Patient Mortality, Staff Resources and Workload in the ICU
Patient Mortality, Staff Resources and Workload in the ICU
This multicenter study proposes evidence-based thresholds of five patients to two nurses and 14 patients to one physician, above which there is an increase in ICU mortality. Those shifts with inadequate staffing resources, given the patients' needs, occurred mostly during weekends for nurses and at nights for physicians. In addition, higher risk of death was strongly influenced by heavy workload during shifts based on increased patient turnover and volume of LSPs performed by ICU teams.
Although some subsets of these parameters have been explored previously, the literature is scarce regarding the shift-by-shift analysis of both staffing and workload measures in a multicenter setting. Studies are traditionally based on fixed levels of staff (ie, ratios fixed a priori for periods of a few months), instead of considering daily staff variations. This lack of granularity may explain why there is currently inconsistent association between medical staffing and patient outcome. In agreement with the guidelines of the Society of Critical Care Medicine for safe care, the present results clearly highlight a threshold effect regarding medical staff size relative to the number of patients and their needs. The present results also support previous observations, suggesting a potential relationship between ICU mortality and nurse staffing.
This study opens the way to an automated monitoring system. All types of data computed in the present work were collected routinely. Therefore, automating the process to provide a continuous follow-up of the adequacy of staffing levels and workload is possible. Such a monitoring tool would help manage staffing adequately and optimize patient flow. However, using routinely collected data to investigate preventable deaths caused by failures in ICU organization have clear limitations. In addition to excluding deaths with DFLST orders from our dataset, a solution would be to collect specific causes of death, such as "failure to rescue," which may reflect an unbalanced staffing level. In addition, we primarily used a combination of patient turnover and LSPs to assess work intensity at the team level. In the studied ICUs and in the majority of French ICUs, there is no consult team to take care of less-sick patients in ICUs. The same team is in charge of new admissions and other patients at the same time. So the observed workload is the sum of the patients in the ICU and new admissions. Representing the workload as a combination of LSPs, patient severity and turnover allowed us to take into account both patients present in the ICU and new admissions. Tracking the caregivers' well-being and how they are experiencing the burdens of daily activities may provide additional information. Furthermore, several nursing workload scores have been previously developed, such as the therapeutic intervention scoring system, the nursing activities score, or the nine equivalents of nursing manpower use score. Unfortunately, these metrics were not present in available databases.
In terms of generalizability, this study was performed over eight closed ICUs in four academic hospitals. Despite a limited sample size, we think that the findings can probably be generalized to the other French academic hospitals given that their organization does not vary much. Also, our analyses showed no influence of the number of residents per physician on patient mortality. Therefore, we can argue that our findings may also apply to nonacademic hospitals. Although any ICU with an organization similar to the ICUs from this study could benefit from the present results, it would be interesting to validate our findings although replication studies in other countries. The optimal P/P ratios may be different in the context of open ICUs, where the physician formally responsible for the patient is not the intensivist and physicians from outside of the ICU may participate in patient care. Another limitation to this study is that no adjustment was feasible regarding the specialty of ICU physicians (ie, intensivist, anesthesia, and mixed) that may have influenced patients' outcomes.
Representing a real picture of daily workload in the ICU, this study raises further unresolved questions. What are the exact conditions of excessive workload and insufficient staffing that lead to avoidable deaths in the ICU? Ideally, investigating shift-to-shift variations of caregivers staffing and patient turnover would allow identification of which caregiver is assigned to a given patient at any time in a particular ICU. Here, we provided this information at the unit level at each time period. The next step would be to introduce the linking of individual data between patients and caregivers, allowing for a dynamic analysis of their interactions. Indeed, workload may not be uniformly distributed over time across different team members. For example, two patients assigned to the same caregiver may need urgent care, whereas other caregivers might simultaneously experience a lower workload. In this situation, it is likely that the latter helps the former. A solution to this issue was proposed in some ICUs. Teams dedicated to managing new ICU admissions have been implemented in a delimited ICU zone. The performance of such organizations, which aim to prevent ICU malfunction that results from excessive turnover, should be assessed. Furthermore, what are the determinants of clinical team performance, and how can we make efficient teams? Quantifying the patient-to-caregiver ratio in real time provides an overall view of the appropriate staffing level. A more accurate evaluation of the capability of a team to properly handle difficult situations represents the next step. Analysis of individual characteristics and interactions among team members should be considered because team composition and familiarity might influence its resilience to intense workload variations. Thus, high-performance teams would maintain high levels of quality when exposed to stressful situations, and teamwork skills may surpass the sum of individual talent. Staff experience, or the number of shifts involving the same colleagues, may reflect expertise and how well people communicate with each other through the acquisition of skills that allow for quick responses that can guarantee patient safety. In the same manner, safety culture in the team may play a role in patient safety. Methods such as crew resource management imported from aviation were implemented in surgical settings. Team training might be useful to improve patient outcome in ICUs.
