Integrated Pulmonary Index during procedural sedation and analgesia

A cluster-randomized trial

Aaron Conway

Peter Munk Cardiac Centre, UHN & University of Toronto

Mohammad Goudarzi Rad

Peter Munk Cardiac Centre, UHN & University of Toronto

Sebastian Mafeld

Joint Department of Medical Imaging, UHN & University of Toronto

Matteo Parotto

Department of Anesthesiology and Pain Medicine, UHN & Interdepartmental Division of Critical Care Medicine, University of Toronto

July 20, 2023

Procedural sedation is very safe and highly effective

Using capnography for respiratory monitoring during sedation can provide an extra layer of safety

The Canadian Anesthesiologists’ Society’s position statement recommends that capnography should be available wherever moderate or deep sedation is used.

CAS guidelines do not provide specific recommendations for how capnography should be implemented for nurse-administered sedation

Alarm management

Capnography is a very sensitive device so we need to ensure the alarm conditions set will not disrupt procedures unnecessarily

Every alarm that gets triggered from a capnography monitor during a procedure should be important enough to trigger an intervention

The intervention should be initiated as soon as possible after the alarm is triggered

Optimizing alarm conditions for capnography

Integrated pulmonary index

Index of physiological parameters related to the assessment of respiratory function (etCO₂, SpO₂, RR, HR) based on a fuzzy-logic inference model

Simplifies the interpretation of respiratory monitoring by assigning clinical responses for given scores

Shows promise as a strategy to optimize the implementation of capnography for respiratory monitoring during nurse-administered sedation

Aim is to determine if using the Integrated Pulmonary Index is an effective way to implement capnography during nurse-administered sedation.

The study

Hybrid implementation-effectiveness cluster randomized trial design

• evidence for the safety and efficacy of using capnography exists
• implementation of this monitoring device for sedation is recommended in clinical guidelines

Nurses who choose to participate were randomized to:

Capnography with IPI

Capnography without IPI

Hybrid implementation-effectiveness cluster randomized trial design

Establishment

2 months

Stability

2 months

Evaluation

8 months

Establishment period

• Start using the Capnostream 35p devices
• Education to nursing staff
• Invited nurses to take part in the randomized controlled trial

Stability period

Goal was to optimize the alarm conditions

Audit and feedback process to evaluate the performance characteristics of the capnography alarms:

• Alarm burden (total number of alarms triggered).
• Number of appropriate alarms (defined as an alarm that triggered an intervention).
• Number of inappropriate alarms (defined as alarms that were triggered but manually dismissed by silencing the alarm).
• Duration of alert conditions (defined as the total time that an alert condition was active).

• Performance characteristics of alerts should be evaluated after implementation and that clinicians should revise alerts based on the performance characteristics.

The Society for Critical Care Medicine Alarm and Alert Fatigue Task Force

Audit and feedback processes are commonly used and can improve practice.

Evaluation period

Alarm performance was compared between nurses randomized to use IPI with those who were not.

Primary outcome

Number of seconds in an alert condition state without an intervention being applied

• Scored as zero if a patient did not have any alarms during the procedure
• Used a mixed effects model that accounted for the large number of zero scores, being the zero-inflated negative binomial mixed effects model
• The two parts of the zero-inflated model we used were a logit model to model which of the two processes the zero outcome was associated with, and a negative binomial model for the count process
• Fixed effect in the model was the randomization and the random effect was the cluster

Secondary outcomes

• Alarm burden (total number of Medtronic Capnostream 35p monitor alarms).
• Number of appropriate alarms (defined as an alarm that triggered an intervention).
• Number of inappropriate alarms (defined as alarms that were triggered but manually dismissed by silencing the alarm).
• Total duration of alert conditions (defined as the total time that an alert condition was active inclusive of the time to an intervention and the time until the alert conditions are resolved).
• TROOPS adverse sedation events.
• Area under the curve of oxygen desaturation, calculated by taking the difference between a threshold (SpO2 90%) and current oxygen saturation summed each second while oxygen saturation was below the threshold.

Results

CONSORT diagram - Data collection for this study took place from June 2022 to June 2023

Sample

Characteristic	IPI disabled, N = 1751	IPI enabled, N = 2151
Age (years)	63 (50, 72)	65 (55, 71)
Sex
Female	81 (47%)	82 (39%)
Male	93 (53%)	129 (61%)
Total midazolam (mg)	1.00 (1.00, 1.50)	1.00 (1.00, 2.00)
Total fentanyl (mcg)	50 (50, 100)	75 (50, 100)
Oxygen device
mask	0 (0%)	2 (0.9%)
nasal prongs	173 (100%)	210 (99%)
Procedure duration (minutes)	31 (20, 51)	31 (17, 53)
1 Median (IQR); n (%)

Alarm settings used by nurses during the study

Parameter	Lower threshold	Upper threshold
Pulse rate	40 beats per minute	120 beats per minute
Respiratory rate	6 breaths per minute	50 breaths per minute
End-tidal carbon dioxide	8 mmHg	60 mmHg
Oxygen saturation	90%
No breath detected	30 seconds
Integrated pulmonary index	4

Primary outcome

No difference between groups (IRR 1.79; 95% CI 0.93 to 3.45)

Alarm-related secondary outcomes

	Negative binomial component			Zero component
Outcome	IRR	95% CI	p-value	OR	95% CI	p-value	ICC
Total alarm duration	1.77	0.96 to 3.27	0.068	0.94	0.56 to 1.59	0.83	0.02
Number of appropriate alarms	0.63	0.17 to 2.37	0.495	0.37	0.1 to 1.4	0.14	<0.01
Number of inappropriate alarms	13.13	1.4 to 122.95	0.024	38.04	0.07 to 20945.46	0.26	<0.01
OR = Odds ratio; IRR = Incidence rate ratio; ICC = Intra-cluster correlation

Area under the SpO₂ 90% curve

Lower for the IPI-enabled group (𝛽 -0.82; 95% CI -1.41 to -0.22)

Adverse events

• There was one intermediate severity respiratory adverse event observed in the IPI disabled group.
• There were two intermediate-severity adverse events related to sedation quality in the IPI enabled group.
• The remaining 13 adverse events were minor severity related to breathing where oxygen desaturation was addressed with minor interventions.
• There were no serious adverse events observed in either group.
• The odds ratio for the occurrence of an adverse event in the IPI enabled group compared to the IPI disabled group was 1.43 (95% CI 0.51 to 4.01).

Discussion

We found that the use of the Integrated Patient Index did not reduce the number of seconds that alarms were triggered without intervention.

Therefore, results from the study do not indicate that integrating multiple physiological parameters related to respiratory assessment into a single index help to lower the threshold for intervention by nurses.

Although we didn’t specify thresholds in the protocol, nurses did not change alarm settings during the evaluation period.

In effect, this was a comparison of IPI at a threshold of 4 versus no IPI. It is possible that results would differ at a lower IPI threshold.

Although oxygen desaturation was not common in either group, the IPI-enabled group had lower levels of the area under the SpO₂ desaturation curve

It is possible that the actions taken in response to alarms in the IPI-enabled group mediated the effect on oxygen saturation, which potentially signals a safety benefit for using this approach for capnography alarm management

The total doses of sedative and analgesic medication for many patients was small (median 1mg midazolam and 50-75mcg fentanyl)

May not be relevant to situations where higher doses of sedative and analgesic medication are used (higher number of alarms)

The cluster design (by nurses) was a very efficient way to conduct this study

Only possible due to the hybrid implementation-effectiveness design

Further exploratory analyses are possible with the data collected in this study

• More detailed temporal exploration of types of interventions and alarm conditions
• Exploring relationships between alarm conditions and adverse events
• Dataset can be used for external validation of capnography waveform classification algorithms

Limitations

• Imprecision may be a concern, arising from the fact that:
  • a smaller number of clusters were included than we estimated for the sample size calculation due to the restricted number of sites available at the time of data collection.
  • The intra-cluster correlation for the primary outcome was also higher (0.03) than we estimated for the sample size calculation.
• The research assistant performing outcome measurement was not blinded to the allocation of the nurse because it was not practical to obscure the Capnostream monitor from their view without also impeding the nurses own use of the device.

Conclusion

Enabling the Integrated Pulmonary Index did not lower the time taken by nurses to intervene after an alarm was triggered by the Capnostream monitor

Potential of improved safety with lower severity of oxygen desaturation in the IPI-enabled group (unclear if the difference is clinically significant)

Questions