Tiered escalation response systems in practice: A post hoc analysis examining the workload implications

Abstract

Objective

Many rapid response systems now have multiple tiers of escalation in addition to the traditional single tier of a medical emergency team. Given that the benefit to patient outcomes of this change is unclear, we sought to investigate the workload implications of a multitiered system, including the impact of trigger modification.

Design

The study design incorporated a post hoc analysis using a matched case–control dataset.

Setting

The study setting was an acute, adult tertiary referral hospital.

Participants

Cases that had an adverse event (cardiac arrest or unanticipated intensive care unit admission) or a rapid response team (RRT) call participated in the study. Controls were matched by age, gender, ward and time of year, and no adverse event or RRT call. Participants were admitted between May 2014 and April 2015.

Main outcome measures

The main outcome measure were the number of reviews, triggers, and modifications across three tiers of escalation; a nurse review, a multidisciplinary review (MDT—admitting medical team review), and an RRT call.

Results

There were 321 cases and 321 controls. Overall, there were 1948 nurse triggers, of which 1431 (73.5%) were in cases and 517 (26.5%) in controls, 798 MDT triggers (660 [82.7%] in cases and 138 [17.3%] in controls), and 379 RRT triggers (351 [92.6%] in cases and 28 [7.4%] in controls). Per patient per 24 h, there were 3.03 nurse, 1.24 MDT, and 0.59 RRT triggers. Accounting for modifications, this reduced to 2.17, 0.88, and 0.42, respectively. The proportion of triggers that were modified, so as not to trigger a review, was similar across all the tiers, being 28.6% of nurse, 29.6% of MDT, and 28.2% of RRT triggers. Per patient per 24 h, there were 0.61 nurse reviews, 0.52 MDT reviews, and 0.08 RRT reviews.

Conclusions

Lower-tier triggers were more prevalent, and modifications were common. Modifications significantly mitigated the escalation workload across all tiers of a multitiered system.

1. Introduction

Track and trigger and response systems are the tools used to detect and guide the response to acute clinical deterioration in hospital patients. They highlight to clinicians when a trigger indicative of clinical deterioration has been achieved and then guide clinicians to the type of response required. Tracking refers to vital sign measurement and documentation. A trigger occurs when a vital sign meets a certain predetermined threshold—either a single vital sign parameter or an aggregate score can be used as a trigger. Many of these systems now incorporate a tiered escalation response based upon the degree of physiological disturbance.¹ The ‘lower’ tiers of response can range from an increased frequency of observation, a review by a senior nurse, or a home team/cover doctor through to the ‘highest’ tier, being a rapid response team (RRT). Track and trigger systems form the basis of the afferent limb of the rapid response system, while the efferent arm is the response that occurs.

Trigger thresholds at the lower tiers are more prevalent² and are likely to be more sensitive but less specific for detecting patients at a risk of further deterioration.³ In the lead up to this study, we noted in a local audit (O'Connell, unpublished data) the lower-tier triggers had a sensitivity of 74% and specificity of 49% for predicting a subsequent RRT call, unexpected intensive care unit (ICU) admission, or a cardiac arrest. In contrast, triggers for the highest tier of escalation had a sensitivity of 35% and a specificity of 97%. Triggers set at lower thresholds are thus more likely to initiate an escalation response and be associated to increase in staff workload.

Desensitisation to triggers is a risk if there is a large burden of low-specificity triggers occurring that distracts the clinician from the true deterioration. This, in turn, can lead to a delayed or absent response to triggers, known as alarm fatigue.[4], [5], [6] Alarm fatigue has been well studied in critical care areas and areas where electronic alarms from physiological monitoring devices are used. Frequent alarms that prove to be false positive lead to staff disregarding subsequent alarms. This results in a delay in recognition of deterioration and in turn leads to patient harm. Alarm fatigue can be one of the causes of afferent limb failure (ALF), where an (unmodified) trigger occurs but is not responded to; this has traditionally been used to describe failure to respond to an RRT trigger, but with the introduction of lower tiers, this could be applied to these new tiers also.⁴ Multiple triggers can also interrupt the flow of ‘usual’ work. An interruption to a task can increase errors, again leading to patient harm.^7,8 Thus, we may expect the same in the ward environment, where manually collected vital sign measurements are gathered on a regular basis.

