Medical Harm: Patient Perceptions and Follow-up Actions

Much research has been conducted to describe medical mistakes resulting in patient harm using databases that capture these events for medical organizations. The objective of this study was to describe patients’ perceptions regarding disclosure and their actions after harm.

We analyzed a patient harm survey database composed of responses from a voluntary online survey administered to patients by ProPublica, an independent nonprofit news organization, during a 1-year period (May 2012 to May 2013). We collected data on patient demographics and characteristics related to the acknowledgment of patient harms, the reporting of patient harm to an oversight agency, whether the patient or the family obtained the harm-associated medical records, as well as the presence of a malpractice claim.

There were 236 respondents reporting a patient harm (mean age, 49.1 y). In 11.4% (27/236) of harms, an apology by the medical organization or the clinician was made. In 42.8% (101/236) of harms, a complaint was filed with an oversight agency. In 66.5% (157/236) of harms, the patient or the family member obtained a copy of the pertinent medical records. A malpractice claim was reported in 19.9% (47/236) of events.

In this sample of self-reported patient harms, we found a perception of inadequate apology. Nearly half of patient harm events are reported to an oversight agency, and roughly one-fifth result in a malpractice claim.

From the *Department of Surgery, School of Medicine, †Departments of Anesthesiology/Critical Care Medicine and Surgery, School of Medicine, and ‡Departments of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland.

Correspondence: Martin A. Makary, MD, MPH, Department of Surgery, Johns Hopkins Hospital, Halsted 610, 600 N. Wolfe St, Baltimore, MD 21231 (e-mail:

The authors disclose no conflict of interest.

Magnitude of Anemia at Discharge Increases 30-Day Hospital Readmissions

Anemia during hospitalization is associated with poor health outcomes. Does anemia at discharge place patients at risk for hospital readmission within 30 days of discharge? Our objectives were to examine the prevalence and magnitude of anemia at hospital discharge and determine whether anemia at discharge was associated with 30-day readmissions among a cohort of hospitalizations in a single health care system.

From January 1, 2009, to August 31, 2011, there were 152,757 eligible hospitalizations within a single health care system. The endpoint was any hospitalization within 30 days of discharge. The University HealthSystem Consortium’s clinical database was used for demographics and comorbidities; hemoglobin values are from the hospitals’ electronic medical records, and readmission status was obtained from the University HealthSystem Consortium administrative data systems. Mild anemia was defined as hemoglobin of greater than 11 to less than 12 g/dl in women and greater than 11 to less than 13 g/dl in men; moderate, greater than 9 to less than or equal to 11 g/dl; and severe, less than or equal to 9 g/dl. Logistic regression was used to assess the association of anemia and 30-day readmissions adjusted for demographics, comorbidity, and hospitalization type.

Among 152,757 hospitalizations, 72% of patients were discharged with anemia: 31,903 (21%), mild; 52,971 (35%), moderate; and 25,522 (17%), severe. Discharge anemia was associated with severity-dependent increased odds for 30-day hospital readmission compared with those without anemia: for mild anemia, 1.74 (1.65–1.82); moderate anemia, 2.76 (2.64–2.89); and severe anemia, 3.47 (3.30–3.65), P

Anemia at discharge is associated with a severity-dependent increased risk for 30-day readmission. A strategy focusing on anemia treatment care paths during index hospitalization offers an opportunity to influence subsequent readmissions.

From the *Department of Cardiothoracic Anesthesia, Heart and Vascular Institute, †Quality and Patient Safety Institute, ‡Department of Quantitative Health Sciences, Research Institute, §Business Intelligence, Medical Operations, ∥Department of Thoracic and Cardiovascular Surgery, Heart and Vascular Institute, and ¶Department of General Surgery, Digestive Disease Institute, Cleveland Clinic, Cleveland, Ohio.

Correspondence: Colleen G. Koch, MD, MS, MBA, Department of Cardiothoracic Anesthesia, Heart and Vascular Institute, Cleveland Clinic, 9500 Euclid Ave/Mail Stop J-4, Cleveland, OH 44195 (e-mail:

The authors disclose no conflict of interest.

Efficacy and Cost-Benefit Analysis of a Global Environmental Cleaning Algorithm on Hospital-Acquired Infection Rates

This study evaluates clinical outcomes and cost-benefit analysis before and after implementation of a global environmental cleaning algorithm on all hospital-acquired infection (HAI) rates.

A retrospective, quasi-experimental study design was used to review the hospital’s procedure and infection rate database for all HAIs from January 1, 2009, through June 30, 2011. We calculated the infection rates and did a cost-benefit analysis before and after the environmental cleaning algorithm was instituted on July 19, 2010.

The methicillin-resistant Staphylococcus aureus rates per 1000 patient days decreased 63%. The central line–associated bloodstream infection rate had a 72% reduction. The catheter-associated urinary tract infection rate dropped 79%. The vancomycin-resistant Enterococcus rate went down 53%. The hospital-acquired Acinetobacter baumanii infection rate had a 65% reduction. The medical intensive care unit ventilator-associated pneumonia rate was reduced 72%. Cardiothoracic sternal wound surgical site infection (SSI) rate dropped 93%, spinal fusion SSI decreased 56%, and total knee arthroplasty SSI was eliminated with a 100% reduction. The hospital avoided an estimated 13 deaths and $5,800,526 in costs during a 1-year period.