This study proposes evidence-based ratios of patients per nurse and physician in the context of ICUs. Our findings support recommendations for adapting caregivers' resources to patients' needs in real time. Insufficient staffing above the observed maximum thresholds showed an increased risk of mortality. Particular attention should be paid to critical periods identified to be at risk of high patient-to-caregiver ratios (ie, on weekends for nurses and at night for physicians). Moreover, identification of patient turnover as an independent risk factor of mortality should lead to a thoughtful management of patient influx during a single shift. Delaying admissions during periods when teams are experiencing a heavy workload with unbalanced patients-to-caregivers ratios could prevent ICU disorganization. However, the heterogeneity staffing patterns in ICUs around the world cannot be overlooked: larger studies involving different countries will be needed to validate these findings. Because all data used in this study were routinely collected in hospital information systems, real-time monitoring of staffing levels and workload with dedicated alarms is feasible. Such monitoring of patient-to-caregiver ratios would help not only to have sufficient resources for guaranteeing patient safety when needed but also to avoid wasting in case of temporary overstaffing. Hence, continuous balancing between staffing resources and workload may increase care efficiency in ICUs. Otherwise, a cost-effective solution would consist of smoothing activity and staff presence over time according to threshold recommendations.
Discussion
This multicenter study proposes evidence-based thresholds of five patients to two nurses and 14 patients to one physician, above which there is an increase in ICU mortality. Those shifts with inadequate staffing resources, given the patients' needs, occurred mostly during weekends for nurses and at nights for physicians. In addition, higher risk of death was strongly influenced by heavy workload during shifts based on increased patient turnover and volume of LSPs performed by ICU teams.
Although some subsets of these parameters have been explored previously, the literature is scarce regarding the shift-by-shift analysis of both staffing and workload measures in a multicenter setting. Studies are traditionally based on fixed levels of staff (ie, ratios fixed a priori for periods of a few months), instead of considering daily staff variations. This lack of granularity may explain why there is currently inconsistent association between medical staffing and patient outcome. In agreement with the guidelines of the Society of Critical Care Medicine for safe care, the present results clearly highlight a threshold effect regarding medical staff size relative to the number of patients and their needs. The present results also support previous observations, suggesting a potential relationship between ICU mortality and nurse staffing.
This study opens the way to an automated monitoring system. All types of data computed in the present work were collected routinely. Therefore, automating the process to provide a continuous follow-up of the adequacy of staffing levels and workload is possible. Such a monitoring tool would help manage staffing adequately and optimize patient flow. However, using routinely collected data to investigate preventable deaths caused by failures in ICU organization have clear limitations. In addition to excluding deaths with DFLST orders from our dataset, a solution would be to collect specific causes of death, such as "failure to rescue," which may reflect an unbalanced staffing level. In addition, we primarily used a combination of patient turnover and LSPs to assess work intensity at the team level. In the studied ICUs and in the majority of French ICUs, there is no consult team to take care of less-sick patients in ICUs. The same team is in charge of new admissions and other patients at the same time. So the observed workload is the sum of the patients in the ICU and new admissions. Representing the workload as a combination of LSPs, patient severity and turnover allowed us to take into account both patients present in the ICU and new admissions. Tracking the caregivers' well-being and how they are experiencing the burdens of daily activities may provide additional information. Furthermore, several nursing workload scores have been previously developed, such as the therapeutic intervention scoring system, the nursing activities score, or the nine equivalents of nursing manpower use score. Unfortunately, these metrics were not present in available databases.
