RRT activation rates
RRT activation rates were analysed for seven of the nine hospitals (figure 1). Two hospitals were unable to provide data on the monthly admissions and RRT calls. The average RRT activation rate per hospital over 2016 varied between 12 to 68 per 1000 admissions. All hospitals showed lower call rates after hours (18:00-08:00) and during 24-hour weekend days (8 to 33 per 1000 admissions) which was comparable with about twothirds of total RRT calls. Figure 1 (upper panel) shows the average number of RRT calls per hospital. Hospital 2 had the highest RRT call rates (surgical 64, medical 73 and after hours 60 and 59 per 1000 admissions, respectively) compared with the other hospitals. In the other six hospitals, RRT call rates were similar and ranged between 8 and 25/1000 admissions. After hours, this ranged between 7 and 16 per 1000 admissions. The lower panel in figure 1 indicates the highest and lowest average RRT call rate per specialty on a monthly basis. Irrespective of specialty, hospital 2 had the highest utilisation of RRT services. Minimal average use was 7.5/1000 admissions and highest 86/1000.

Discussion

After mandatory introduction of RRS in the Netherlands in 2008, mortality and patient harm significantly decreased.[5] The current study shows that the reported uptake of RRS is high and escalation of care is appropriate within the two-tier system. The main outcomes of the study point towards increased patient-tonurse ratios in the majority of studied hospitals with over 84% of nurses taking care of more than eight patients, and often up to 15 patients, during night shifts. The physical and mental strain of work is deemed relatively balanced by both nurses and registrars but increasing to higher workloads during shift (i.e. off-hours) work. Self-reported adherence to escalation of care is high despite the workload, although RRT activations reflect an almost 50% reduction of after-hours RRT activation rate (8-33/1000 versus 12-68/1000 admissions). Interestingly, the hospital with the highest nurse staffing also showed the highest rate of RRT activation. Unfortunately, within the current study, we were unable to correlate RRT activation rates with clinical outcome.

Statistically significant differences in RRT call rates were observed between hospitals (two hospitals were unable to provide the data). Hospital 2 showed consistently higher activation rates together with generally lower patient-tonurse ratios compared with the other hospitals. The number of RRT activations per 1000 admissions has been designated as ‘RRT dose’ and increasing ‘the dose’ is directly related to better outcomes. In this context, a dose of over 40 per 1000 admissions is deemed optimal.[12] The degree to which this can be extrapolated to the Dutch situation is not precisely known, because case-mix, trigger and activation protocols for RRTs are different and therefore care should be taken with these extrapolations.

As an RRS is built from multiple components, focusing solely on the efferent arm is an atomistic approach. Before the escalation to the RRT, a multitude of processes occur. Therefore, the effectiveness of the entire system is influenced by many factors and each step is as important as the previous one. As previously shown, delayed recognition of deterioration and activation of the RRT is associated with poor clinical outcome.[13] Potential causes for system delays have been attributed to impaired recognition, failure to escalate, lack of clinical urgency and monitoring failures.[9,14,15] As also observed in our study, diurnal patterns have been observed for RRT activation rates. Activation rates out of office hours are lower compared with daytime and are associated with significant clinical burden.[16,17] Unfortunately, our study was not designed to analyse correlation of RRT activation rates with clinical outcome. Interestingly, higher patient-to-nurse ratios at night may be an important aspect which may be related to reduced recognition of deteriorating patients and thus declining activation rates at night. The landmark study by Aiken et al. in 2002 reported that in hospitals with high patient-to-nurse ratios, surgical patients had increased (risk-adjusted) 30-day mortality and failure-to-rescue rates. After adjustment, each additional patient per nurse increased the likelihood of death within 30 days of admission by 7% (OR 1.07, 95% CI 1.03-1.12).[10] A second study by the same group identified a correlation between the degree of education of nursing staff and clinical outcome in nine European countries.[18] Later published studies have shown similar outcomes, adding to the burden of evidence.[19] Through which specific interaction this occurs, remains unknown. One possibility, tightly linked to the failure-of-rescue principle and described in the above papers is the late detection of complications and clinical deterioration.[20,21] Therefore, increased patient-to-nurse ratios could impede appropriate and timely response and have previously been shown to be linked to monitoring failures.[15]