This global environmental cleaning protocol was associated with decreased HAIs and hospital costs.

From the *Quality, Safety & Risk, and †Infection Prevention and Control, Riverside Community Hospital, Riverside, California.

Correspondence: Barbara R. Everett, RN, BS, MSN, CIC, Riverside Community Hospital, 4445 Magnolia Ave, Riverside, CA 92501 (e-mail:

The authors disclose no conflict of interest.

From Bad to Worse: Anemia on Admission and Hospital-Acquired Anemia

Anemia at hospitalization is often treated as an accompaniment to an underlying illness, without active investigation, despite its association with morbidity. Development of hospital-acquired anemia (HAA) has also been associated with increased risk for poor outcomes. Together, they may further heighten morbidity risk from bad to worse.

The aims of this study were to (1) examine mortality, length of stay, and total charges in patients with present-on-admission (POA) anemia and (2) determine whether these are exacerbated by development of HAA.

In this cohort investigation, from January 1, 2009, to August 31, 2011, a total of 44,483 patients with POA anemia were admitted to a single health system compared with a reference group of 48,640 without POA anemia or HAA.

Data sources included the University HealthSystem Consortium database and electronic medical records. Risk-adjustment methods included logistic and linear regression models for mortality, length of stay, and total charges. Present-on-admission anemia was defined by administrative coding. Hospital-acquired anemia was determined by changes in hemoglobin values from the electronic medical record.

Approximately one-half of the patients experienced worsening of anemia with development of HAA. Risk for death and resource use increased with increasing severity of HAA. Those who developed severe HAA had 2-fold greater odds for death; that is, mild POA anemia with development of severe HAA resulted in greater mortality (odds ratio, 2.57; 95% confidence interval, 2.08–3.18; P 0.001), increased length of stay (2.23; 2.16–2.31; P P 0.001).

Present-on-admission anemia is associated with increased mortality and resource use. This risk is further increased from bad to worse when patients develop HAA. Efforts to address POA anemia and HAA deserve attention.

From the *Department of Cardiothoracic Anesthesia, Heart and Vascular Institute, †Quality and Patient Safety Institute, ‡Department of Quantitative Health Sciences, Research Institute, §Business Intelligence, Medical Operations, ∥Department of Thoracic and Cardiovascular Surgery, Heart and Vascular Institute, and ¶Department of General Surgery, Digestive Disease Institute, Cleveland Clinic, Cleveland, Ohio.

Correspondence: Colleen G. Koch, MD, MS, MBA, Department of Cardiothoracic Anesthesia, Heart and Vascular Institute, and Quality and Patient Safety Institute, Cleveland Clinic, 9500 Euclid Ave/Mail Stop J-4, Cleveland, OH 44195 (e-mail:

The authors disclose no conflict of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (

Screening Questions for Nonsteroidal Anti-inflammatory Drug Risk Knowledge

The aim of this study was to evaluate screening questions for estimating nonsteroidal anti-inflammatory drug (NSAID) risk knowledge.

Cross-sectional data from a telephone interview of NSAID users 50 years or older from 39 physician practices in Alabama were used. Patient-reported awareness of prescription NSAID risk and health literacy were the independent variables, and a cumulative index score of objectively tested knowledge of 4 prominent NSAID risks was the dependent variable. General linearized latent and mixed model ordered logistic regression was used to estimate associations among the independent variables, covariates, and objectively tested NSAID risk knowledge. Population-averaged probabilities for levels of objectively tested NSAID risk knowledge were subsequently estimated.

Subjective awareness of any prescription NSAID risk (adjusted odds ratio [AOR], 2.40; 95% confidence interval [CI], 1.55–3.74), adequate health literacy (AOR, 1.71; 95% CI, 1.04–2.83), and physician counseling about 1 or more NSAID risks (AOR, 1.69; 95% CI, 1.09–2.61) were significantly and positively associated with NSAID risk knowledge. The probability of correctly answering at least 1 of the 4 NSAID risk knowledge questions was 70% in the absence of any subjective risk awareness and in less than adequate health literacy. Whereas the probability of correctly answering at least 1 of the 4 NSAID risk knowledge questions increased to 86% in the presence of subjective awareness of any prescription NSAID risk and adequate health literacy.

Screening questions for subjective NSAID risk awareness and health literacy are predictive of objectively tested NSAID knowledge and can be used to triage patients as well as subsequently initiate and direct a conversation about NSAID risk.

From the *Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, Utrecht, The Netherlands; †College of Pharmacy, The University of Oklahoma College of Pharmacy, Tulsa, Oklahoma; ‡American Society of Health-System Pharmacists (ASHP) Research and Education Foundation, Bethesda, Maryland; §Center for Education and Research on Therapeutics of Musculoskeletal Disorders (CERTs), University of Alabama at Birmingham, Birmingham, Alabama; and ∥Division of Health Informatics and Implementation Science, Department of Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, Massachusetts.

Correspondence: Michael J. Miller, RPh, DrPH, FAPhA, College of Pharmacy, The University of Oklahoma, 4502 East 41st St, Tulsa, OK 74135 (e-mail:

Dr Allison serves on the Medical Education External Review Panel for Pfizer. All other authors disclose no conflict of interest.

Supported, in part, by the Agency for Healthcare Research and Quality (AHRQ) Centers for Education and Research on Therapeutics cooperative agreement (U18HS016956).

