| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Corresponding author: Dr Milisa Manojlovich, University of Michigan School of Nursing, 400 N Ingalls, Room 4306, Ann Arbor, MI 48109-0482 (e-mail: mmanojlo{at}umich.edu).
 |
Abstract |
|---|
 Top Abstract ObjectivesMethodsResultsDiscussionReferences |
|---|
Objectives To determine the relationships between patients’ outcomes and (1) nurses’ perceptions of elements of communication between nurses and physicians and (2) characteristics of the practice environment.
Methods A cross-sectional survey design was used. Information on ventilator-associated pneumonia, bloodstream infection associated with a central catheter, and pressure ulcers was collected from 25 intensive care units in southeastern Michigan. Simultaneously, 462 nurses in those units (response rate, 53.3%) were anonymously surveyed. The Conditions for Work Effectiveness Questionnaire-II and the Practice Environment Scale of the Nursing Work Index were used to measure characteristics of the practice environment. The Intensive Care Unit Nurse-Physician Questionnaire was used to measure communication between nurses and physicians. Statistical tests included correlation and multiple regression. Analyses were conducted at the unit level.
Results Unit response rates varied from 6% to 100%. Together, variability in understanding communication and capacity utilization were predictive of 27% of the variance in ventilator-associated pneumonia. Timeliness of communication was inversely related to pressure ulcers (r= –0.38; P=.06), and workplace empowerment and scores on the Acute Physiology and Chronic Health Evaluation III were positive predictors of ventilator-associated pneumonia (R2=0.36; P=.005).
Conclusions Not all elements of communication were related to the selected adverse outcomes. The connection between characteristics of the practice environment at the unit level and adverse outcomes remains elusive.
Another way to minimize adverse outcomes is through nursing care processes such as communication. Nurses promote patients’ safety in part by communicating with physicians.8,9 Physicians and nurses vary in their perspectives of what constitutes good communication,10 a situation that makes it difficult to build consensus between the groups on how to improve that communication. Even when physicians and nurses were surveyed together, although they agreed that disruptive behavior of physicians was directly linked to nurses’ satisfaction, the 2 groups differed in their beliefs about responsibility, barriers to progress, and possible solutions to the problem.11
| Hospital environments with empowering social structures are linked to better nursing outcomes.
|
In the classic study on communication between nurses and physicians in intensive care units (ICUs), Shortell et al12 examined 4 elements of communication—openness, timeliness, accuracy, and understanding—because of the importance of the elements in contributing to effective communication in the complex, fast-paced ICU environment. A study that examines nurses’ perceptions of communication with physicians overall as well as specific elements of communication may provide better knowledge of how communication can be modified. Specific strategies to improve communication may be more effective than more general prescriptions. Before such specific interventions can be developed, a clearer understanding of nurses’ perceptions of elements of communication between nurses and physicians is warranted.
Knowledge of the relationships between the environment, nursing care processes, and specific outcomes is needed. According to the Nursing Role Effectiveness Model,13 which was used to guide this study, both characteristics of the practice environment and nursing care processes such as communication with physicians contribute to patients’ outcomes that are sensitive to nursing care. Outcomes related to patients’ safety such as pressure ulcers and nosocomial infections have been linked consistently to various aspects of nursing practice.14 Nosocomial pneumonia is the leading cause of mortality due to hospital-acquired infections,15 and hospital stays increase by 63% for patients with pressure ulcers.16 Therefore, we focused on 2 specific nosocomial infections, ventilator-associated pneumonia (VAP) and bloodstream infections associated with central catheters (BSI), and on the prevalence of pressure ulcers.
|
Objectives |
|---|
|
TopAbstractObjectivesMethodsResultsDiscussionReferences |
|---|
|
Methods |
|---|
|
TopAbstractObjectivesMethodsResultsDiscussionReferences |
|---|
Information on survey methods has been published elsewhere.17 Surveys were distributed to a total of 866 registered nurses. At the same time that surveys were being completed, data on other structure and outcome variables at the ICU level of analysis were also collected. Because survey distribution did not begin simultaneously at all sites, outcome data were collected for a 5-month period in each ICU from April through August 2005. This period covers the time frame for survey completion across the 25 ICUs (2
months) as well as the month before and the month after nurse surveys were conducted.
Study site coordinators, assigned to each participating hospital, were responsible for distributing surveys and collecting outcome data (ie, prevalence of pressure ulcers, VAP, and BSI). ICU data on mean length of stay, occupancy rate, patient days, and staffing variables also were collected to control for the influence of these factors on proposed relationships. The individual patients, as well as the participating nurses, units, and hospitals, all remained anonymous. The study was approved for protection of human subjects by the institutional review board of the University of Michigan and the institutional review boards of the other participating institutions.
