ORIGINAL RESEARCH

Modeling the Cost Savings of Continuous PulseOximetry and Capnography Monitoring of UnitedStates General Care Floor Patients Receiving OpioidsBased on the PRODIGY Trial

Ashish K. Khanna . Carla R. Jungquist . Wolfgang Buhre .

Roy Soto . Fabio Di Piazza . Leif Saager on behalf of the PRediction of Opioid-induced respiratory Depression In

patients monitored by capnoGraphY (PRODIGY) Group Investigators

Received: April 7, 2021 /Accepted: May 6, 2021 / Published online: May 24, 2021� The Author(s) 2021

ABSTRACT

Introduction: Despite the high incidence ofrespiratory depression on the general care floorand evidence that continuous monitoringimproves patient outcomes, the cost–benefit ofcontinuous pulse oximetry and capnography

monitoring of general care floor patientsremains unknown. This study modeled the costand length of stay savings, investment break-even point, and likelihood of cost savings forcontinuous pulse oximetry and capnographymonitoring of general care floor patients at riskfor respiratory depression.Methods: A decision tree model was created tocompare intermittent pulse oximetry versuscontinuous pulse oximetry and capnographymonitoring. The model utilized costs andThe members of the PRediction of Opioid-induced

respiratory Depression In patients monitored bycapnoGraphY (PRODIGY) Group Investigators are listedin acknowledgements.

Supplementary Information The online versioncontains supplementary material available at https://doi.org/10.1007/s12325-021-01779-7.

A. K. Khanna (&)Department of Anesthesiology, Section on CriticalCare Medicine, Wake Forest School of Medicine,Wake Forest Center for Biomedical Informatics,Critical Illness, Injury and Recovery Research Center(CIIRRC), Medical Center Boulevard, Winston-Salem, NC 27157, USAe-mail: akhanna@wakehealth.edu

A. K. KhannaOutcomes Research Consortium, Cleveland, OH,USA

C. R. JungquistUniversity at Buffalo School of Nursing, Buffalo, NY,USA

W. BuhreDepartment of Anesthesiology, University MedicalCenter, Maastricht, The Netherlands

R. SotoDepartment of Anesthesiology, Beaumont Hospital,Royal Oak, MI, USA

F. Di PiazzaMedtronic Core Clinical Solutions, Study andScientific Solutions, Rome, Italy

L. SaagerDepartment of Anesthesiology, University ofMichigan Medical School, Ann Arbor, MI, USA

L. SaagerDepartment of Anesthesiology, University MedicalCenter Goettingen, Goettingen, Germany

Adv Ther (2021) 38:3745–3759

https://doi.org/10.1007/s12325-021-01779-7

outcomes from the PRediction of Opioid-in-duced respiratory Depression In patients moni-tored by capnoGraphY (PRODIGY) trial, andwas applied to a modeled cohort of 2447patients receiving opioids per median-sizedUnited States general care floor annually.Results: Continuous pulse oximetry andcapnography monitoring of high-risk patients isprojected to reduce annual hospital cost by$535,531 and cumulative patient length of stayby 103 days. A 1.5% reduction in respiratorydepression would achieve a break-even invest-ment point and justify the investment cost. Theprobability of cost saving is C 80% if respiratorydepression is decreased by C 17%. Expansion ofcontinuous monitoring to high- and interme-diate-risk patients, or to all patients, is projectedto reach a break-even point when respiratorydepression is reduced by 2.5% and 3.5%,respectively, with a C 80% probability of costsavings when respiratory depression decreasesby C 27% and C 31%, respectively.Conclusion: Compared to intermittent pulseoximetry, continuous pulse oximetry andcapnography monitoring of general care floorpatients receiving opioids has a high chance ofbeing cost-effective.Trial Registration: www.clinicaltrials.gov,Registration ID: NCT02811302.

Keywords: Break-even analysis; Capnography;Continuous monitoring; Cost savings;Economic model; General care floor;Healthcare economics; Pulse oximetry;Respiratory compromise; Respiratory depression

Key Summary Points

Why carry out this study?

Respiratory depression occurs in 46% ofpatients receiving opioids on the generalcare floor, where standard of caremonitoring consists of intermittent pulseoximetry spot-checks.

Continuous pulse oximetry andcapnography monitoring can detectrespiratory depression, but thecost–benefit of continuous pulse oximetryand capnography monitoring isunknown.

