Prospective randomised trial of the Integrated Pulmonary Index™ in low-acuity inpatients

Editor

Opioid-induced respiratory impairment is a significant patient safety concern and increases healthcare costs.¹ Whilst oxygen saturation (SpO₂) has been the mainstay of respiratory monitoring since the early 1980s,² end-tidal capnography (EtCO₂) may be superior to pulse oximetry in detecting respiratory compromise in patients receiving supplemental oxygen, even in low-acuity areas.^3,4 However, EtCO₂ monitoring has been largely limited to the operating room, procedural areas, and the ICU, partially out of concern that additional surveillance monitoring of inpatients may present interpretation difficulties for providers in low-acuity areas, whilst increasing alarm fatigue.⁴ To address these concerns, the Integrated Pulmonary Index (IPI™, Medtronic, Dublin, Ireland) integrates four parameters—EtCO₂, ventilatory frequency (VF), SpO₂, and pulse rate (PR)—into one unitless score.⁵ A single monitor that incorporates and weighs the contribution of multiple real-time vital signs might allow the more robust monitoring of patients, whilst addressing concerns about increased workload and alarm fatigue. We hypothesised that IPI-based monitoring is non-inferior to the standard vital-sign monitoring. A pre–post study was conducted on four general care floors at the Vanderbilt University Medical Center (Nashville, TN, USA). A waiver of the requirement for informed consent was granted by the Human Research Protection Program, as the risk of changes in surveillance monitoring and alerting in a population not traditionally monitored was determined to be minimal (Institutional Review Board #161809). This trial was published on clinicaltrials.gov before initiation (NCT03050983). A copy of the full trial protocol is available upon request.

The nursing staff received extensive pre-implementation training on the device. Spontaneously breathing adult patients, with an expected minimum duration of continuous capnography and pulse oximetry monitoring for the initial 24 h of admission, were eligible for enrolment. In Phase 1, each subject was monitored with continuous SpO₂, PR, VF, and EtCO₂; these monitors were also used to generate alerts. In Phase 2, the monitoring remained unchanged, but the IPI algorithm was enabled and generated the alerts, in lieu of the standard monitors. Of note, alerts were not generated until an alert condition had been met for at least 30 s. After 24 h, the decision to remove capnography and pulse oximetry was left to the clinician. The primary outcome was hospital resource length of stay. Patient characteristics were compared across study phases using a Wilcoxon test or Pearson test, as appropriate. Linear regression was utilised to quantify the effect of the IPI intervention on the primary outcome, and the cost and escalation of care, adjusting for potential confounders. The use of IPI during Phase 2 was considered non-inferior to the standard of care during Phase 1 if the mean hospital resource length of stay was not larger by more than 20%. In a bootstrap-based post hoc analysis, we found that allocating 200 patients in each arm of the study provided approximately 85% power to conclude non-inferiority of the IPI intervention with respect to the primary outcome.

A total of 217 patients were included in the analysis from each phase. Patients in each group were similar (Table 1). On multivariate analysis, there was no significant difference in Phase 2 compared with Phase 1 in the following outcome measures: mean hospital resource length of stay: –2.1% [95% confidence interval (CI): –14.1, 9.0; P=0.713], total hospital cost: 3.6% (95% CI: –5.1, 13.2; P=0.428), and variable direct technical cost: 5.7% (95% CI: –4.2, 16.7; P=0.269). Rapid response team (RRT) calls and ICU escalations were uncommon and did not significantly differ between groups [9% vs 11% (P=0.422) for RRT and 3% vs 3% (P=1.000) for ICU escalation].

Table 1.

Cohort characteristics and outcomes, n=434. Data are presented as median (lower quartile–upper quartile) for continuous variables. Numbers after percentages are counts. ASA status, ASA physical status classification; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; LOS, length of stay; OSA, obstructive sleep apnoea; RRT, rapid response team; VDTC, variable direct technical cost. Tests used: *Wilcoxon test; ^†Pearson test; ^‡Adjusted regression

Variable	Phase 1 (n=217)	Phase 2 (n=217)	P-value
Age (yr)	47/61/71	51/61/69	0.841*
Sex: female	48% (105)	31% (68)	<0.001†
Case-mix index	1.66 (1.35–2.07)	1.63 (1.35–2.07)	0.313*
BMI	28.4 (25.1–32.9)	29.0 (25.4–33.2)	0.461*
CAD	11% (24)	20% (43)	0.012†
CHF	6% (14)	4% (9)	0.284†
DM	21% (46)	19% (42)	0.633†
OSA	16% (34)	16% (34)	>0.99†
COPD	5% (11)	9% (19)	0.130†
Surgical admit	92% (200)	94% (204)	0.449†
ASA physical status
1	2% (5)	1% (3)	0.168†
2	28% (60)	23% (49)
3	60% (131)	70% (152)
4	10% (21)	6% (13)
Hospital LOS	2.43 (1.31–4.23)	2.10 (1.25–4.21)	0.713‡
Alert frequency	20 (4–35)	21 (0–43)	0.165‡
RRT call	9% (19)	11% (24)	0.422†
ICU escalation	3% (7)	3% (7)	>0.99†
Total cost	Reference	3.6% (–5.1 to 13.2%)	0.428
VDTC	Reference	5.7% (–4.2 to 16.7%)	0.269

The IPI appears to be non-inferior to multi-parameter monitoring with respect to hospital resource length of stay and other important secondary outcomes, including cost and need for escalation of care. The IPI is designed to simplify the interpretation of vital signs in low-acuity inpatients and potentially address concerns around alarm fatigue with introducing respiratory monitoring. Given the complexity of interpreting EtCO₂ values and waveforms, which may be poorly understood and valued by providers in low-acuity settings, simplified alerts that offer potential to achieve a similarly safe level of monitoring are of interest. The IPI further simplifies the process, as it is calculated automatically by the bedside monitor, displayed in real time, and does not require any manual calculation by providers.

The strengths of this study are that it was prospective, groups were well matched, and there was no selection bias aside from timing of intervention. A major limitation is that the study was observational and non-randomised. However, we demonstrate a reasonable safety profile for the use of IPI in low-acuity inpatients, with the potential to improve patient monitoring and prevent adverse events, whilst addressing the important issue of alarm fatigue. We propose conducting larger randomised controlled trials to compare IPI, and possibly other innovative digital monitoring technologies,⁵ to multi-parameter monitoring and better identify patients most likely to benefit from capnography and IPI.

Authors' contributions

Conception and design: R.E.F., B.S.R., M.S.S., W.S.S., P.P.P., L.M.W.

Data acquisition: R.E.F., J.P. Walco, D.M.M., J.P. Wanderer, M.S.S., P.P.P., L.M.W.

Analysis and interpretation of data: R.E.F., J.P. Walco, D.M.M., J.P. Wanderer, M.S.S., P.P.P., L.M.W.

Statistical conduct: M.S.S.

Drafting of manuscript: R.E.F., J.P. Walco.

Manuscript revision for important intellectual content: D.M.M., J.P. Wanderer, B.S.R., M.S.S., W.S.S., P.P.P., L.M.W.

All authors provide final approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Declaration of interest

The conduct of this study was funded by a grant from Medtronic. The investigators were responsible for all data collection and statistical analysis, which were performed independent of Medtronic.

Funding

Medtronic (COVMOPO0525); KL2 grant via the Vanderbilt Faculty Research Scholars Program (1KL2 TR002245) to R.E.F.