fact that so many of the potential factors did not independently affect the risk for patient harm (Table 6) underscores the multifactorial nature of medication safety and the complexity of medication error. For example, although the univariate analysis suggested a strong impact of certain factors (eg, geriatric age and decreased creatinine clearance) on patient harm, many were not independently noted to increase risk when impact of all factors was considered.

[ CLOSE WINDOW ]

Table 6. Multivariate Analysis

CrCl = creatinine clearance; ISMP = Institute for Safe Medication Practices.

There was no significant difference between groups in terms of patient length of stay at the time of the error or number of medications per patient; this may be because of the academic medical setting and the complexity of most patients. When assessing individual patient risk, we also did not identify a particular population as being at higher (or lower) risk for harm due to medication error with regard to the primary problem or reason for admission. However, the impact of 30-day readmissions on the risk for patient harm, regardless of the primary problem, was striking and underscores the importance of this metric from a clinical and reimbursement standpoint. As organizations expand clinical services provided by the pharmacy department, a targeted focus on patients readmitted following a recent hospital stay may aid in prevention of harmful medication errors. Effort should also be made to minimize the rate of 30-day readmissions.

We found that the timing of the medication error had little impact on harm. The exception to this finding was the increased risk of harm suggested for errors that occurred between 0300 and 0659. While it is theoretically possible that errors occurring during early-morning hours are more harmful, overnight staff may simply be more likely to report harmful errors than nonharmful ones. This finding highlights the criticality of education regarding medication error reporting and the importance of reporting near misses and nonharmful medication errors. When institutions are promoting the practice of medication error reporting, it is vital to target individuals who may not routinely interact with medication safety experts (eg, night shift workers and those in ancillary or outpatient areas). Education regarding medication error reporting should include an overview of the process as well as a description of why this practice is important and the value it brings to the patients and the organization. It is more difficult to explain the findings regarding error reporting in February and April; this study was conducted in a very large health system and it is possible that there were unaccounted trends in error reporting that also affected these results.

The impact of pharmacists on medication safety is highlighted in the provider factor results; the decreased risk of harm found with reported dispensing errors suggests that pharmacists in our system were able to prevent, or identify and correct, significant dispensing errors. Conversely, the dispensing process may benefit from the security of an additional layer of protection (ie, administration) prior to the medication reaching the patient. There was a significant decrease in the risk of patient harm when a pharmacist reviewed a medication order prior to administration. This underscores the importance of pharmacy involvement in the medication use process.

One limitation of our study is that, as an organization, we do not require medication error reporters to disclose their profession and chose not to collect this information. A theoretical bias could limit the applicability of these results if most errors were reported by pharmacists; however, when we reviewed data from the study time period for those who did self-identify, we found that, overall, 55% of medication errors were reported by nurses compared to 39% by pharmacists, 1% by physicians, and 5% by other groups.

Our study also suggests implications pertaining to the use of override functionality of automated dispensing cabinets. Based on the identified impact of pharmacists, override lists should be limited to emergency medications for which a time delay due to pharmacist review would be dangerous to a patient; institutions should routinely assess their override lists, examining both local and national data. Additionally, pharmacy departments should assess their distribution process and technology to ensure that medications are not available for use prior to pharmacist approval of a medication order via other ways (eg, nonsecure medication cabinets).

We found that, contrary to our expectations, patients who experienced more than 1 medication error during the reporting period were less likely to be harmed. The most likely reason for this is that the root cause in the majority of these errors was usually identified as medication reconciliation. At our institution, failures in medication reconciliation usually affected several medications (eg, medication reconciliation was not performed, the wrong patient’s list was used), but in a manner that did not translate to patient harm.

We did not identify any particular medication class, route, frequency, or root cause independently affecting risk for harm during medication error; however, as in the case with individual error types, it is likely that there were too few patients in each group to detect meaningful impact in these areas. Despite this fact, emphasis on medication safety programs for insulin, injectable routes of administration (ie, intravenous infusion and injection, subcutaneous injection), and continuously administered medications is warranted. An important related finding was the dramatic increased risk for harm associated with ISMP high-alert medications as a whole; conscientious use of these medications should be emphasized in medication safety programs, organizational metrics, and pharmacy operations processes. Additionally, it is vital that personal responsibility be taken by individual practitioners when ordering, dispensing, and administering these medications to prevent harmful medication errors.

Our results differed somewhat from those of a similar study. Zaal et al. conducted a 2-way case-control study to assess risk factors for harm from medication errors in a general medicine inpatient population.^[7] They reviewed 5724 medication errors from 592 patient admissions and found that 102 (1.8%) contributed to patient harm. The authors did not establish any independent risk factors for harm or no harm; however, the study assessed a homogenous population and only 102 errors. In our study, we assessed a wider variety and larger number of patients, which may explain why we were able to establish several independent factors. A greater number and percentage (182; 4%) of our medication errors were considered harmful; this may also reflect our more diverse population of patients.

A second study that had objectives similar to ours was a cross-sectional study performed by Stavroudis et al.^[8] This study assessed neonatal intensive care unit medication error reports submitted to MEDMARX; of 6749 reports, 4% were determined to be harmful (ie, preventable ADEs). The authors found a correlation between ISMP high-alert medications, prescribing errors, and equipment/device failure and patient harm; however, the authors did not perform a multivariate analysis or adjust for the impact of confounding factors. We found a similar rate of reported preventable ADEs in our study and our results echo the importance of the ISMP high-alert medication list as a predictor of patient harm.