Triggers however can be modified at the discretion of the treating clinician, so as not to generate the prescribed escalation response and/or alter the threshold at which an escalation response occurs. These modifications can be used as a strategy to mitigate against alarm fatigue by reducing potential escalation workload. This process allows a formal structure to adapt the afferent limb and does not represent ALF.

There are mixed results in studies examining the impact of tiered escalation response systems on outcomes and even less information on the workload generated.[9], [10], [11], [12] Given the unclear benefits of multitiered systems over a single-tiered system, we sought to investigate the actual and potential workload implications of a multitiered system, including the impact of trigger modification upon escalation workload.

2. Methods

2.1. Study design

This is a nested, matched case–control study. A post hoc analysis was performed on data collected for a prior study examining the predictive value of tiered escalation system triggers. The medical records, including the observation charts, were reviewed for triggers, escalation responses, and modifications to triggers. A modification is a written medical order so that a particular trigger(s) does not initiate the required escalation response (can be ordered by any medical officer above the level of an intern).

2.2. Setting

The setting incorporated a public, acute, adult quaternary referral hospital with 700 inpatient overnight beds.

2.3. Population

Cases were patients who had an adverse event (cardiac arrest or unanticipated ICU admission from the ward) or an RRT call. If a patient had more than one event, the last event was considered. Controls were patients matched one to one by age, gender, ward, and time of year who did not have an adverse event or an RRT call. Patients were identified from hospital (discharge data from patient administration database) and RRT (locally held ICU and RRT data collected by hand at RRT calls and entered into an internally managed database) databases, and their medical records were reviewed. The controls were found by manual review of hospital admission data. Patients were admitted during the period from May 2014 to April 2015.

For the cases, the 24-hour period up to 15 min before an event was reviewed. For the controls, a random 24 h period of their admission was selected, during which they remained in the general ward.

2.4. Nature of data collected

The patient observations documented include respiratory rate, oxygen flow rate, oxygen saturation, systolic and diastolic blood pressure, pulse rate, temperature (measured in degrees Celsius), sedation score, and pain score. The sedation scoring ranged from 0, meaning awake; 1, easily roused and stays awake; 2, being easily roused but cannot maintain alertness for more than 10 s; and 3, being unrousable. Pain score is a patient subjective score, with 0 being no pain and 10 being the worst pain imaginable. Any single vital sign, once a predetermined threshold was met, could trigger an escalation response. All vital signs documented in the 24 h period were reviewed, and all triggers in that period were recorded.

There were three tiers of escalation, the lowest being a nurse review. This mandates a senior registered nurse to review the patient within 30 min and document the outcome for the review. The next tier, a multidisciplinary review (MDT), requires a review and documentation by a medical officer within 30 min, usually attended by the admitting medical team or if after hours by a cover medical officer. The final tier is an RRT call. All reviews that were documented during the 24 h period were recorded.

2.5. Statistical analysis

Data were entered into an Access database. Data were extracted, and simple descriptive statistics were applied. Analysis was performed in Microsoft Excel.

The study received ethics approval, HREC/14/RAH/163.

3. Results

There were 321 cases and 321 controls, the median age was 77 (interquartile range: 63–85) y in the cases and 76 (interquartile range: 63–85) y in the control group (p = 0.63). Overall, 50.2% of cases and 49.8% of controls were male (p = 0.93) and were similarly matched forward. Of the cases, 8 (2%) had a cardiac arrest, 17 had (5%) an unanticipated ICU admission, and 296 (92%) had an RRT call. During their hospital admission, 95 (29.6%) cases and 9 (2.8%) controls died. A not-for-cardiopulmonary resuscitation order was documented for 103 (32.1%) cases and 49 (15.3%) controls.

