Defining and Measuring Patient Safety in the Critical Care Unit
Alan M. Fein
Steven Y. Chang
Sara L. Merwin
David Ost
John E. Heffner
Patient safety has become a major concern of the general public, policy makers, and local, state, and national government. Frequent news coverage has been devoted to individuals who were victims of serious medical errors. In the 1999 publication of the Institute of Medicine, To Err Is Human: Building a Safer Health Care System [1], the risks of medical care were highlighted, particularly the nearly 100,000 deaths per year that could be attributed to medical errors. A general sense of the importance of a safety culture in the intensive care unit (ICU) is increasing, as suggested by the multiple reports and publications in the lay and scientific media devoted to this topic [2].
The high-risk environment of the ICU benefits from integrated and coordinated systems that identify patient safety problems and report them to providers so they can improve
their performance. To maintain high-quality care, critical care teams need to know not only what to do but also how they are doing and what they need to do to improve their structure, processes, and outcomes of care. Donabedian [3] first described these three domains—structure, process, and outcome—as necessary elements for measuring the quality of health care. They also serve as a conceptual framework for measuring patient safety in the ICU.
their performance. To maintain high-quality care, critical care teams need to know not only what to do but also how they are doing and what they need to do to improve their structure, processes, and outcomes of care. Donabedian [3] first described these three domains—structure, process, and outcome—as necessary elements for measuring the quality of health care. They also serve as a conceptual framework for measuring patient safety in the ICU.
On a broad scale, ICU patient safety-reporting systems identify trends and patterns allowing health care organizations, governmental agencies, and private accreditation organizations to monitor the quality and safety of health care delivery, which facilitates public reporting of data and increases transparency [4]. Patient safety-reporting systems also have the potential to create large data repositories that inform the development of strategies that reduce the risk of preventable medical incidents [5,6].
Effective reporting systems require definitions and methods that are standardized throughout the community of providers, so that information can be shared and meaningful comparisons can be made. In the 2003 report, Patient Safety: Achieving a New Standard of Care, the Institute of Medicine (IOM) emphasized the importance of standardizing and better managing information on patient safety to improve outcomes of care [5]. A critical element of this standardization is the development of a common taxonomy of patient safety terms. In the absence of standardized terminology, health care providers have no way to know what events to capture and how to describe those events in consolidated reports [7]. Also, fragmented approaches for defining and classifying near misses, adverse events, and other patient safety concepts prevent aggregation of data in formats that allow analysis and summary reporting [1,8]. To date, governmental and private sector accrediting bodies have not coordinated their efforts to develop actionable, integrated, validated, and reliable systems to measure and report medical errors and patient safety [9].
Safety Lessons from Other Industries
Safety and error prevention in the health care setting compares unfavorably with that in aviation, banking, chemical manufacturing, and military services in peacetime. Lessons from these industries are now being applied to the health care industry. Approaches to safety in these industries are characterized by well-defined strategies to protect workers and customers. Technology-based approaches are part of this strategy, but organizational and psychologic aspects are contributing factors as well. For example, developing a culture of safety has been identified as one important method of improving safety. The aviation industry has focused on the importance of teamwork in reinforcing a safety culture.
Although technical, organizational, and psychologic interventions are effective, it is also worth noting the limits of the existing method. Persistence of fatalities in aviation and auto transportation suggest that safety efforts may be counterbalanced by other competing risk factors such as high volumes, greater complexity of the product, cost-pressure, and rapidly changing designs. This is particularly relevant to health care because the population is changing (higher number of increasingly older and higher-risk patients) and the technology is changing at a very rapid rate [10].
Thus, there is probably an upper limit in terms of cost-effective health care safety that can be reached, but has not been attained. Health services are being encouraged by the IOM report to aim for an error rate of less than 3.4 per million, that is, “six-sigma quality.” The discipline of anesthesiology in particular has made substantial contributions through its development of a safety culture and equipment-manufacturing standardization that resulted in a reduction in anesthesia-related deaths to 4.4 per million, that is, “five-sigma standard.”
To achieve this standard in the ICU, there must be a precise definition of the terms needed to study patient safety, their methods of measurement, how these can be applied to the special problems of ICU organization, physician training, and development of a culture of safety, and finally how these concepts apply to governmental regulations.
Definitions
The basic terms in common use to define concepts of patient safety are listed in Table 210.1 and show the working definitions that have entered into the lexicon of the patient safety industry [11]. Health care quality is defined by the IOM as “the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge” [12]. This definition conforms to two (process and outcome) of the three constructs (structure, process, and outcome) proposed by Donabedian [3] to be necessary for measuring the quality of health care. The IOM has also listed several attributes of quality care that define quality care as being safe, patient-centered, timely, effective, efficient, and equitable [13]. Thus, patient safety is one domain within the broader concept of quality.