After appropriate triggering, adequate escalation and management is required. As described previously, ward teams have an opportunity to manage their patients after which RRT can be activated. Inefficiency in management and inappropriate escalation has been referred as ‘clinically futile cycles’.[22] This concept refers to periods in which patient adverse events occur not because of ignorance or inability, but solely because healthcare providers apply unwritten rules and cultural practices delaying effective and efficient care. Examples of these were shown in this study (table 4) in which the RRT is not always called, especially by registrars and consultants who do not appropriately escalate in the first place. This circular way of thinking and acting prevents staff from effectively managing the clinical problem, with a potential delay or absence of RRT activation. Reasons for this futility remain not fully elucidated and in accordance with Shearer et al., a small subset of healthcare providers still revert to either a colleague or consultation of another ‘organ’ specialist.[23] Overall, and within the context of this study, there has never been a reliable comparison of the ‘dose’ of the RRT between various hospitals, countries or hospital cultures. As seen in the current study, only one of the hospitals (hospital 2) is subsequently deemed ‘mature’ (i.e. well-established) although cardiac arrest rates in all these hospitals decreased significantly after introduction of RRS (data not shown).[24]

The current study has limitations due to its retrospective nature (RRT activation rates) and the utilisation of a questionnaire. With RRT activation data from only seven hospitals, it was not feasible to perform regression analysis to explore possible correlation between questionnaire results and activation rates. Secondly, although the questionnaire was anonymous and we assured respondents that their responses could not be traced back to a person or department, responder bias remains a potential confounder. Overall, the response rate was just under 31%, which is reasonable considering the large group of people who were potential subjects. It is worth considering that our main objective was to reach out to all healthcare providers who may have worked on the wards rather than looking at a single well-defined group, but this may have decreased our response rate. Thirdly, as the objective of the questionnaire was to measure compliance with the RRS protocol (which could be easily deducted from the questionnaire and case vignettes), reporting bias is possible as respondents could have been more inclined to provide the ‘desired’ answer. Adequate management for a deteriorating patient relies on synchronisation of a multitude of factors and the different healthcare providers working in unison. It is impossible to include all of the potential confounders as one would need to have detailed information on the adequacy of triggering systems,[25] adequacy of measurement of vital signs (continuous versus intermitted non-invasive measurements)[26] and adequacy of medical management initiated by the teams.[27] Hence, our focus was on the general process and not to specific factors relating to potential delays.

Conclusion

This study has shown successful implementation of RRS in the Netherlands. Self-reported compliance with the required steps in the escalation protocol is high, and the RRT activation rates as part of the two-tier system seem adequate. The RRT should not evolve into a rescue mechanism which compensates for system failures.[28] Future studies should focus on factors responsible for differences in activation rates, futility in escalation and assess if lower activation rates during nights and weekends are related to outcome and how to improve this situation.

Disclosures

All authors declare no conflicts of interest. No funding or financial support was received.

RRT activation rates
RRT activation rates were analysed for seven of the nine hospitals (figure 1). Two hospitals were unable to provide data on the monthly admissions and RRT calls. The average RRT activation rate per hospital over 2016 varied between 12 to 68 per 1000 admissions. All hospitals showed lower call rates after hours (18:00-08:00) and during 24-hour weekend days (8 to 33 per 1000 admissions) which was comparable with about twothirds of total RRT calls. Figure 1 (upper panel) shows the average number of RRT calls per hospital. Hospital 2 had the highest RRT call rates (surgical 64, medical 73 and after hours 60 and 59 per 1000 admissions, respectively) compared with the other hospitals. In the other six hospitals, RRT call rates were similar and ranged between 8 and 25/1000 admissions. After hours, this ranged between 7 and 16 per 1000 admissions. The lower panel in figure 1 indicates the highest and lowest average RRT call rate per specialty on a monthly basis. Irrespective of specialty, hospital 2 had the highest utilisation of RRT services. Minimal average use was 7.5/1000 admissions and highest 86/1000.

Discussion

After mandatory introduction of RRS in the Netherlands in 2008, mortality and patient harm significantly decreased.[5] The current study shows that the reported uptake of RRS is high and escalation of care is appropriate within the two-tier system. The main outcomes of the study point towards increased patient-tonurse ratios in the majority of studied hospitals with over 84% of nurses taking care of more than eight patients, and often up to 15 patients, during night shifts. The physical and mental strain of work is deemed relatively balanced by both nurses and registrars but increasing to higher workloads during shift (i.e. off-hours) work. Self-reported adherence to escalation of care is high despite the workload, although RRT activations reflect an almost 50% reduction of after-hours RRT activation rate (8-33/1000 versus 12-68/1000 admissions). Interestingly, the hospital with the highest nurse staffing also showed the highest rate of RRT activation. Unfortunately, within the current study, we were unable to correlate RRT activation rates with clinical outcome.