3.1. Triggers

Across all patients, there were 1948 nurse triggers, of which 1431 (73.5%) were in cases and 517 (26.5%) in controls, 798 MDT triggers (660 [82.7%] in cases and 138 [17.3%] in controls), and 379 RRT triggers (351 [92.6%] in cases and 28 [7.4%] in controls). On average, per patient per 24 h, there were 3 nurse, 1.24 MDT, and 0.59 RRT triggers.

Amongst both groups, 238 (74%) cases and 163 (50%) controls (62% overall) had at least one nurse trigger, 164 (51%) cases and 56 (18%) controls (34% overall) had an MDT trigger at least once, and 113 (35%) cases and 15 (5%) controls (20% overall) had at least one RRT trigger. For cases, triggers that occurred in the 24 h period prior to their event were included (excluding the triggers that resulted in their ‘event’).

3.2. Trigger modification

The proportion of triggers that were modified was similar across all the tiers of escalation, being 28.6% of nurse, 29.6% of MDT, and 28.2% of RRT triggers. However, there was a difference in the proportion of triggers modified for the cases compared to that of the controls in the two higher tiers (Table 1).

Table 1.

Number of triggers documented in cases and controls in each tier and the percentage modified.

Trigger type	RRT	MDT	Nurse
Number of cases (% modified)	351 (26.5%∗)	660 (28.2%)	1431 (28.2%)
Number of controls (% modified)	28 (50.0%)	138 (36.2%)	517 (29.6%)
Total (% modified)	379 (28.2%)	798 (29.6%)	1948 (28.6%)

This table presents, for each of the three tiers of response to deterioration, the number of triggers documented for cases and controls together with the proportion of these triggers that were modified.

∗p < 0.05 for the percentage of triggers in cases that were modified compared to controls for that tier.

RRT, rapid response team; MDT, multidisciplinary team.

The number of each trigger and the percentage that was modified both in cases and controls are shown in Table 2.

Table 2.

Workload based upon reviews and triggers and following exclusion of triggers modified so as not to initiate an escalation response.

	Cases	Controls	Total
Median number of nurse triggers per patient per 24 h (range)	2 (0–40)	1 (0–26)	1 (0–40)
Median number of MDT triggers per patient per 24 h (range)	1 (0–22)	0 (0–15)	1 (0–22)
Median number of RRT triggers per patient per 24 h (range)	0 (0–20)	0 (0–9)	0 (0–20)
Nonmodified triggers (modified triggers excluded)
Median number of nonmodified nurse triggers per patient per 24 h (range)	2 (0–22)	0 (0–13)	1 (0–22)
Average number of nonmodified MDT triggers per patient per 24 h	1 (0–18)	0 (0–5)	1 (0–18)
Average number of nonmodified RRT triggers per patient per 24 h	0 (0–11)	0 (0–5)	0 (0–11)

RRT, rapid response team; MDT, multidisciplinary team.

The number of reviews across the cohorts as well as the average workload per 24 h period is presented in Table 3.

Table 3.

Number of reviews that occurred and average number of reviews per patient per 24 h.

	Cases	Controls	Total
Number of nurse reviews	267	127	394
Number of MDT reviews	265	66	331
Number of RRT reviews	50	0	50
Average number of escalation response based upon a nurse trigger per patient per 24 h	0.83	0.40	0.61
Average number of escalation response based upon MDT triggers per patient per 24 h	0.83	0.21	0.52
Average number of escalation response based upon RRT triggers per patient per 24 h	0.16	0	0.08

RRT, rapid response team; MDT, multidisciplinary team.

Modifications to triggers so as to not trigger an escalation response reduced workload; however, the magnitude of the gap between modifications and expected escalation (i.e., ALF) was greater, across all tiers.

Fig. 1 demonstrates the proportion between all triggers, the triggers that were modified and actual escalations that occurred. In Fig. 2, the relationship of frequency of each trigger is compared to the percentage of modifications of each trigger.

Fig. 1.

Total triggers, the proportion of total triggers that are not modified (unmodified triggers/all triggers), and triggers that resulted in a review. RRT, rapid response team; MDT, multidisciplinary team.

Fig. 2.

Relationship between the number of each type of triggers vs the percentage of triggers modified.