The purpose of this study was to model thecost and length of stay savings,investment break-even point, andlikelihood of cost savings for continuouspulse oximetry and capnographymonitoring of general care floor patientsat risk for respiratory depression.

What was learned from the study?

Continuous pulse oximetry andcapnography monitoring of high-riskpatients could reduce annual hospital costby $535,531 and cumulative patientlength of stay by 103 days, reaching abreak-even investment point when theincidence of respiratory depressiondecreases by 1.5%.

3746 Adv Ther (2021) 38:3745–3759

Compared to intermittent pulse oximetry,continuous pulse oximetry andcapnography monitoring of general carefloor patients receiving opioids has a highchance of being cost-effective.

PLAIN LANGUAGE SUMMARY

Respiratory depression occurs when a personhas an abnormally slow breath rate, low oxygensaturation, low or high concentration ofexhaled carbon dioxide, or stops breathingintermittently. This condition occurs in 46% ofpatients receiving opioids in medical and sur-gical hospital units. Respiratory depression canbe detected using continuous respiratory mon-itoring to measure respiratory rate, heart rate,blood oxygen saturation, and exhaled carbondioxide. However, the cost–benefit of continu-ous respiratory monitoring in patients hospi-talized in medical and surgical units isunknown. We created an economic model topredict differences in hospital cost and patientlength of stay when using continuous respira-tory monitoring, compared to intermittentblood oxygen saturation and heart rate moni-toring. This model predicts the break-evenpoint where the cost of investing in monitoringtechnology will equal the costs saved by pre-venting respiratory depression, and evaluatesthe chance that continuous respiratory moni-toring will be cost saving. If patients at highestrisk for respiratory depression are continuouslymonitored, the model projects annual hospitalcost will decrease by $535,531, and total lengthof stay of all patients will decrease by 103 days.The cost of investing in monitoring is predictedto equal the cost savings if respiratory depres-sion decreases by 1.5%, and there is a highprobability of the investment being cost saving.The model also predicts that continuous mon-itoring will reduce annual cost and length ofstay if patients at high and intermediate risk, orif all patients, undergo continuous respiratorymonitoring. Overall, continuous respiratorymonitoring has a high chance of being cost-effective.

DIGITAL FEATURES

This article is published with digital features,including a summary slide and plain languagesummary, to facilitate understanding of thearticle. To view digital features for this article goto https://doi.org/10.6084/m9.figshare.14541750.

INTRODUCTION

Respiratory depression is common on hospitalgeneral care floors, where nurses traditionallyassess vital signs every 4–6 h [1–3]. However,compared to intermittent assessment of respi-ratory status, continuous electronic monitoringdetects significantly more cardiorespiratoryevents [1, 4]. For example, 20% of all postop-erative patients experience significant timeunder hypoxemic thresholds, and almost all goundetected with intermittent monitoring [1].Low respiratory rate occurs in 41% of patients inthe post-anesthesia care unit and general carefloor [5].

Recently, the PRediction of Opioid-inducedrespiratory Depression In patients monitored bycapnoGraphY (PRODIGY) trial reported that46% of postsurgical and medical patientsreceiving parenteral opioids experienced respi-ratory depression episodes [2]. This prospectivetrial comprehensively assessed respiratorydepression, monitoring patients with blindedcontinuous pulse oximetry and capnography,including strict time and threshold cutoffs todefine respiratory depression [2]. The PRODIGYtrial found that adverse events requiring rescueaction or prolonged hospitalization occurredmore often in patients with monitor-detectedrespiratory depression episodes [2, 6]. This isconsistent with prior reports using continuouspulse oximetry or capnography only [1, 5], fur-ther substantiating that respiratory depressionmay be preventable with better monitoring andearlier intervention [2, 6, 7].

Respiratory opioid-related adverse events(ORADE) are costly to healthcare systems[8–10]. In particular, unrecognized respiratorydepression contributes to poor patient out-comes and longer lengths of stay that increase

Adv Ther (2021) 38:3745–3759 3747

healthcare cost [6, 11–13]. Compared to UnitedStates (US) general care floor patients withoutrespiratory depression, hospital cost is $3686higher in patients with respiratory depression[6]. Although professional organizations andtheir clinical practice guidelines recommendthe use of electronic respiratory monitoring, themajority of US hospitals do not have an ade-quate supply of devices [14, 15]. This is becausedespite current evidence that early recognitionof respiratory depression can potentiallydecrease hospital cost, instituting continuousmonitoring is perceived as requiring significantfinancial investment [16].