Statistically significant differences in RRT call rates were observed between hospitals (two hospitals were unable to provide the data). Hospital 2 showed consistently higher activation rates together with generally lower patient-tonurse ratios compared with the other hospitals. The number of RRT activations per 1000 admissions has been designated as ‘RRT dose’ and increasing ‘the dose’ is directly related to better outcomes. In this context, a dose of over 40 per 1000 admissions is deemed optimal.[12] The degree to which this can be extrapolated to the Dutch situation is not precisely known, because case-mix, trigger and activation protocols for RRTs are different and therefore care should be taken with these extrapolations. As an RRS is built from multiple components, focusing solely on the efferent arm is an atomistic approach. Before the escalation to the RRT, a multitude of processes occur. Therefore, the effectiveness of the entire system is influenced by many factors and each step is as important as the previous one. As previously shown, delayed recognition of deterioration and activation of the RRT is associated with poor clinical outcome.[13] Potential causes for system delays have been attributed to impaired recognition, failure to escalate, lack of clinical urgency and monitoring failures.[9,14,15] As also observed in our study, diurnal patterns have been observed for RRT activation rates. Activation rates out of office hours are lower compared with daytime and are associated with significant clinical burden.[16,17] Unfortunately, our study was not designed to analyse correlation of RRT activation rates with clinical outcome. Interestingly, higher patient-to-nurse ratios at night may be an important aspect which may be related to reduced recognition of deteriorating patients and thus declining activation rates at night. The landmark study by Aiken et al. in 2002 reported that in hospitals with high patient-to-nurse ratios, surgical patients had increased (risk-adjusted) 30-day mortality and failure-to-rescue rates. After adjustment, each additional patient per nurse increased the likelihood of death within 30 days of admission by 7% (OR 1.07, 95% CI 1.03-1.12).[10] A second study by the same group identified a correlation between the degree of education of nursing staff and clinical outcome in nine European countries.[18] Later published studies have shown similar outcomes, adding to the burden of evidence.[19] Through which specific interaction this occurs, remains unknown. One possibility, tightly linked to the failure-of-rescue principle and described in the above papers is the late detection of complications and clinical deterioration.[20,21] Therefore, increased patient-to-nurse ratios could impede appropriate and timely response and have previously been shown to be linked to monitoring failures.[15] After appropriate triggering, adequate escalation and management is required. As described previously, ward teams have an opportunity to manage their patients after which RRT can be activated. Inefficiency in management and inappropriate escalation has been referred as ‘clinically futile cycles’.[22] This concept refers to periods in which patient adverse events occur not because of ignorance or inability, but solely because healthcare providers apply unwritten rules and cultural practices delaying effective and efficient care. Examples of these were shown in this study (table 4) in which the RRT is not always called, especially by registrars and consultants who do not appropriately escalate in the first place. This circular way of thinking and acting prevents staff from effectively managing the clinical problem, with a potential delay or absence of RRT activation. Reasons for this futility remain not fully elucidated and in accordance with Shearer et al., a small subset of healthcare providers still revert to either a colleague or consultation of another ‘organ’ specialist.[23] Overall, and within the context of this study, there has never been a reliable comparison of the ‘dose’ of the RRT between various hospitals, countries or hospital cultures. As seen in the current study, only one of the hospitals (hospital 2) is subsequently deemed ‘mature’ (i.e. well-established) although cardiac arrest rates in all these hospitals decreased significantly after introduction of RRS (data not shown).[24] The current study has limitations due to its retrospective nature (RRT activation rates) and the utilisation of a questionnaire. With RRT activation data from only seven hospitals, it was not feasible to perform regression analysis to explore possible correlation between questionnaire results and activation rates. Secondly, although the questionnaire was anonymous and we assured respondents that their responses could not be traced back to a person or department, responder bias remains a potential confounder. Overall, the response rate was just under 31%, which is reasonable considering the large group of people who were potential subjects. It is worth considering that our main objective was to reach out to all healthcare providers who may have worked on the wards rather than looking at a single well-defined group, but this may have decreased our response rate. Thirdly, as the objective of the questionnaire was to measure compliance with the RRS protocol (which could be easily deducted from the questionnaire and case vignettes), reporting bias is possible as respondents could have been more inclined to provide the ‘desired’ answer. Adequate management for a deteriorating patient relies on synchronisation of a multitude of factors and the different healthcare providers working in unison. It is impossible to include all of the potential confounders as one would need to have detailed information on the adequacy of triggering systems,[25] adequacy of measurement of vital signs (continuous versus intermitted non-invasive measurements)[26] and adequacy of medical management initiated by the teams.[27] Hence, our focus was on the general process and not to specific factors relating to potential delays.

Conclusion

This study has shown successful implementation of RRS in the Netherlands. Self-reported compliance with the required steps in the escalation protocol is high, and the RRT activation rates as part of the two-tier system seem adequate. The RRT should not evolve into a rescue mechanism which compensates for system failures.[28] Future studies should focus on factors responsible for differences in activation rates, futility in escalation and assess if lower activation rates during nights and weekends are related to outcome and how to improve this situation.

Disclosures

All authors declare no conflicts of interest. No funding or financial support was received.