4. Discussion

4.1. Main study findings

This study has reported on the prevalence of triggers for acute deterioration, episodes of escalation, and potential (all triggers that could result in an escalation response) and actual workload (triggers that did result in an escalation response) within a system of a multitiered escalation response amongst hospital inpatients. This study showed that lower-tier triggers were most prevalent, and the use of modifications, so that a particular trigger did not initiate an escalation response, was common, particularly amongst highly prevalent triggers. As a consequence, modifications significantly mitigated the expected occurrence of escalation of care, and in turn, staff workload. Furthermore, even after accounting for modifications, ALF further reduced escalation workload.

4.2. Workload and outcomes, comparison with other evidence

Introduction of tiered escalation and response systems increases the workload for hospital staff. In one study, the effect of introducing a two-tier system across four wards resulted in a significant increased rate of RRT calls (2.2% of all patients to 3.7%) and the frequency of vital sign monitoring (3.4 full sets per d to 4.5 per d).¹⁰ In another, the implementation of a two-tier system was associated with a significant increase in the RRT rate from 18/1000 separations to 43/1000 separations (136% increase over 5 y) and a subsequent decrease in the rate of cardiac arrests by 42%.⁹ Previously, we showed an 82% increase in the RRT call rate (28 calls per 1000 admissions per month to 50 RRT calls per 1000 admissions [an 82% increase]) after the implementation of a three-tiered system (the same time period as this study), but no associated change was observed in the trend of the rate of cardiac arrests or hospital mortality when using an interrupted time series analysis.¹² Thus, it appears that tiered escalation systems increase staff workload but may not consistently lead to improved patient outcomes.

4.3. Effect of modifications

Modifying a trigger so that it does not’ result in an escalated response has the potential to reduce overall workload as well as reduce alarm fatigue. Modifications rely on clinicians to make a judgement on the need for escalation responses in the setting of certain triggers. However, given that there may be more than one trigger at one time point, this is not a direct relationship. With modifications, the potential workload decreased across all escalation tiers, with a reduction in the average number of triggers that should result in an escalation response per 24 h per patient (average RRT triggers per 24 h reduced from 0.59 to 0.42, MDT triggers from 1.24 to 0.88, and nurse triggers from 3.03 to 2.17). However, when contrasting expected to actual escalation responses, ALF had an additional, and in magnitude, greater reduction in the expected workload. Modifications may also have a variable impact on the workload, with one study showing that they did not reduce repeat RRT calls.¹³ This study sought to evaluate the effect on workload and is not designed to assess the effect of modifications on patient outcomes.

The proportion of triggers that were modified was similar across all the tiers of escalation. However, there was a higher percentage of RRT triggers modified for controls than for cases (50% vs 26%) than the difference in modification to MDT and nurse triggers, which were similar for both groups. This suggests that, at the higher tiers, clinicians were more likely to make a modification for patients inherently less likely to go on and deteriorate further. There also seems to be an influence on the use of modifications based on the prevalence of the trigger occurring. Frequently occurring triggers were very likely to be modified, and triggers with low frequency had a variable rate of modification. For example, the most commonly occurring trigger of oxygen saturations being 90–94% was modified 45% of the time, and the second most common trigger of pulse rate being 100–120 beats per min was modified 37% of the time. At lower frequency of triggers though, this relationship was not seen. For example, the least frequent trigger of respiratory rate of less than eight breaths per min was still modified in 40% of trigger occurrences. A respiratory rate of 26–30 breaths per minute was the most modified trigger (49% modified). This suggests that clinicians are selective with modifications and how they may be applied to the workload. In contrast to ALF, which has been associated with adverse patient outcomes, the^14,15 consequence of modifications is not as well documented and clearly needs further evaluation.

In summary, modifications reduce the potential workload by reducing the number of triggers that require a response, but it is unclear the effect this has on patient outcomes. This study has shown that modifications are often applied to high-frequency triggers and are more often applied to RRT triggers for patients that are less likely to deteriorate.