In terms of generalizability, this study was performed over eight closed ICUs in four academic hospitals. Despite a limited sample size, we think that the findings can probably be generalized to the other French academic hospitals given that their organization does not vary much. Also, our analyses showed no influence of the number of residents per physician on patient mortality. Therefore, we can argue that our findings may also apply to nonacademic hospitals. Although any ICU with an organization similar to the ICUs from this study could benefit from the present results, it would be interesting to validate our findings although replication studies in other countries. The optimal P/P ratios may be different in the context of open ICUs, where the physician formally responsible for the patient is not the intensivist and physicians from outside of the ICU may participate in patient care. Another limitation to this study is that no adjustment was feasible regarding the specialty of ICU physicians (ie, intensivist, anesthesia, and mixed) that may have influenced patients' outcomes.
Representing a real picture of daily workload in the ICU, this study raises further unresolved questions. What are the exact conditions of excessive workload and insufficient staffing that lead to avoidable deaths in the ICU? Ideally, investigating shift-to-shift variations of caregivers staffing and patient turnover would allow identification of which caregiver is assigned to a given patient at any time in a particular ICU. Here, we provided this information at the unit level at each time period. The next step would be to introduce the linking of individual data between patients and caregivers, allowing for a dynamic analysis of their interactions. Indeed, workload may not be uniformly distributed over time across different team members. For example, two patients assigned to the same caregiver may need urgent care, whereas other caregivers might simultaneously experience a lower workload. In this situation, it is likely that the latter helps the former. A solution to this issue was proposed in some ICUs. Teams dedicated to managing new ICU admissions have been implemented in a delimited ICU zone. The performance of such organizations, which aim to prevent ICU malfunction that results from excessive turnover, should be assessed. Furthermore, what are the determinants of clinical team performance, and how can we make efficient teams? Quantifying the patient-to-caregiver ratio in real time provides an overall view of the appropriate staffing level. A more accurate evaluation of the capability of a team to properly handle difficult situations represents the next step. Analysis of individual characteristics and interactions among team members should be considered because team composition and familiarity might influence its resilience to intense workload variations. Thus, high-performance teams would maintain high levels of quality when exposed to stressful situations, and teamwork skills may surpass the sum of individual talent. Staff experience, or the number of shifts involving the same colleagues, may reflect expertise and how well people communicate with each other through the acquisition of skills that allow for quick responses that can guarantee patient safety. In the same manner, safety culture in the team may play a role in patient safety. Methods such as crew resource management imported from aviation were implemented in surgical settings. Team training might be useful to improve patient outcome in ICUs.
This study proposes evidence-based ratios of patients per nurse and physician in the context of ICUs. Our findings support recommendations for adapting caregivers' resources to patients' needs in real time. Insufficient staffing above the observed maximum thresholds showed an increased risk of mortality. Particular attention should be paid to critical periods identified to be at risk of high patient-to-caregiver ratios (ie, on weekends for nurses and at night for physicians). Moreover, identification of patient turnover as an independent risk factor of mortality should lead to a thoughtful management of patient influx during a single shift. Delaying admissions during periods when teams are experiencing a heavy workload with unbalanced patients-to-caregivers ratios could prevent ICU disorganization. However, the heterogeneity staffing patterns in ICUs around the world cannot be overlooked: larger studies involving different countries will be needed to validate these findings. Because all data used in this study were routinely collected in hospital information systems, real-time monitoring of staffing levels and workload with dedicated alarms is feasible. Such monitoring of patient-to-caregiver ratios would help not only to have sufficient resources for guaranteeing patient safety when needed but also to avoid wasting in case of temporary overstaffing. Hence, continuous balancing between staffing resources and workload may increase care efficiency in ICUs. Otherwise, a cost-effective solution would consist of smoothing activity and staff presence over time according to threshold recommendations.
Source...