Implementation of Crew Resource Management: A Qualitative Study in 3 Intensive Care Units

Crew resource management (CRM) has been increasingly applied in health care to improve safe patient care.1 It consists of a team training that was developed in the aviation field and aims to increase participants’ understanding of how certain threats, such as miscommunication, can develop2 as well as provides tools and skills to respond to such threats. The focal point of the training is nontechnical skills, such as communication, teamwork, leadership, situational awareness, decision making, and problem solving.3 These nontechnical skills complement the technical abilities and contribute to safe and efficient task performance.4 At long last, CRM should be embedded in the organizational culture as a way of doing things.2

The findings of evaluations of CRM as a classroom-based training are promising but inconclusive with regard to behavioral change.5 For instance, McCulloch and colleagues6 found an increase in the use of nontechnical skills for nurses but not for anesthetists or surgeons. Rabol et al5 recommend more qualitative research to get a deeper insight into why these mixed results occur. The effects of CRM—and interventions in general—are determined by the persuasiveness of its program, as well as the quality of the implementation.7,8 Therefore, the quality of the implementation might explain the mixed results of CRM.

To date, implementation has never been the main focus of CRM evaluation research. It has mainly been described alongside the quantitative results in 2 ways. It has been described, first, as a predefined part of the training, expressed in the Methods section. For instance, Stead et al9 state that the implementation of CRM comprises 3 phases: site assessment, training, and sustainment. Second, it has been described by discussing the main barriers and facilitators perceived by the researchers while conducting their study. Morey et al,10 for example, stress that support from management was a prime facilitator in the implementation of CRM.

Although both ways of describing the implementation yield valuable information, they do not depict the whole process of implementation and overlook the underlying vision, structure and follow-up. An exception is the study by Marshall and Manus,11 in which they described the important characteristics, goals, changes, barriers, and facilitators for each participating department. They did not, however, distinguish different phases of implementation.

The present research systematically describes the implementation from the perspective of 3 trained intensive care units (ICUs) based on interviews with implementation leaders. The choices, rationales, and consequences that played a role in the implementation process will be assessed, whereas existing system change frameworks present in each unit will be used to characterize the implementation in each ICU. In this way, we aim to gain insight into contextual factors influencing the effects of CRM and to present practical examples to readers interested in the implementation of CRM. In addition, increased understanding of the implementation process of CRM in 3 ICUs might help explain the effects of CRM. In short, the present study takes a first glance at the implementation of CRM.

Back to Top | Article Outline


Design and Setting

The design of the study was built around 3 stages of implementation: (1) the preparation, (2) the actions after the CRM training, and (3) the plans for the future. To assess all 3 stages, semistructured interviews with implementation leaders were conducted. In addition, after the training, a questionnaire was administered to measure the reaction of the participants about CRM training. Finally, we collected all plans of action that had been formulated during the training sessions.

Three Dutch ICUs participated in the present study and received CRM training. The ICUs were part of nonacademic teaching hospitals in The Netherlands, with a mean of 872 beds, all located in an urban environment. The ICUs had 12 to 14 beds and 65 to 79 unique employees. All units delivered level 2 care, which implies a 24-hours-per-day availability of an IC physician for the care of patients, and structurally around 0.40 full time-equivalent IC physicians per bed.12 The medical staff of the ICUs included IC physicians, residents, nurses, nurse trainees, and members of the management. The training was not free of charge; therefore, the ICUs had to be able to make the necessary financial and organizational arrangements. The efforts to make these arrangements ensured that the ICUs were willing to receive CRM training. The selection process of the participating ICUs is described elsewhere.13

All ICUs started discussing CRM seriously within their unit after being approached to participate in the larger effectiveness study13 of which the present research is a part. This provided the opportunity to monitor these ICUs from the point at which the first person—the pioneer—tried to convince staff about the benefits of CRM until 15 months after receiving the CRM training. The larger study compared 3 ICUs that received CRM training with 3 comparable ICUs in a controlled trial with 1 premeasurement and 2 follow-up measurements. The trial aimed to assess all levels of the Kirkpatrick evaluation framework for training programs (reaction, learning, behavioral change, and organizational impact). The study was approved by the Ethical Committee of the VU Medical Center. Participation was confidential, and all gathered data were stored entirely anonymously.

Back to Top | Article Outline

Intervention: CRM Training

The CRM training was classroom based and consisted of 2 consecutive days from 9 AM until 5 PM. Because a maximum of 15 participants per session had been set, several trainings were organized to include all members of the IC staff. Two ICUs received 4 sessions, and 1 ICU received 6. In total, 14 CRM sessions were organized, and in such a way that each discipline was represented during the sessions, guaranteeing a multidisciplinary audience. Each ICU was trained separately.

The training aimed to improve patient safety on all fronts of the ICU by creating awareness regarding the threats of unsafe behavior on the individual, team, and organizational level. Subsequently, the participants were stimulated to develop ways to recognize these threats and to prevent negative consequences on each of these levels. These solutions were continuously recorded during the 2 days of training by means of writing down concrete plans of action. An example of an action on the team level is to explicitly appoint a coordinator of the day to improve and clarify leadership, teamwork, and communication.