Instruments
Workplace empowerment is defined as 4 social structures—opportunity, information, support, and resources—that are embedded in any work environment and, when accessed, are sources of power. Workplace empowerment was measured with 3 empowerment scales, which are used to assess various aspects of Kanter’s concept of empowerment.18 The 3 empowerment scales were the Conditions for Work Effectiveness, version II; the Job Activities Scale II; and the Organizational Relationships Scale II. The Conditions for Work Effectiveness, version II, is a 12-item scale with 4 subscales: opportunity, information, support, and resources. The Job Activities Scale II is a 3-item measure of Kanter’s concept of formal power; the Organizational Relationships Scale II consists of 4 items and is used to measure Kanter’s concept of informal power. Individual items for all scales ranged from 1 (none) to 5 (a lot). The mean of items in each of the 3 scales was calculated, then a total empowerment score was created by summing the 3 scale means (score range, 6–30). All 3 empowerment scales, 19 items in total, have established reliability19 and validity.20 The 
reliability of the combined empowerment scales was 0.92 in this study.
| Physicians and nurses vary in their ideas about what constitutes good communication.
|
Magnet-Hospital Properties
Five key domains of the nursing work environment are consistent with magnet-hospital properties and support professional nursing practice: nurses’ participation in hospital affairs; nursing foundations for quality care; nurse manager ability, leadership, and support of nurses; staffing and resource adequacy; and collegial nurse-physician relations.21 The Practice Environment Scale of the Nursing Work Index21 is based on magnet-hospital properties and uses 5 subscales to measure those key domains in the hospital environment. The Practice Environment Sub-scale uses a 4-point Likert-type scale ranging from 1 (strongly disagree) to 4 (strongly agree). The mean of item scores in each subscale was calculated. The overall scale score is the mean of the 5 subscale scores. Construct validity has been established, and confirmatory factor analysis supports the 5-subscale structure of the tool.21 The Cronbach for the overall scale has been reported as 0.82,21 but in this study, the reliability of the entire Practice Environment Scale of the Nursing Work Index was 0.95.
Communication Between Nurses and Physicians.
Communication was measured by using part of the ICU Nurse-Physician Questionnaire.22 The 4 scales used to measure communication between nurses and physicians were openness (4 items), accuracy (5 items), timeliness (4 items), and understanding of the communication that occurs between nurses and physicians (8 items). Openness refers to the extent to which nurses believe they can speak to physicians without fear of repercussions; accuracy refers to the extent to which nurses believe that the information conveyed to them by physicians is accurate; timeliness measures the degree to which nurses believe that information about patient care is relayed promptly to physicians; and understanding refers to the extent to which nurses believe that communication on the unit is effective and comprehensive.22 Answers for items in the scale ranged from 1 (strongly disagree) to 5 (strongly agree). The 4 scales were scored by taking the mean score of the items in each scale. The overall communication score is the mean of the scores for the 4 scales. In addition, the variability in communication was calculated as the standard deviation of responses within each of the 4 scales. The overall scale has adequate evidence of good reliability and validity.22 For this study scale, reliability was 0.92. The Cronbach ’s for the subscales were 0.75 for accuracy, 0.90 for openness, 0.58 for timeliness, and 0.90 for understanding.
Patients’ Outcomes Operational definitions from the National Quality Forum were used for all outcomes.23 For pressure ulcers, the numerator was the number of ICU patients with a stage II or greater pressure ulcer that was not present on admission to the ICU. The denominator was the total number of patients in the unit on the day of the prevalence study. For VAP, the numerator was the number of cases of VAP and the denominator was the total number of ventilator days for a unit. Finally, for BSI, the numerator was the number of BSI associated with central catheters on a unit (as confirmed by laboratory results or clinical sepsis) and the denominator was the total number of central-catheter days for that unit.
| As the timeliness of communication increased, the prevalence of pressure ulcers decreased.