Patient safety has been variously defined by the Agency for Health Care Research and Quality (AHRQ) as “the absence of the potential for, or the occurrence of, healthcare associated injury to patients created by avoiding medical errors as well as taking action to prevent errors from causing injury” [14] and “freedom from accidental or preventable injuries produced by medical care” [15].
Within this context of safety, medical errors are defined as “mistakes made in the process of care that result in or have the potential to result in harm to patients. Mistakes include the failure of a planned action to be completed as intended or the use of a wrong plan to achieve an aim. These can be the result of an action that is taken (error of commission) or an action that is not taken (error of omission)” [14]. Errors of commission (e.g., ordering an incorrect drug dose) as compared with errors of omission (e.g., failure to order heparin for venous thromboembolism prophylaxis) are more readily noted. Errors are further classified as active or latent [16,17]. Active errors occur at the interface between a human provider and a care-delivery system (e.g., mechanical ventilator, intravenous pump) and typically involve readily apparent actions (e.g., adjusting a dial incorrectly). Latent errors define a less obvious failure of a health care organization or structure that contributed to errors or allowed the errors to harm patients. An example of a latent error would be understaffing of nurses in an ICU. Other typologies include domains that ascribe characteristics of preventability, seriousness and whether the error was intercepted before affecting a patient [18] (Table 210.1).
Errors have also been classified as slips or mistakes. Slips are failures of automatic behaviors, or lapses in concentration (e.g., forgetting to perform a routine task due to a lapse in memory) and often occur from fatigue or distractions in the workplace. Mistakes represent incorrect choices, such as choosing the wrong drug for a clinical condition, and typically result from inexperience or lack of knowledge or training. The remedies for these two types of errors differ, with slips being more responsive to removing distractions from the workplace or automating monotonous tasks and mistakes respond to increased training or supervision.
Incidents are defined as unexpected or unanticipated events or circumstances not consistent with the routine care of a particular patient, which could have or did lead to an unintended
or unnecessary harm to a person, or a complaint, loss, or damage. Adverse events are different, and are defined as an “untoward and usually unanticipated outcome that occurs in association with health care” [14] or more broadly stated by the IOM as “an injury resulting from a medical intervention” [1]. The Critical Care Safety Study defines adverse events as “Any” injury due to medical management, rather than the underlying disease [18]. Describing an event as an adverse event does not imply poor-quality care or that an error occurred. An adverse event only indicates that an undesirable outcome resulted from a medical intervention rather than an underlying disease process [19]. As an example, if proper procedures are followed for central line placement but the patient develops a pneumothorax, this would constitute an adverse event even though all the elements of quality care were met.
or unnecessary harm to a person, or a complaint, loss, or damage. Adverse events are different, and are defined as an “untoward and usually unanticipated outcome that occurs in association with health care” [14] or more broadly stated by the IOM as “an injury resulting from a medical intervention” [1]. The Critical Care Safety Study defines adverse events as “Any” injury due to medical management, rather than the underlying disease [18]. Describing an event as an adverse event does not imply poor-quality care or that an error occurred. An adverse event only indicates that an undesirable outcome resulted from a medical intervention rather than an underlying disease process [19]. As an example, if proper procedures are followed for central line placement but the patient develops a pneumothorax, this would constitute an adverse event even though all the elements of quality care were met.
Table 210.1 General Terms Used in Patient Safety | ||
|---|---|---|
|
Most existing typologies of definition related to patient safety pertain to medical interventions. Errors of diagnosis are emerging as relatively uninvestigated but equally important causes of unsafe patient management in the ICU [20].
Measurement of Safety in the Intensive Care Unit
The science of measuring and reporting patient safety remains immature and can be viewed from the perspective of whether the measure identifies a structure, process, or outcome related to safety. Different methods of measurement focus on one or more of these elements and may be more or less efficient at identifying safety risks in one or more of these domains. The primary methods of measurement include incident reporting, targeted monitoring, use of discharge data sets, process of care measurement, trigger tools, ICU audits, and direct observation [18].
Incident Reporting
In terms of collecting safety measurement data, traditional methods based on incident reporting of specific adverse events have been largely ineffective for several reasons [21]. First, reports have been generated in a punitive environment that focuses on the provider who committed an error rather than systems of care and discourages self-reporting of errors [5]. Second, each report of an error represents a “numerator” value that does not give insight into the denominator pool of patients at risk for similar errors. In the absence of these values, the incidence of errors and the overall safety of an ICU cannot be assessed. Third, definitions of errors used in incident-reporting systems vary, which impedes data synthesis, analysis, collaborative work, and evaluation of the impact of changes in health care delivery [22]. And fourth, appropriate functional data spanning the domains of structure, process, and outcome are not collected, which impedes the ability to “deconstruct” an error to understand its root causes and patient impact.