The ICUs were free to choose how they wished to organize their implementation of CRM and the plans of action that were formulated during the training. The 2 CRM instructors were available as consultants for a period of 1 day after the ICU had been trained. It was up to the ICU to decide on how to use this help (e.g., get organized, implement changes, reiterate theory). Furthermore, ICUs were encouraged to form a CRM change team. A detailed description of the training can be found in Kemper et al.13

Back to Top | Article Outline

Measurement and Participants


Twelve semistructured interviews were conducted to assess the progression of the implementation process. These interviews were held at 3 moments in accordance with the 3 stages of implementation: (1) just before the CRM training, (2) 4 to 6 months after the training, and (3) approximately 15 months after the training. The content of the interviews was based on the phases of Grol and Wensing’s14 implementation model. The first interview focused on the “orientation” phase, the first phase of the model, which included a raised awareness and interest for the intervention. The second interview focused on “change”—phase 4 of Grol and Wensing—in which CRM is tried or used. Objective data on the types and number of changes that were implemented were recorded and categorized for each ICU. The third interview focused on maintenance—the fifth phase of Grol and Wensing—in which the integration into the daily routine and sustainment of CRM are the central themes. The phases of insight and acceptance—phases 2 and 3 of Grol and Wensing—were not used for the interviews because these are part of the CRM training, rather than part of its implementation by the ICUs. All interviews were conducted and transcribed by the first author.

The participants were all implementation leaders in each phase of the implementation, being identified through the various contact moments related to the starting up of the study. For the preparation phase, the pioneers were interviewed. These were the persons who first brought up the topic of CRM and convinced staff and management to train the whole ICU. When the pioneers were not responsible or involved with the further implementation of CRM, a second interview was conducted with the person who had in fact prepared the implementation. For the stage after CRM training and plans, the chair of the CRM change team or the person responsible for the implementation of CRM projects was interviewed. Some persons fulfilled several of these roles (e.g., pioneer and chair of the change team) and were therefore interviewed several times. In total, 12 interviews were conducted with 8 persons (Table 1). Besides implementation leaders, regular participants were initially interviewed as well. The forthcoming information was, however, too detailed and intertwined with specific initiatives that it lost its value to describe the general implementation. Therefore, these interviews were not continued and not included in the present study.

The interviews were digitally recorded and were worked out as chronological narratives, which were then presented to the interviewed persons, so-called respondent validation,15 to minimize a biased interpretation of the interviewer. The narratives were then split up and categorized into the relevant stage of the implementation process, followed by a further subdivision across the topics within each stage (Table 1). The topics were based on aspects of the implementation model of Grol and Wensing and related implementation literature16,17 as well as previous CRM evaluations.11,18,19

Back to Top | Article Outline

End-of-Course Critique

The reaction of the participants was assessed using the mean score on the End-of-Course Critique (ECC). All participants were asked to fill out the ECC at the end of the training. The ECC was originally developed by Grogan and colleagues20 and adapted for use in the ICU. It measures the reaction immediately after the training, expressed as the extent of the perceived relevance and utility of the specific topics covered in the CRM training (e.g. “The session about ‘Human Factors’ was relevant and useful”). The ECC consists of 14 statements that are to be rated on a 5-point scale, varying from “strongly disagree” to “strongly agree.”

Back to Top | Article Outline

Plans of Action

Throughout the CRM training (i.e., for each of the topics covered), the participants were asked to write down concrete issues that they wanted to address in their ICU. These plans of action provide an insight into what kind of quality improvement initiatives CRM evokes.

Back to Top | Article Outline


Before the Training: Orientation

Table 2 shows that the 3 ICUs in the present study approached CRM in a similar way. All report a professional interest in quality and, in particular, CRM. The pioneers of all 3 ICUs indicated that they thought CRM was a new and promising opportunity for quality improvement. Their interest in quality improvement influenced the decision to place CRM on the agenda. In addition, all ICUs indicated that communication, whether it is multidisciplinary, between departments, or as part of teamwork, was something that could be improved. These 2 aspects provided the necessary momentum to initiate CRM training. The process of getting the green light for CRM in all 3 ICUs comprised several stages of convincing relevant stakeholders. All ICUs mention the costs of the training and staff hours as a barrier during this stage.

Back to Top | Article Outline

After the Training: Reaction (ECC) and Plans of Action

The participants rated the CRM training very positively in the ECC (mean, 4.47; SD, 0.45). In total, 94 different initiatives were formulated in the plans of action during the 14 training sessions of the 3 ICUs. Because reporting all initiatives would be too comprehensive, we focused on the ones that were mentioned by all 3 ICUs. These initiatives were most common at the unit level. The organizational initiatives concerned structural changes, such as assigning a coordinator of the day. The team level comprises initiatives that require teamwork, for instance, creating situational awareness by using a time-out procedure. The smallest number of common initiatives was found at the individual level. The participants indicated that they wanted to further develop nontechnical skills and to give feedback to each other. Table 3 summarizes those plans of action that were mentioned by all ICUs.

Back to Top | Article Outline

After the Training: Change

All 3 ICUs report that several initiatives were implemented. The number of implemented initiatives designed to promote quality was recorded and categorized. In addition to these planned changes, they also report alterations that were not part of any plan. A common theme is openness in communication because patient safety is discussed more in ICU A, there is a receptive environment for a debriefing in ICU B, and ICU C reports enhanced mutual communication.