Finally, a cross-sectional study was performed by Crespin et al. in the setting of long-term care.^[10] The objective of this study was to identify factors that contribute to both repeat errors and patient harm. The authors assessed 15,037 medication errors reported to a state agency that represented approximately 26% of the patients cared for at enrolled facilities. Of the medication errors, 123 (0.8%) were determined to contribute to patient harm. Using a multivariate analysis, the authors identified repeat medication errors, errors that occurred between 1500 and 2300, errors in documentation, and temporary staff members as predictors of harm. These factors differed somewhat from our results, highlighting the difference in practice setting and the importance of assessing local trends.

There are a number of limitations to our study that could impact interpretation of the results. Principal among these is that this study aimed to quantify specific factors that affect the risk for harm due to a medication error; while several of these were identified, the multifactorial nature of medication errors should be taken into account when applying these results to practice. For a medication error to harm a patient, multiple steps within the medication use processes must break down. Even though a factor may not have been found to independently affect the risk for harm in this study, it should not be precluded as a factor to be considered by individual practitioners or safety specialists managing medication use systems. Additionally, observational results such as these should not be considered confirmatory. With this in mind, a proactive approach to medication safety using health care failure modes and effects analysis and other tools should be used on a regular basis to minimize the risk of harmful medication errors.

Another limitation is that the data used in the study were collected by a single pharmacist reviewer (RDB); this could increase the risk of bias from a pharmacist perspective. This risk is partially mitigated by the fact that the data were primarily obtained from standard medication error reports prepared by separate individuals with input from relevant medical, nursing, and other staff. Additionally, this individual had training and experience in performing medication error evaluation.

Another important limitation is the breadth of data collected. There are certainly other potential factors affecting the risk of patient harm, such as the training level of individual providers, the familiarity of individual providers with the practice area, time of day as measured by shift, impact of technology, and additional patient-specific information, which were not collected as part of this study.

Finally, as with all single-system studies of this nature, these results may not be generalizable to all hospitals, including other facilities within our own system. Each institution has a responsibility to its patients to carefully evaluate potential failures in the medication use process that apply at a local level, considering the unique nature of its own patient population. While our results may represent potential target points, we strongly suggest, based on these results, that key institutional stakeholders in medication safety evaluate local factors affecting the risk of harm from medication errors. This analysis should focus on the factors assessed in this study, as well as those mentioned as being omitted. The impact of the institution-specific culture of reporting (voluntary, in our case) should also not be underestimated. While our results reflect factors affecting risk for harm associated with reported medication errors, they may not be truly reflective of those errors that remain unreported. Considering that a number of factors appeared to potentially affect risk of harm but were identified too infrequently to assess that risk adequately, we echo previous calls to improve the rate of error reporting by enabling a culture of safety.

In conclusion, preventable ADEs represented 4% of medication errors reported at our institution; approximately 91% of these were NCC MERP category E. Our study suggested a number of potential factors that could affect the risk for a preventable ADE, relative to a medication error that did not cause harm, and suggested that several of these independently affect that risk. Factors associated with increased independent risk of harm were 30-day readmissions, time of day 0300–0659, and ISMP high-alert medications. Factors associated with decreased independent risk of harm were having experienced more than 1 medication error, errors occurring during February or April, dispensing errors, and pharmacist review of the medication order.

Medication safety directors and pharmacy operations managers should develop programs to promote safe, conscientious use of ISMP high-alert medications, promote pharmacist review of orders, and direct attention toward high-risk patients. Efforts should be made to improve the rate of medication error reporting, as it is likely there were too few patients in some groups of factors (ie, medication class, frequency, and route of administration) to assess their impact on the risk of harm, and to identify factors associated with harm at local levels.

[ CLOSE WINDOW ]

References

1. Kohn LT, Corrigan JM, Donaldson MS, eds. To err is human: building a safer health system. Washington, DC: National Academy Press, 1999:1–8.
2. Krahenbuhl-Melcher A, Schlienger R, Lampert M, et al. Drug-related problems in hospitals: a review of the recent literature. Drug Saf 2007;30:397–407.
3. Bates DW, Boyle DL, Vander Vliet MB, Schneider J, Leape L. Relationship between medication errors and adverse drug events. J Gen Intern Med 1995;10:199–205.
4. Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA 2001;285:2114–20.
5. Ninno SD, Ninno MA. Adverse drug events. In: Mueller BA, Bertch KE, Dunsworth TS, et al., eds. Pharmacotherapy self-assessment program. 4th ed. Kansas City, MO: American College of Clinical Pharmacy, 2002.
6. Morimoto T, Gandhi TK, Seger AC, Hsieh TC, Bates DW. Adverse drug events and medication errors: detection and classification methods. Qual Saf Health Care 2004;13:306–14.
7. Zaal RJ, van Doormaal JE, Lenderink AW, et al. Comparison of potential risk factors for medication errors with and without patient harm. Pharmacoepidemiol Drug Saf 2010;19:825–33.
8. Stavroudis TA, Shore AD, Morlock L, et al. NICU medication errors: identifying a risk profile for medication errors in the neonatal intensive care unit. J Perinatal 2010;30:459–68.
9. Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events: implications for prevention. JAMA 1995;274:29–34.
10. Crespin DJ, Modi AV, Wei D, et al. Repeat medication errors in nursing homes: contributing factors and their association with patient harm. Am J Geriatr Pharmacother 2010;8:258–70.
11. National Coordinating Council for Medication Error Reporting and Prevention. Index for categorizing medication errors. http://www.nccmerp.org/pdf/indexBW2001–06–12.pdf (accessed 2011 Oct 28).

AcknowledgmentsWe thank EJ Last PharmD, who provided valuable editorial support during the data collection and development of the manuscript.

The Annals of Pharmacotherapy. 2012;46(5):634-641. © 2012 Harvey Whitney Books Company