The purpose of this analysis was to evaluatethe economic value of continuous monitoringby modeling the investment break-even pointand likelihood of cost savings with continuouspulse oximetry and capnography monitoringcompared to intermittent pulse oximetry mon-itoring of general care floor patients receivingopioids.

METHODS

A model was developed to estimate the invest-ment break-even point and likelihood of costsavings associated with implementation ofcontinuous pulse oximetry and capnographymonitoring of US general care floor patients atrisk for respiratory depression. The model,which follows good practice guidelines (Inter-national Society of Pharmacoeconomics andOutcomes Research) [17], was created using adecision tree framework in Excel (Microsoft,Redmond, WA). The model considers a budgetholder perspective for a median-sized US hos-pital and projects cost over a 1-year timehorizon.

Model Structure

The model was created to compare costs andoutcomes for (1) standard of care intermittentpulse oximetry monitoring versus (2) continu-ous respiratory monitoring with pulse oximetryand capnography of medical and surgicalpatients receiving opioids on the general carefloor. Within the continuously monitored

group, the model simulates outcomes for threescenarios: monitoring of patients at (1) highrisk, (2) high and intermediate risk, or (3) anylevel (high, intermediate, or low) risk of respi-ratory depression, determined using thePRODIGY score (Fig. 1) [2]. This model consid-ers three PRODIGY trial outcomes: (1) incidenceof respiratory depression episodes, (2) length ofstay, and (3) total hospital admission costs.Owing to lack of interventional data, the effectof implementing continuous monitoring wasmodeled from 0% to 100% reduction in respi-ratory depression.

PRODIGY Trial

The PRODIGY trial enrolled 1495 patients ageC 18, 20, and 21 across the USA and Europe,Japan, and Singapore, respectively. This posthoc analysis included only US patients whoreceived parenteral opioids and underwentblinded continuous pulse oximetry andcapnography monitoring on the general carefloor (N = 758) [2, 18]. Continuous monitoringcontinued up to 48 h using the CapnostreamTM

20p or 35 portable bedside monitor (Medtronic,Boulder, CO). The PRODIGY trial was con-ducted following institutional review board orethics committee approval, with writteninformed consent collected from all patients.Respiratory depression episodes included respi-ratory rate B 5 breaths/minute, oxygen satura-tion B 85%, or end-tidal carbon dioxide B 15or C 60 mmHg for C 3 min; apnea episodefor[ 30 s; or any respiratory opioid-relatedadverse event [2, 18]. Each patient was retro-spectively assessed for the risk of respiratorydepression using the PRODIGY score, whichwas derived from the PRODIGY dataset using amultivariable logistic regression model [18]. Themodel includes five independent patient char-acteristics, including age C 60 in decades, malesex, sleep disordered breathing, opioid naivety,and chronic heart failure [2]. Each variable isassigned a point value, and patients can beeasily evaluated for their risk of respiratorydepression based on the sum of the pointsassigned, with low (\8 points), intermediate(C 8 and \15 points), and high (C 15 points)

3748 Adv Ther (2021) 38:3745–3759

risk categories [2, 6]. Since the PRODIGY datasetwas used to derive the PRODIGY score, for thepurposes of this analysis, each PRODIGY patientwas retrospectively assigned a PRODIGY scorebased on his or her age, sex, opioid naivety, andpresence of sleep disordered breathing andchronic heart failure [6].

Patients missing length of stay and hospitalcost data (N = 11) or data required to determinePRODIGY score (N = 9) were excluded [6]. Todetermine the influence of patient outliers onthe model, the analysis was performed using allpatients with hospital costs available, and sep-arately using a population excluding patientoutliers for hospital cost or length of stay,identified using Cook’s distance [6].

Modeled Cohort

The model extrapolates PRODIGY outcomes toa cohort of general care floor patients receivingopioids in a median-sized US hospital. On thebasis of the 2018 Premier� Healthcare Database,90% of surgical patients and 45% of medicalpatients on US general care floors receive opi-oids, with a median 2447 opioid-receivingpatients per hospital per year (95% CI2092–2802) (Table S1 in Supplementary Mate-rial). This is consistent with recent literature[10, 19–23].