The results of the interviews at this stage (Table 4) reveal that all 3 ICUs had a clear, yet different, vision on what they wanted to accomplish with CRM. This vision determined their follow-up process and resulting changes. The strategy of ICU A showed a strong resemblance to the system change framework of Kotter.22 In line with Kotter’s framework, the change team thoroughly prepared CRM, by first formulating a mission statement and formally involving the management. They communicated their vision to the personnel regarding the envisioned change by means of a weekly newsletter and used CRM training to empower the staff to act on this vision. After the training, they started implementing small adjustments, so-called quick wins, such as an extra whiteboard and the mounting of photos, to pave the way for larger projects.

Intensive care unit B started with the implementation of some of the easier initiatives that were mentioned in the plans of action. Their further implementation was characterized by a focus on the development of checklists for high-risk situations that were identified during the training. To develop these checklists, ICU B explicitly chose to integrate CRM for a large part into the Comprehensive Unit-based Safety Program (CUSP).23 The CUSP is a strategic framework comprising 5 steps specifically designed to reduce preventable harm through the development of tools, such as checklists. The CUSP has been successfully applied in the ICU.24 Intensive care unit B used the CRM training to educate the staff on safety science, which was the first step of their CUSP program. With the use of a prospective risk analysis method, ICU B developed several checklists for standard operating procedures and handovers.

The implementation by ICU C can be characterized with the social movement approach.25 Within this approach, implementation is an unstructured, self-organizing, and auto-catalyzing process. The commitment of the staff to the intervention is essential. In the eyes of the implementation leaders of ICU C, the CRM training elicited this kind of commitment. Especially, the development of a mutual CRM vocabulary helped the staff to express and address issues of quality and safety themselves. Besides allocating structural time to elaborate ideas, there was no structured guidance regarding which issues, derived from the plans of action or newly developed, should be chosen. Furthermore, no change team was put together to aid the implementation.

Although the ICUs differed in how they organized their changes, some resemblances were found. All ICUs used the plans of action as a starting point for change after the CRM training. After prioritizing and categorizing, these plans formed the input for the first initiatives. An overlap in themes was seen with regard to the implemented changes. All ICUs revised the role of the coordinating nurse and developed checklists, for instance, for the transport of a patient.

An important barrier, mentioned by all ICUs, was the lack of implementation knowledge and skills. For instance, how do you get and keep the staff involved, especially the less “CRM-enthused” part? This might explain why some initiatives were not well received, despite the bottom-up approach of the training, or perhaps the development of “implementation fatigue” played a role.

Back to Top | Article Outline

Plans: Maintenance

Intensive care units A and B indicated that they continued working with their change teams as a structural part of their unit. The goal of ICU A was to change the safety culture, whereas ICU B wanted to continue the development of checklists for high-risk situations. Intensive care unit C wanted to integrate the CRM mechanism of recognizing risks and addressing them, as a “normal” way of doing things, rather than as a “special” project. There were no concrete plans to structurally sustain CRM in ICU C. The results are summarized in Table 5.

Back to Top | Article Outline


The results of the present study demonstrate that all 3 ICUs successfully launched several initiatives, each using a different implementation strategy. Furthermore, all ICUs have taken several steps to sustain their approach for the foreseeable future. Despite the variety in strategies, 3 similarities can be seen between all 3 implementation processes that were crucial at the start of the implementation. First, all units reported problems with communication during the orientation phase. This acknowledgment of a performance gap is an indication that the participating ICUs, or at least the pioneers, possess a sense of urgency to change. This is an important first step for further implementation.16 Second, all ICUs allocated structural time for quality improvement before CRM and for change after CRM. Third, despite having different strategies, all units had a clear vision regarding their goals and strategies concerning CRM.

All ICUs indicated that they would use the plans of actions that were formulated during the CRM training as a starting point for their follow-up initiatives. The role of the coordinating nurse and the development of checklists are themes that recur in each ICU and are in line with CRM topics such as leadership and standardized communication.26 All ICUs mention the costs of CRM and a lack of implementation expertise as important barriers during the orientation and the change phase, respectively. The fact that CRM was perceived by the implementation leaders as a new and promising way to improve patient safety, as well as educating the whole staff, were regarded by all 3 ICUs as facilitating factors ro receive CRM training. Finally, all ICUs reacted very positively in the ECC.

The flexibility of the CRM follow-up initiatives provides opportunities but also creates pitfalls. An advantage of the flexibility is that the initiatives can be tailored to the specific situation and can be integrated in existing programs. For instance, Tapson et al27 used CRM training to successfully enhance the appropriate use of venous thromboembolism prophylaxis in surgery. A pitfall of the flexibility is that a large number of CRM initiatives can be overwhelming and may lead to implementation exhaustion. Box 1 presents suggestions proposed by the implementation leaders to get the best out of the initiatives.

Text Box 1

Suggestions of the participating ICUs

  • Be ready to solve problems, not just to identify them
  • Get advice from implementation experts
  • Formulate end goals and evaluate them
  • Start directly after the CRM
  • Schedule enough time in advance
  • Be aware that CRM can easily lose momentum

The diversity in follow-up initiatives could explain the mixed results of classroom-based CRM training on a behavioral level as reported by Rabol et al.5 First of all, as a result of the diversity, it is possible that an outcome that is used in an evaluation is not applicable for each site that is being studied. For instance, the use of a checklist is an often-used endpoint in CRM evaluation.5 When applied to the present study, we probably would have found a large effect in ICU B, a small effect in ICU A, and no effect in ICU C. This exemplifies the difficulty in defining an outcome that is applicable in all units, especially when the evaluation is designed even before the training, in accordance with scientific discourse.