|
Information on the staffing mix (the ratio of registered nurses to nonlicensed nursing personnel), hours of nursing care provided per patient day (HPPD), and capacity utilization in each ICU was collected to explore the possibility that these variables might influence variability in patients’ outcomes. Capacity utilization is the ratio of occupancy rate to length of stay and was constructed from unit occupancy rate as the numerator and length of stay as the denominator. Units with high occupancy rates and short lengths of stay are indicative of greater efficiency (better capacity utilization) from an administrative point of view. From a nursing perspective, however, an ICU with a high occupancy rate and short lengths of stay involves more nursing work than does an ICU with a low occupancy rate and longer lengths of stay, because nurses admit and discharge patients more frequently. Better capacity utilization may be associated with fewer adverse events, but only when staffing is adequate.24
Procedures
Rather than depending on each unit to provide its own outcome data, members of the research team constructed control variables and outcome data from numerators and denominators requested from each unit, so that measures were uniform across all sites. For example, HPPD is a staffing variable that can be defined and constructed differently across sites.25 Without common definition and measurement, it would not be possible to compare HPPD measures or to use HPPD in analyses of data from all 25 ICUs. Information on VAP, BSI, and pressure ulcers was available from administrative databases because data on these outcomes are routinely collected for the Joint Commission and the National Quality Forum.
In order to control for patient severity, a risk adjustment measure was included in the data analysis. Severity-adjustment methods have not been standardized across the health care industry, and in at least 1 study,26 researchers found that mortality rates across hospitals varied by type of severity-adjustment method used. Therefore, research assistants computed scores on the Acute Physiology and Chronic Health Evaluation III (APACHE III) for a random sample of patients from 24 of the ICUs during all 5 months of data collection (April through August 2005). A total of 1090 patients’ charts were randomly sampled, a mean of about 45 charts from each ICU. The APACHE III value for one ICU was imputed on the basis of the mean APACHE III score for participating ICUs in that hospital. APACHE III scores range from 0 to 299, with higher scores indicative of a higher risk of hospital death.27
Data were aggregated by unit. The ICU was chosen as the unit of analysis for several reasons. First, patients’ data were provided by the hospital at the ICU level rather than at the lower patient level. Second, data from the nurses were collected from individual nurses but were aggregated up to the ICU level by computing mean scores across the nurses in an ICU, so that appropriate relationships between nurse-generated and outcome variables could be modeled. Finally, the impact of nurse staffing is most direct at the unit level.28
All data were examined for accuracy as soon as they were submitted, so that errors were corrected and outliers verified immediately. Data cleaning was done after each data entry session by checking frequencies and descriptives for discrepancies. Scatterplots, skewness, and kurtosis were examined to determine the shape of the data distribution. On the basis of this information, data were determined to be fairly normally distributed, so no transformations were required.
Pearson correlations in the form of a correlation matrix were used to estimate associations between pairs of predictors (workplace empowerment, magnet-hospital properties, communication), control variables (APACHE III scores, capacity utilization, staffing), and outcomes (VAP, BSI, pressure ulcers). Multiple regression was used to estimate models when significant correlations were found. Pearson’s correlations do not indicate direction of causality in the relationship between 2 variables, but by using multiple regression it was possible to predict the probability of an outcome variable given several predictor or control variables.29
|
Results |
|---|
|
TopAbstractObjectivesMethodsResultsDiscussionReferences |
|---|
2 tests, and the groups did not differ significantly on most variables. All 25 ICUs were included to conserve data. It was not possible to get any information about nurses who chose not to respond to the survey, because respondents’ anonymity had been assured. The mean percentage of nurses in each unit responding to the survey was 57.5% (SD, 18.0%; range, 6%–100%).
Overall, 462 of 866 nurses (53.3%) completed usable surveys. The nurses in the sample were mainly female (84%) and white (78%), ranging in age from 22 to 64 years (mean, 39.3). The nurses had a mean of 13 years of nursing experience, had spent a mean of 10 years in their institutions, and a mean of about 8 years in ICUs. Most nurses were educationally prepared at the baccalaureate level (59%); the rest were prepared at associate’s (30%), diploma (8%), or master’s (3%) levels. A total of 17% of the sample had specialty certification. Summary statistics of the main study variables are presented in Table 1
.
|
).
|
Because several associations were significant, multiple regression models were generated to test possible predictors of the 3 outcomes (VAP, BSI, and pressure ulcers). APACHE III scores and capacity utilization emerged as control variables that were significantly associated with VAP in bivariate tests. A regression model is also presented for pressure ulcers and communication timeliness, even though the bivariate association was not significant. HPPD was positively associated with BSI (r = 0.46, P = .02), but because BSI did not correlate significantly with any other predictor, it was not modeled by using regression. Neither staffing mix nor HPPD was associated with any other outcome. Each regression model was limited to 2 independent variables because of the small sample (25 ICUs), so various combinations of independent variables were entered in models 1 through 3 in Table 3. Model 2 explained about 27% of the variance in VAP, with the predictors of variability in understanding communication and capacity utilization.