Recent advances to incident reporting have enhanced the detection and analysis of errors. Internet-based systems allow anonymous reporting of errors to encourage providers who have either committed an error or have knowledge of an error to enter related information into a central data repository. Institutional commitment to a “culture of safety” has a motivational effect on error reporting because health care providers note the impact that a reported error can generate in terms of improved quality of care. This culture requires several essential process elements to enhance error reporting: A team (a) convenes to develop preventative solutions to a reported error, (b) generates plans to improve the care, and (c) has a method for implementing and measuring the impact of their plan [23]. The Intensive Care Unit Safety Reporting System (ICUSRS) is
an anonymous reporting system that focuses on “systems factors” rather than “person factors” and provides expert analysis with feedback and guidance to improve processes of care and prevent error recurrences [11,24]. The University Health Systems Consortium’s Safety Net reporting system can generate consolidated reports with application to the ICU [24].
an anonymous reporting system that focuses on “systems factors” rather than “person factors” and provides expert analysis with feedback and guidance to improve processes of care and prevent error recurrences [11,24]. The University Health Systems Consortium’s Safety Net reporting system can generate consolidated reports with application to the ICU [24].
However, problems remain with incident reporting in terms of the taxonomy used to describe errors and adverse events. The Joint Commission (JC) published a patient safety event taxonomy and classification schema for near misses, errors, and adverse events [11]. The taxonomy was designed to conform to an analytical framework and common word usages to promote its use and the understanding of its output. Data entered allows classification of a patient safety event within five complementary primary groups: impact—the outcome or effects of medical error and systems failure, commonly referred to as harm to the patient; type—the implied or visible processes that were faulty or failed; domain—the characteristics of the setting in which an incident occurred and the type of individuals involved; cause—the factors and agents that led to an incident; and prevention and mitigation—the measures taken or proposed to reduce incidence and effects of adverse occurrences.
The ICUSRS reporting platform similarly uses a framework for evaluating factors that contribute to an incident [11]. Both the JC and ICUSRS systems recognize that errors are multifactorial and therefore include multiple variables along the three domains of structure, process, and outcomes, such as caregiver performance, systems of care, resource availability, functioning of teams, and the environment of care. These systems describe events with a multidimensional taxonomy to allow the comprehensive description and full deconstruction of errors to determine their root causes [9]. However, even if taxonomy issues of incidence reporting are improved, the problem of determining the true incidence rate remains. A comprehensively described and deconstructed incident only gives insight into the numerator; it does not provide information on the number of patients at risk and does not allow determination of true incidence rates.
Targeted Monitoring
A complementary approach to incident reporting is targeted monitoring. ICUs can measure their patient safety outcomes by monitoring a specific indicator, such as the incidence of Clostridium difficile infection in the ICU or ventilator-associated pneumonia. In so doing, ICUs are challenged to define their denominators and select indicators that can be readily detected and counted to provide an accurate numerator. The denominator is especially difficult to determine because the measurement has major impact on interpretation [11]; for instance, C. difficile infection rates can be described per ICU patient, patient ICU days, or at-risk patient ICU days. The numerator data are equally challenging because of the time and expense of chart extraction needed for their collection. If the characteristics of the patient population change over time, then these factors must be accounted for as well. For example, if the patient population changes or new services such as transplant are offered by a given hospital, then the patient mix will change and adjusted hazard rates will be needed. Thus, for this approach to work, a multidisciplinary team that includes people with ICU training, organizational skills, database management, and epidemiology are needed.
Discharge Data
Discharge data represents a potential source of information to allow the retrospective collection of quality and safety indicators to profile ICU performance [25,26,27]. Recently, AHRQ has developed empiric measures of quality and safety from multistate discharge data in a redesign of the original Healthcare Cost and Utilization Project Quality Indicators [28]. The Patient Safety Indicators are relevant to ICU safety of care. Although most of these indicators relate to surgical patients, newer indicators are being designed to measure the safety of care for medical patients with critical illnesses, such as myocardial infarction, stroke, and congestive heart failure.