Second, it can be questioned whether behavior in general is likely to change as a direct result of diverse, smaller interventions within the evaluation period of 1 year. It can be argued that these initiatives influence behavior by changing the safety culture. When the implementation of CRM initiatives is perpetuated over time, it will change the way people think about issues regarding safety and quality. This resonates in the social norms, which partly determine behavior.28 Behavioral change through culture takes much longer than 1 year; for instance, in aviation, it took approximately 40 years to gradually, but steadily, establish the safe culture that exists today.29 However, once established, the change will be very robust.

On the basis of the present study, we are unable to recommend one implementation strategy over the other. Although the ICUs share some similar initiatives (e.g., role of the coordinating nurse), the execution was always a bit different. Therefore, we could not compare the ICUs on the same endpoints and make valid assumptions on which strategy led to the best results. In addition, it can be questioned whether 1 implementation strategy would have led to similar results in all ICUs. The ICUs chose their strategy on the basis of a clear vision on what they wanted to achieve with CRM. Therefore, the strategies were highly dependent on the context to which they were applied. Existing literature also shows that context is an important aspect when determining which implementation strategy should be favored.30,31 However, more research on this topic would be an interesting venue for future studies.

Back to Top | Article Outline


The present article is an exploration to qualitatively study the implementation process of CRM. The number of ICUs is limited; therefore, the results should be considered preliminary. In addition, the interviewees were all actively involved with CRM; therefore, their perception might be slightly positively biased. To counteract this bias, the interviews were focused on objective results, such as which projects were actually implemented.

The external validity is limited by the number of ICUs that participated in the present study. Because this study was conducted as part of a larger effectiveness study,13 it was bound to the number of ICUs that participated in that study. Furthermore, participating in the effectiveness study required organizational and financial commitment, reflecting a willingness to receive CRM. This willingness should be taken into account when considering the external validity.

Back to Top | Article Outline


This research shows that CRM requires preparation and implementation, both of which require time and dedication. Consequently, it involves more than 2 days of training. The study illustrates that, despite the differences in vision concerning how to approach CRM, all 3 ICUs in the current research developed and implemented their own locally owned initiatives. The multitude and diversity in initiatives reflect the catalyzing effect of CRM on new and existing quality initiatives. Furthermore, it indicates that CRM helps participants to recognize, address, and handle safety issues. Finally, the diversity in initiatives may help explain the mixed results in outcomes in the present CRM evaluation research.

The results of the present study suggest that units that are considering CRM should base their strategy on a clear vision. The implementation strategy should probably be close to their own previous experience with the implementation of other projects. Structural time should be made available for preparation and implementation. The implementation should be tailored to the specific situation, depending on what goals are to be achieved using CRM.

All in all, it is promising to note that all 3 ICUs in the current research, despite their own barriers, visions, and strategies, developed multiple quality improvement initiatives and aim to continue doing so.