4.4. Limitations

This is a single-centre, retrospective study that may limit it’s generalisability. Limitations of this study include a sampling issue. While the cases and controls are well matched, the sample may not be representative of all patients in the hospital. A well-matched control group that did not go on to have an event likely represents a group of ‘well’ inpatients that had a lower mortality and lower rate of limitations to therapy. Furthermore, the controls had a random 2- h period selected for review; most deterioration occurs within the first 24 h of admission, and so it may underestimate the triggers in the control group. However, the control group did not have any RRT calls and still had a large number of triggers, suggesting there is a significant workload associated with a tiered system across all patients. We cannot quantify the gap between triggers and number of reviews, i.e., ALF, in this study because one patient may have had multiple triggers at one time point. Further evidence is needed to investigate the degree of ALF in a tiered escalation response system. ALF may in part be a way by which the ward staff managers manage their workload.

5. Conclusion

In conclusion, modifications reduce the potential workload imposed upon hospital staff by a large amount. This effect was seen across all the three tiers of a multitiered system and across a breadth of patient groups.

Conflict of interest

Dr. O'Connell has nothing to disclose.

CRediT author statement

Alice O’Connell: Conceptualization, methodology, investigation, data curation, writing -original draft preparation and editing and visualisation. Arthas Flabouris: Conceptualization, methodology, writing – review and editing, supervision Suzanne Edwards: Software, formal analysis, data curation, writing – review and editing. Campbell Thompson: Conceptualization, methodology, writing – review and editing, supervision

Appendix 1. Individual triggers, their frequency, and percentage modified.

	Number of triggers: total	Number of triggers: cases	Number of triggers: controls	% of triggers modified : total	% of triggers modified: cases	% of triggers modified : controls
SpO2 90–94	518	373	145	44.9	45.5	43.4
Pulse rate 100–120 beats per min	460	375	85	37.2	34.6	48.2
Respiratory rate 21–25 breaths per min	352	272	80	28.4	27.6	31.2
Systolic blood pressure 90–100 mmHg	226	159	67	33.6	27	49.2
Pain score 5–7	162	89	73	1.2	0	2.7
Respiratory rate 26–30 breaths per min	133	121	12	48.9	46.3	75
Pulse rate 120–140 beats per min	132	130	2	29.6	29.2	50
Systolic blood pressure <90 mmHg	120	111	9	28.3	26.1	55.5
Pulse rate 50–60 beats per min	101	37	64	22.8	24.3	21.8
SpO2 < 90	98	89	9	34.7	32.6	55.5
Oxygen flow = 6 L per min	98	92	6	4.1	4.4	0
Oxygen flow >6L per min	95	88	7	21.1	21.6	14.2
Systolic blood pressure 170–180 mmHg	92	60	32	15.2	15	15.6
Respiratory rate >30 breaths per min	79	79	0	38	38	0
Systolic blood pressure 180–200 mmHg	78	59	19	12.8	13.5	10.5
Temperature 38.1–38.5	72	53	19	11.1	9.4	15.7
Temperature >38.5	59	47	12	28.8	27.6	33.3
Sedation score = 2	58	52	6	1.7	1.9	0
Pain score 8–10	45	29	16	0	0	0
Pulse rate >140 beats per min	37	37	0	8.1	8.1	0
Temperature 35.1–35.5	35	28	7	2.9	3.5	0
Systolic blood pressure >200 mmHg	30	29	1	6.7	6.9	0
Temperature <35	21	21	0	42.8	42.8	0
Respiratory rate <8 breaths per min	10	1	9	40	0	44.4
Respiratory 8–10 breaths per min	6	4	2	0	0	0
Pulse rate 40–50 beats per min	3	2	1	0	0	0
Sedation score = 3	3	3	0	0	0	0
Pulse rate <40 beats per min	0	0	0	0	0	0
Bleeding	1	1	0	0	0	0
Seizure	1	1	0	0	0	0

Listed in order of descending frequency of trigger over both groups. Colour indicates tier of the individual trigger (yellow being a nurse, orange being an MDT, and red being an RRT trigger).

RRT, rapid response team; MDT, multidisciplinary team; SpO₂, oxygen saturation.