Although this method is powerful and can be quite useful, it is important to also recognize its limitations. Discharge data analysis gives insight into outcomes, but little information on structure or process. Large datasets such as these also have limited data quality for clinically relevant covariates, so controlling for confounders is difficult. Because all of the clinically relevant covariates are not included, the problem of residual confounding is always a problem and caution should be exercised when interpreting results. Making interinstitutional comparisons is therefore difficult, and even when trending data over time, results must be analyzed with caution. When patient populations and their problems are relatively homogenous and stable over time, this is a good system (e.g., surgical patients). When there is marked heterogeneity in terms of clinical problems and rapid changes in process of care over time, this approach will have difficulty. Having said this, discharge datasets can be an important tool for hypothesis generation so that ICU leaders can then launch more systematic studies into particular problems.
Process of Care Measurement
Safety can also be measured through determination of the proportion of patients who receive certain processes of care that have a strong evidentiary base for improving clinical outcomes. However, it may be difficult to isolate and to ascertain the contributory effect of influential factors, that is, adherence to best practice by the caregiving team, the role of complications, or level of care. Physicians and other clinicians often have a stronger sense of accountability toward a process measure than an outcome measure because the process measure can be more strongly linked to a particular care provider or team [29]. Also, physicians may believe that outcomes can be overly influenced by severity of disease and prove resistant to quality improvement efforts. To serve as an accurate measure of safety and to influence quality improvement, process measures must have a causal relationship with the outcome they are intended to represent.
Examples of process measures include approaches for ordering therapy in the ICU. Medication errors and adverse drug events occur commonly in the ICU [30] and can be limited by the use of formatted drug-ordering forms [31]. Computerized physician order entry for drugs has the potential to decrease the rate of serious medication errors [32] and to improve clinical outcomes when applied to antibiotic prescribing [33]. Additional care processes that should be in place to support patient safety can be constructed by reviewing evidence-based clinical practice guidelines [34], such as standardizing orders for ventilator management in the ICU [35].
Process of care measurement is often very effective for certain types of problems, like computerized order entry, but it is important to recognize some of the limitations and difficulties inherent in the system when applied to more complex problems. When strong evidence-based clinical practice guidelines are available, this is a feasible strategy, but often this is not the case. In addition, properly identifying those patients eligible for a particular protocol in the appropriate time period is critical. Examples include the use of thrombolytics for myocardial infarction and stroke, as well as recombinant activated protein C
for sepsis. Determining the numerator for such process of care measures is fairly easy (who actually received the drug), but determining the denominator can be more difficult and can be costly because of the time and expense needed for data collection (e.g., reviewing every chart in the emergency department of a patient presenting with angina or suspected myocardial infarction). In addition, chart abstraction in such cases usually requires a high level of expert judgment, which makes it even more difficult. Thus, process of care measurement, because of cost and time considerations, may be a suitable approach to improving safety for those problems in which there is a strong-evidence base and in which the costs of identifying the patient population (both numerator and denominator) are sustainable and warranted by the value of information obtained.
for sepsis. Determining the numerator for such process of care measures is fairly easy (who actually received the drug), but determining the denominator can be more difficult and can be costly because of the time and expense needed for data collection (e.g., reviewing every chart in the emergency department of a patient presenting with angina or suspected myocardial infarction). In addition, chart abstraction in such cases usually requires a high level of expert judgment, which makes it even more difficult. Thus, process of care measurement, because of cost and time considerations, may be a suitable approach to improving safety for those problems in which there is a strong-evidence base and in which the costs of identifying the patient population (both numerator and denominator) are sustainable and warranted by the value of information obtained.
Intensive Care Unit Audits
An audit of the existing structure of an ICU can also measure patient safety. Evidence supports improved outcomes in ICUs staffed by sufficient numbers of board-certified intensivists [36,37]. Additional structure measures of safety include the presence of resources to establish ongoing competency of medical staff and residents [38], adequate nurse staffing and skill sets [39,40], and appropriate technology resources, such as smart pumps and bar coding [41]. And, most importantly, the presence of a culture of safety represents a central structure within an ICU for promoting safety. Such a culture emerges from the presence of leaders who are committed to safety and staff who understand that errors are inevitable, acknowledge that errors are to be reported, dedicate time to learn about new risks and hazards, support teamwork and open communication, and upgrade procedures and implement safeguards on a continuing basis [16]. Organizational characteristics of safe programs with low accident rates include successful safety programs with strong management commitment, safety training as part of new employee’s training, frequent open contacts between workers and management, general environmental control and good housekeeping, a stable workforce, and positive reinforcement for good safety practices. Surveys exist that allow ICU directors to assess the status of their units’ culture of safety [42,43,44,45]. Pronovost and Sexton [44] recommend measuring the entire hospital annually with the full Safety Attitudes Questionnaire, which has construct validity and sufficient reliability for measuring the single construct of safety culture. Once measured, the culture of safety can be improved with focused interventions for any low-scoring hospital areas, such as an ICU.
Related posts:
Full access? Get Clinical Tree