Back to Top | Article Outline


1. Zeltser MV, Nash DB. Approaching the evidence basis for aviation-derived teamwork training in medicine. Am J Med Qual. 2010;25:13–23.
2. Oriol MD. Crew resource management: applications in healthcare organizations. J Nurs Adm. 2006;36:402–406.
3. McConaughey E. Crew resource management in healthcare: the evolution of teamwork training and MedTeams. J Perinat Neonatal Nurs. 2008;22:96–104.
4. Flin R, O’Conner P, Crichton M. Safety at the Sharp End: A Guide to Non-technical Skills, Farnham, England: Ashgate; 2008.
5. Rabol LI, Ostergaard D, Mogensen T. Outcomes of classroom-based team training interventions for multiprofessional hospital staff. A systematic review. Qual Saf Health Care. 2010;19:e27.
6. McCulloch P, Mishra A, Handa A, et al. The effects of aviation-style non-technical skills training on technical performance and outcome in the operating theatre. Qual Saf Health Care. 2009;18:109–115.
7. Real K, Poole MS. Innovation implementation: conceptualization and measurement in organizational research. In: Woodman RW, Pasmoore WA, eds. Research in Organizational Change and Development. Bingley: Emerald Group Publishing Limited; 2005:63–134.
8. Koppelaar E, Knibbe JJ, Miedema HS, et al. Determinants of implementation of primary preventive interventions on patient handling in healthcare: a systematic review. Occup Environ Med. 2009;66:353–360.
9. Stead K, Kumar S, Schultz TJ, et al. Teams communicating through STEPPS. Med J Aust. 2009;190(suppl):S128–S132.
10. Morey JC, Simon R, Jay GD, et al. Error reduction and performance improvement in the emergency department through formal teamwork training: evaluation results of the MedTeams project. Health Serv Res. 2002;37:1553–1581.
11. Marshall DA, Manus DA. A team training program using human factors to enhance patient safety. AORN J. 2007;86:994–1011.
12. The Netherlands Society of Anaesthesiologists. Organisatie en werkwijze op intensive care-afdelingen voor volwassenen in Nederland [Organization and working method in the intensive care units for adults in the Netherlands]. Alphen aan den Rijn, The Netherlands: Van Zuiden Communication BV; 2006.
13. Kemper PF, De Bruijne M, Van Dyck C, et al. Effectiveness of classroom based crew resource management training in the intensive care unit: study design of a controlled trial. BMC Health Serv Res. 2011;11:304.
14. Grol R, Wensing M. What drives change? Barriers to and incentives for achieving evidence-based practice. Med J Aust. 2004;180(suppl):S57–S60.
15. Maxwell JA. Qualitative Research Design: An Interactive Approach. Thousand Oaks, CA: Sage Publications Inc; 2005.
16. Rogers EM. Diffusion of Innovations. 5th ed; New York: Free Press; 2003.
17. Greenhalgh T, Robert G, Macfarlane F, et al. Diffusion of innovations in service organizations: systematic review and recommendations. Milbank Q. 2004;82:581–629.
18. Taylor CR, Hepworth JT, Buerhaus PI, et al. Effect of crew resource management on diabetes care and patient outcomes in an inner-city primary care clinic. Qual Saf Health Care. 2007;16:244–247.
19. Sehgal NL, Fox M, Vidyarthi AR, et al. A multidisciplinary teamwork training program: the Triad for Optimal Patient Safety (TOPS) experience. J Gen Intern Med. 2008;23:2053–2057.
20. Grogan EL, Stiles RA, France DJ, et al. The impact of aviation-based teamwork training on the attitudes of health-care professionals. J Am Coll Surg. 2004;199:843–848.
21. Agency for Healthcare Research and Quality. TeamSTEPPS Instructor Guide. Rockville, MD: AHRQ; 2006.
22. Kotter JP. Leading change—why transformation efforts fail. Harv Bus Rev. 1995;73:59–67.
23. Pronovost P, Weast B, Rosenstein B, et al. Implementing and validating a comprehensive unit-based safety program. J Patient Saf. 2005;1:33–40.
24. Pronovost PJ, Berenholtz SM, Goeschel C, et al. Improving patient safety in intensive care units in Michigan. J Crit Care. 2008;23:207–221.
25. Bate P, Robert G, Bevan H. The next phase of healthcare improvement: what can we learn from social movements? Qual Saf Health Care. 2004;13:62–66.
26. Musson DM, Helmreich RL. Team training and resource management in health care: current issues and future directions. Harv Health Policy Rev. 2004;5:25–35.
27. Tapson VF, Karcher RB, Weeks R. Crew resource management and VTE prophylaxis in surgery: a quality improvement initiative. Am J Med Qual. 2011;26:423–432.
28. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50:179–211.
29. Sexton JB, Thomas EJ, Helmreich RL. Error, stress, and teamwork in medicine and aviation: cross sectional surveys. BMJ. 2000;320:745–749.
30. Buchanan D, Fitzgerald L, Ketley D, et al. No going back: a review of the literature on sustaining organizational change. Int J Manag Rev. 2005;7:189–205.
31. Jansson N. Organizational change as practice: a critical analysis. J Organ Change Manag. 2013;26:1003–1019.

Prescriptions for Bedtime Sedatives After the Introduction of a General Admission Order Set at an Academic Health Center: The Potential and Pitfalls of Order Sets

This study describes the impact of modifications to a general admission order set on physician prescribing of 2 as-needed or pro re nata (PRN) bedtime sedatives.

The hospitalists at our institution have used a general medical admission order set since 2005. Zolpidem was the only as-needed (PRN) bedtime sedative option on the order set until trazodone was added in December 2008. Trazodone is preferred over zolpidem in the geriatric population. We identified patients admitted by the hospitalists between January 2007 and August 2013 who were prescribed with either zolpidem or trazodone as a PRN sedative. Patient demographics, date and time of the order, and number of sedative doses administered during the hospitalization were recorded. Orders placed within 12 hours of admission were attributed to admission orders.

Between 2007 and 2013, the number of patients admitted by the hospitalists with an order for PRN trazodone on admission increased by 18-fold. During the same period, the number of admissions by the hospitalists increased by 2.3 times. Zolpidem orders exceeded those for trazodone in all age groups until 2008. After the addition of trazodone, its use exceeded that of zolpidem. Almost half (48%) of all patients did not have a dose of the PRN trazodone administered.

Although order sets can be leveraged to align practitioners with established guidelines, the expediency of using medications on an order set may overcome physicians’ clinical judgment. The content of an order set therefore deserves careful scrutiny before implementation.

From the *Indiana University Health Physicians; †Department of Medicine, IU School of Medicine; ‡College of Pharmacy and Health Sciences, Butler University; and §Indiana University Health; ∥Department of Pharmacy Practice, College of Pharmacy and Health Sciences, Butler University; and ¶Methodist Hospital, Indiana University Health, Indianapolis, Indiana.

Correspondence: Areeba Kara, MD, MS, FACP, Indiana University Health Physicians, Department of Medicine, IU School of Medicine, 1633 N Capitol Ave, Methodist Medical Tower Suite 640, Indianapolis, IN 46202 (e-mail:;

The authors disclose no conflict of interest.

The hospitalists at the Methodist Hospital are supported in their quality improvement efforts by a grant from the Methodist Health Foundation.

A.N.I. is now with the Department of Pharmacy Practice, College of Pharmacy, Purdue University, West Lafayette, and Eskenazi Health, Indianapolis, Indiana.

Supplemental digital contents are available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (

An Evaluation of Shared Mental Models and Mutual Trust on General Medical Units: Implications for Collaboration, Teamwork, and Patient Safety

This study examines nurse-physician teamwork and collaboration, a critical component in the delivery of safe patient care, on general medical units. To that end, we assess shared mental models and mutual trust, 2 coordinating mechanisms that help facilitate teamwork, among nurses and physicians working on general medical units.