Model Inputs and Data Sources

Twenty percent (N = 171/835), 34% (N = 300/874), and 60% (N = 442/738) of the modeledcohort was allocated to low-, intermediate-, andhigh-risk respiratory depression groups, match-ing the PRODIGY trial distribution (Tables 1and 2). PRODIGY cost and length of stay datawere model inputs (Table 1) [6]. Discount rates,to determine net present value of costs andoutcomes, and inflationary adjustments werenot applied because of the model’s short timehorizon and the recency of the PRODIGY trial.The same model structure was applied sepa-rately utilizing inputs from the PRODIGYcohort excluding outliers (Table S2 in Supple-mentary Material).

For intermittent pulse oximetry monitoring,device pricing consisted of a multiparametermonitor ($3000) prorated over 7 years and areusable pulse oximetry sensor ($65). A UShospital with 27 opioid-receiving general carefloor patients (Premier� Healthcare Database)would need three multiparameter monitors andtwo reusable pulse oximetry sensors per moni-tor, resulting in a device cost of $0.68 perpatient stay. Continuous pulse oximetry andcapnography device pricing consisted of aCapnostreamTM portable respiratory monitor($4300, Medtronic, Boulder, CO) prorated over7 years, a MicrostreamTM capnography filterline

Fig. 1 Model framework, distinguishing between totalannual hospital cost for a standard of care intermittentpulse oximetry monitoring and b implementation of

continuous pulse oximetry and capnography monitoringbased on patient PRODIGY score

Adv Ther (2021) 38:3745–3759 3749

($14.50, Medtronic, Boulder, CO), and a dis-posable NellcorTM pulse oximetry sensor ($8.50,Medtronic, Boulder, CO), for a device cost of$52.73 per monitored patient.

In the base case, the percentage reduction inpatients with C 1 respiratory depression episodewas conservatively assumed at 20% reductionbased on available literature, which reports a34% risk reduction for intensive care unit (ICU)transfer upon implementation of continuouspulse oximetry monitoring, and reduction ofsevere opioid-related adverse events from 3.1/10,000 patients to 0.6/10,000 patients and a50% decrease in rapid response calls due toopioid-induced respiratory depression followingimplementation of continuous capnographymonitoring [16, 24, 25]. Break-even and sensi-tivity analyses did not rely on this assumption.

Modeled Scenarios

In the base case, the model evaluates annualhospital cost and cumulative patient length ofstay using intermittent pulse oximetry moni-toring of all general care floor patients receivingopioids, versus continuous pulse oximetry andcapnography monitoring of general care floorpatients with high respiratory depression riskreceiving opioids (Fig. 1). Additional scenariosevaluate the same outcomes, comparing inter-mittent pulse oximetry monitoring versus con-tinuous respiratory monitoring of general carefloor patients receiving opioids with high andintermediate respiratory depression risk, or anylevel respiratory depression risk. Break-evenanalysis simulated the minimum requirementsneeded to justify the investment, modeling costand length of stay savings as a function of thepercentage respiratory depression reductionfrom 0% to 100%.

Sensitivity Analysis

Probabilistic sensitivity analysis was performedusing 1000 Monte Carlo simulations [26] todetermine the probability of continuous pulseoximetry and capnography monitoring beingcost saving. The model estimates the invest-ment break-even point and likelihood of cost

savings when respiratory depression casesdecrease from 0% to 100%. Parameters wereassigned a probability distribution, includingdistinction of costs (gamma distribution), andepidemiological parameters (beta distributions),and accounted for the standard error of PROD-IGY dataset parameters. One-way deterministicsensitivity analysis was performed to examinethe effects of model parameters on cost andlength of stay outcomes. The uncertainty ofeach parameter was determined using variabil-ity estimated from the PRODIGY dataset.