|
. Models 4 and 5 both predicted 36% of the variance in VAP. Combinations of workplace empowerment, APACHE III scores, and capacity utilization were included as predictors in these models. |
Discussion |
|---|
|
TopAbstractObjectivesMethodsResultsDiscussionReferences |
|---|
Sense making is a theoretical framework31 that is more closely aligned with communication than is the Nursing Role Effectiveness Model. Sense making is an iterative process with a communication component that is used in high-reliability organizations and that cuts across hierarchical boundaries. The purpose of sense making is to foster understanding between parties, so that failures are averted. In some high-risk contexts, people are being trained to improve communication and avoid failures by using a 5-step process known as STICC.32 The 5 components of STICC are situation (Here’s what I think we face), task (Here’s what I think we should do), intent (Here’s why), concern (Here’s what we should keep our eyes on), and calibrate (Now talk to me: tell me if you don’t understand, cannot do it, or see something I do not).32 Although other communication tools such as SBAR (situation, background, assessment, recommendation) are also available, application of the STICC protocol leads to sense making between communicators because it requires feedback.
Sense making is dynamic, in part because of the "calibration" element that permits change and deepening understanding between communicators. Other communication tools currently in vogue ask individuals to come to a decision when they communicate. In decision making, communicators feel ownership for "their" decision, tend to defend that decision, and often do not listen to those who question the decision.33 Because nurses often use silence instead of voice in dealing with physicians,34 adoption of the STICC protocol may allow nurses to be better heard, resulting in improved outcomes for patients through more effective communication.
Variability in understanding communication, as expressed by the standard deviation of the understanding subscale, together with capacity utilization, explained about 27% of the variance in VAP. To have 2 variables predictive of such a large percentage of VAP is stunning and requires that action be taken to decrease variation in understanding communication. One potential strategy may involve the use of a communication protocol such as STICC that improves understanding by insisting on a back-and-forth iterative dialogue between communicators.
Another potential strategy to decrease variability in understanding may be to standardize VAP treatment protocols, although this strategy may be difficult to implement. Causes of VAP differ across populations of patients and type of ICU, because certain types of microorganisms are more prevalent in some populations and ICUs than in others.35 Many cases of VAP are polymicrobial, a situation that further complicates treatment options.15 No single plan for preventing VAP is recommended, however. In a recent evidence-based review36 of strategies used to prevent VAP, researchers recommended additional trials of all preventive strategies. Consistency in understanding communication may best arise from protocols such as STICC, given the current state of the science on prevention and treatment of VAP.
The inverse relationship between timeliness of communication and pressure ulcers trended toward significance, and although nurses independently treat pressure ulcers, timely consultation with physicians may help in preventing pressure ulcers. For example, poor nutrition is a significant risk factor for pressure ulcers,37 yet feeding of critically ill patients is often withheld unnecessarily.38 Even when ordered, enteral feedings are often interrupted as a precautionary measure to decrease the risk of aspiration, resulting in underfeeding.39 Timely communication might consist of notifying the physician during rounds about the number of hours of nonfeeding that have accumulated for a specific patient. A similar communication strategy, one that increases physicians’ awareness on a patient care issue, has been effective in reducing the incidence of urinary tract infections in patients with indwelling urinary catheters.40
| Practice environment characteristics did not contribute to adverse outcomes in the expected manner.
|
None of the outcome variables in our study were correlated with each other, suggesting that at least some of the outcomes chosen for the study may not be particularly sensitive to nursing care. When examined at the unit level, select outcomes that are indicators of the quality of care should be inter-correlated with each other41 because of the influence of nursing care on those outcomes.7 Outcome variables that are closely related to patient acuity measures rather than to each other are indicative of patients’ severity of illness and not of the quality of nursing care.41
We found a significant inverse association between VAP and patient acuity, as measured by APACHE III scores (r = –0.40, P = .05). This finding suggests that VAP may not be a good indicator of the quality of nursing care on a unit. Health care providers other than nurses (ie, respiratory therapists and physicians) often manipulate and manage ventilators, but because no data on respiratory therapists or physicians were collected, the influence of these providers on VAP rates could not be determined. The inverse association between VAP rates and APACHE III scores may initially seem counterintuitive: sicker patients tend to receive mechanical ventilation. However, a possible explanation is that sicker patients who are receiving mechanical ventilation may not live long enough for VAP to develop. This supposition was borne out by data indicating a significant positive correlation between VAP and length of stay (r = 0.25, P = .006).
| As nurses access more information, support, resources and opportunities, they become more effective in their roles.
|
We found no significant relationship between magnet-hospital properties and outcomes, although in the classic study,42 magnet hospitals had lower mortality rates than did other hospitals. We measured magnet-hospital properties, but none of the hospitals in our study was magnet certified. Some other work environment factor may be present in magnet-certified hospitals that we did not measure (eg, flat organizational structure). Perhaps the outcomes in our study are not sensitive to magnet-hospital properties. Although strongly related to nurse-sensitive outcomes, workplace empowerment and magnet-hospital properties have not been linked to patients’ outcomes at the unit level, where nursing care has its greatest impact.28
The strong, positive relationship between work-place empowerment and VAP is difficult to explain. Having 36% of all possible variation in VAP explained by workplace empowerment and APACHE III scores is intriguing. Theoretically, workplace empowerment should be inversely related to adverse outcomes: as nurses access more information, support, resources, and opportunities on the job, they become more effective in their roles, a situation that helps improve outcomes. In a recent study,43 researchers found that nurses’ perceptions of a positive organizational climate were associated with higher odds of BSI, another unexpected relationship. Although previous research has shown the importance of empowerment to nurses’ outcomes,44 our study is the first to try to link workplace empowerment to actual patients’ outcomes. VAP may not be a nursing-sensitive outcome, as discussed earlier, and other variables to which VAP may be more sensitive were not included in the study.
The positive association between HPPD and BSI, although unexpected, is similar to findings in some staffing studies. Recent studies have indicated a relationship between nurse staffing variables and nosocomial infections, providing evidence that the quantity of nurses makes a difference to adverse outcomes. Both Cho et al45 and Blegen et al28 found positive relationships between HPPD and pressure ulcers. Interestingly, these 2 studies showed the expected inverse relationship between registered nurses’ HPPD and pressure ulcers, suggesting that the unique contribution of registered nurses was able to make a difference in the care of patients with pressure ulcers. However, in our study, a positive relationship between registered nurses’ HPPD and BSI remained (r = 0.44, P = .03). As is true for mechanical ventilation, other health care disciplines are involved in the insertion and management of central catheters, so BSI may not be as sensitive to nursing care as are pressure ulcers.
Our study had several limitations. The sample size of 25 ICUs is a limitation, reducing the power of the statistical tests and increasing the possibility of a type II error (failure to find significant relationships). Repeated testing of multiple models increased the chance of type I error, but the small sample size decreased that chance. Perhaps the influence of magnet-hospital properties and communication between nurses and physicians on patients’ outcomes is not large enough to be evident in a sample of only 25 ICUs; perhaps a larger sample of ICUs would be needed for significant findings. A related limitation is low response rates (<50%) on 4 units, although these 4 units did not differ from the units that had higher response rates on most of the variables analyzed.
The facts that a convenience sample was used and that all hospitals were located in the same geographic region pose another limitation. Thus, findings cannot be generalized beyond this area in Michigan. The cross-sectional nature of the study indicates that cause-and-effect statements cannot be made, because longitudinal research would be needed to assess causality. The low internal consistency of the timeliness subscale may have adversely affected results. Because of the broad range of issues on timeliness represented in the 4 items, the low coefficient is not surprising. Finally, evaluating many bivariate associations and testing many regression models increase the likelihood of finding spurious associations. However, the small sample size reduces the likelihood of significance and so reduces the likelihood of such error.
Findings from this study hint at the place to look for answers to the question that plagues us: How can we prevent adverse outcomes in critically ill patients? Results of analysis of the communication subscale suggest that with a larger sample, significant influences of communication on pressure ulcers might have been found. However, engaging in communication with physicians is just one of many processes or functions that nurses perform while caring for patients. Processes other than communication may be more closely linked to adverse outcomes, but specific nursing care processes that improve or worsen patients’ outcomes have rarely been studied.46 The problem of adverse outcomes continues,47 but through studies such as ours, a few answers are beginning to emerge.
|
ACKNOWLEDGMENTS |
|---|
FINANCIAL DISCLOSURES
This study was funded by the Blue Cross Blue Shield of Michigan Foundation.
Now that you’ve read the article, create or contribute to an online discussion on this topic. Visit www.ajcconline.org and click "Respond to This Article" in either the full-text or PDF view of the article.
|
REFERENCES |
|---|
|
TopAbstractObjectivesMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