Data were collected from 37 nurses and 42 physicians at an urban teaching medical center in the Northeastern United States. Shared mental model questionnaire items were iteratively developed with experts’ input to ensure content validity. Mutual trust items were adapted from an existing scale; items were reliable. Data were analyzed using χ2 and independent 2-tailed t tests.

Physicians and nurses reported significant differences in their perceptions of the professional responsible for a variety of roles (e.g., advocating for the patient [P = 0.0007], identifying a near miss/error [P = 0.003]). Medication reconciliation is only role for which nurses perceive less responsibility than physicians perceive nurses have. Regarding mutual trust, both groups reported significantly more trust within their own professions; both groups reported similar levels of trust in physicians, with physicians reporting significantly less trust in their nursing colleagues than nurses perceive (P

Although many efforts have been directed at improving nurse-physician collaboration, more work is needed. To that end, we propose increasing knowledge about their respective roles, providing opportunities for nurse and physician collaboration through rounding or committee work and enhancing the preparedness and professionalism of interactions.

From the *Purdue University, West Lafayette, IN; †Community Health Resources, Windsor, CT; ‡University of Massachusetts, Amherst, MA; and §Baystate Medical Center, Springfield; and ∥Tufts University School of Medicine, Boston, MA.

Correspondence: Sara A. McComb, PhD, Schools of Nursing and Industrial Engineering, Purdue University, Johnson Hall of Nursing, 502 N University St, West Lafayette, IN 47907-2069 (e-mail:

The authors disclose no conflict of interest.

Transparency When Things Go Wrong: Physician Attitudes About Reporting Medical Errors to Patients, Peers, and Institutions

Transparent communication after medical error includes disclosing the mistake to the patient, discussing the event with colleagues, and reporting to the institution. Little is known about whether attitudes about these transparency practices are related. Understanding these relationships could inform educational and organizational strategies to promote transparency.

We analyzed responses of 3038 US and Canadian physicians to a medical error communication survey. We used bivariate correlations, principal components analysis, and linear regression to determine whether and how physician attitudes about transparent communication with patients, peers, and the institution after error were related.

Physician attitudes about disclosing errors to patients, peers, and institutions were correlated (all P’s

Physician attitudes about discussing errors with patients, colleagues, and institutions are related. Several predictors of transparency affect all 3 practices and are potentially modifiable by educational and institutional strategies.

From the *Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA; †Department of Medicine, University of Washington, Seattle, WA; ‡Clinical Research Division, Biobehavioral Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA; §Department of Pediatrics, School of Medicine, University of Washington, Seattle, WA; and ∥Seattle Children’s Research Institute, Seattle, WA.

Correspondence: Sigall K. Bell, MD, BIDMC, Division of General Medicine and Primary Care, 330 Brookline Ave, CO 1309, 2nd Fl, Boston, MA 02215 (e-mail:

The authors disclose no conflict of interest.

This project was supported by grants from the Agency for Healthcare Research and Quality (HS016506, T.H.G.), the Robert Wood Johnson Investigator Award in Health Policy Research (T.H.G.), and the Greenwall Foundation (T.H.G.)

Development of a Medication Safety and Quality Survey for Small Rural Hospitals

We summarize the development and initial implementation of a survey tool to assess medication safety in small rural hospitals.

As part of an ongoing rural hospital medication safety improvement program, we developed a survey tool in all 13 critical access hospitals (CAHs) in Florida. The survey was compiled from existing medication safety assessments and standards, clinical practice guidelines, and published literature. Survey items were selected based on considerations regarding practicality and relevance to the CAH setting.

The final survey instrument included 134 items representing 17 medication safety domains. Overall hospital scores ranged from 41% to 95%, with a median of 59%. Most hospitals showed large variation in scores across domains, with 5 hospitals having at least 1 domain with scores less than 10%. Highest scores across all facilities were seen for safety procedures concerning high-alert or look-alike medications and the assembly of emergency carts. The lowest median scores included availability and consistent use of standardized order sets and the effective implementation of medication safety committees. Most hospitals used the survey results to identify and prioritize quality improvement activities.

The survey can be used to conduct a short medication safety assessment specific to a limited number of areas and services in CAHs. It showed good ability to discriminate medication safety levels across participating sites and highlighted opportunities for improvement. It may need modification if case mix or services differ in other states or if the status quo of medication safety in CAHs or related standards advance. The described process of survey development might be helpful to support such modifications.

From the *Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, †Department of Epidemiology, Colleges of Medicine and Public Health and Health Professions, ‡Department of Pharmacy, UF Health Shands Hospital, University of Florida, Gainesville; and §Florida Medical Quality Assurance, Inc, Tampa; and ∥Office of Rural Health, Florida Department of Health, Tallahassee, Florida.

Correspondence: Almut G. Winterstein, PhD, 1225 Center Dr, College of Pharmacy, University of Florida, Gainesville, FL 32611 (e-mail:

The authors disclose no conflict of interest.

This work was funded by a grant from the Florida Office of Rural Health in collaboration with the Florida Medical Quality Assurance, Inc (FMQAI).

Excerpts of this work have been presented at the American Society for Health Systems-Pharmacists Midyear Meeting in December 2014 in Orlando, Florida.

Supplemental digital contents are available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (