Implementation of Data Drive Heart Rate and Respiratory Rate parameters on a Pediatric Acute Care Unit.

The majority of hospital physiologic monitor alarms are not clinically actionable and contribute to alarm fatigue. In 2014, The Joint Commission declared alarm safety as a National Patient Safety Goal and urged prompt action by hospitals to mitigate the issue [1]. It has been demonstrated that vital signs in hospitalized children are quite different from currently accepted reference ranges [2]. Implementation of data-driven, age stratified vital sign parameters (Table 1) for alarms in this patient population could reduce alarm frequency.

Methods

This is a prospective study of alarm frequency after bedside monitor implementation of age stratified data-driven heart rate (HR) and respiratory rate (RR) parameters on a cardiac medical/surgical step-down unit at a quarternary-care children’s hospital. The data-driven parameters were derived using nurse-documented HR and RR vital signs (62,508 unique measurements) of all non-ICU hospitalized patients <18 yrs during calendar year 2013 at Lucile Packard Children’s Hospital Stanford (LPCH). Alarm frequency data was analyzed over 28 day periods before and after the intervention on 10/27/14. HR and RR alarm data were obtained from the LPCH clinical data warehouse, which stores real-time central monitor system data.

Results

In the pre-intervention study period, there were a total 7,028 HR alarms and 8,400 RR alarms, yielding an average of 14.1 HR alarms/patient day and 16.9 RR alarms/patient day. In the post-intervention period, total HR alarms decreased to 4,844 (9.6/patient day), but RR alarms increased to 20,860 (41.3/patient day).

Discussion

While the frequency of HR alarms decreased following implementation of age-stratified, data-driven vital sign parameters, the frequency of RR alarms markedly increased, particularly for low RR values (Figure 1). Factors contributing to this discrepancy might include: (1) inaccurate creation of data-driven parameters because nurse-charted values were used to calculate them rather than raw monitor data, or because representative patients who spent time in the ICU were excluded from the analysis; and (2) it is possible that pre-intervention monitor parameter settings did not represent pre-intervention reference ranges and so when default monitor settings were updated as part of the intervention, the alarm rate increased as a result of this discrepancy.

Figure 1

Pre- and post-intervention HR & RR alarms.

Conclusion

Our results suggest that adoption of data-driven values for HR alarm parameters will decrease the frequency of HR alarms in a children’s hospital. However, further work to understand the increase in RR alarms is necessary, highlighting the challenges in creation and implementation of data-driven vital sign parameters to reduce alarm fatigue.

Table 1

Data-driven HR and RR limits, by age

Age	Low HR	High HR	Low RR	High RR
<1 month	<115	>170	<30	>60
1- <6mo	<105	>170	<20	>55
6mo- <1yr	<100	>165	<20	>45
1- <2yr	<90	>165	<20	>45
2- <3yr	<85	>155	<18	>40
3- <5yr	<75	>155	<16	>35
5-<9yr	<70	>140	<16	>30
9- <12yr	<65	>130	<14	>30
12- <15yr	<60	>125	<14	>30
≥15yr	<60	>115	<13	>25