RESULTS

Base Case

Across all general care floor patients receivingopioids, continuous pulse oximetry andcapnography monitoring of patients with highrisk for respiratory depression would result in$535,531 annual savings and a cumulativelength of stay decrease of 103 days per year(Table 2), compared to standard of care inter-mittent pulse oximetry monitoring. Thisassumes a 20% respiratory depression reductionand an annual general care floor volume of2447 patients receiving opioids per median-sized hospital (Table S1 in SupplementaryMaterial). In this scenario, the model predictsthat a 1.5% reduction in respiratory depressionwould achieve an investment break-even point,with cost and length of stay savings increasinglinearly as respiratory depression decreases(Fig. 2a–b). Reducing respiratory depression by10%, 20%, and 30% would decrease hospitalcosts by $341, $726, and $1110 per patient,respectively (Fig. S1A in SupplementaryMaterial).

Additional Monitoring Scenarios

A second scenario modeled the cost savings ofcontinuously monitoring patients at high andintermediate respiratory depression risk withpulse oximetry and capnography, compared tointermittent pulse oximetry monitoring of allpatients. In this scenario, cumulative patient

3750 Adv Ther (2021) 38:3745–3759

length of stay would decrease by 152 days, withan annual savings of $606,463 (Table 2). Thereduction in respiratory depression required toreach a break-even point was 2.5% (Fig. 2a).

The third scenario, in which all patientsreceiving opioids, regardless of risk for respira-tory depression, undergo continuous respira-tory monitoring, projected a cumulative lengthof stay savings of 204 days, with $688,221 inannual savings, compared to intermittent pulseoximetry monitoring of the same patients(Table 2). The projected break-even point wouldoccur when respiratory depression decreased by3.5% (Fig. 2a). These continuous monitoringscenarios are expected to linearly decrease per-patient cost and length of stay as respiratorydepression decreases (Fig. S1A in SupplementaryMaterial, Fig. 2b).

Monitoring Scenarios Excluding PatientOutliers

To evaluate the impact of patient outliers, aseparate analysis was performed using thePRODIGY cohort excluding patient outliers.Compared to intermittent pulse oximetrymonitoring on all patients, continuous pulseoximetry and capnography monitoring ofpatients with high respiratory depression riskwould result in a 119-day reduction in cumu-lative length of stay and an annual hospital costsavings of $257,561 (Table S3 in SupplementaryMaterial). A 3% reduction in respiratorydepression would achieve an investment break-even point (Fig. S2A in Supplementary Mate-rial), with a $355 per-patient cost savings ifrespiratory depression decreased 20% (Fig. S1Bin Supplementary Material). Similarly, themodel excluding outliers predicts length of stayreductions of 183 and 216 days and annual

Table 1 Model inputs

PRODIGY risk score Low (< 8 points) Intermediate (‡ 8 and< 15 points)

High (‡ 15points)

Patients in risk group (N = 758) 34% (258/758) 36% (273/758) 30% (227/758)

Patients with respiratory depression in risk group

(N = 758)

20% (52/258) 34% (93/273) 60% (136/227)

Mean length of stay (days) (N = 758)

Patients without respiratory depression episodes 5.2 ± 6.4 6.0 ± 6.1 6.4 ± 7.8

Patients with C 1 respiratory depression episode 6.8 ± 9.4 6.8 ± 10.7 7.5 ± 9.1

Mean hospital cost (N = 411)

Patients without respiratory depression episodes $18,633 ± 14,050 $20,331 ± 14,594 $18,608 ± 9714

Patients with C 1 respiratory depression episode $22,316 ± 13,679 $22,272 ± 14,661 $25,057 ± 19,490

Median number of patients receiving opioids on general

care floor per hospital per year

2447

Monitoring cost per patient

Intermittent pulse oximetry $0.68 per stay

Continuous pulse oximetry and capnography $52.73 per stay

Risk score distributions, length of stay, and healthcare cost are from US patients enrolled in the PRODIGY trial. Mediannumber of patients per hospital per year is sourced from the Premier� Healthcare Database, and Medtronic providedmonitoring cost estimates

Adv Ther (2021) 38:3745–3759 3751

Table2

Modelof

costandlength

ofstay

savingswhencontinuous

pulse

oxim

etry

andcapn

ographymonitoringisim

plem

ented

Patient

mon

itoring

scenario

Allpatients:standard

ofcare

interm

ittent

pulseoxim

etry

mon

itoring

Low

-andinterm

ediate-risk

patients:standard

ofcare

interm

ittent

mon

itoring;

High-risk

patients:continuo

uspu

lseoxim

etry

and

capn

ograph

ymon

itoring

Low

-riskpatients:standard

ofcare

interm

ittent

mon

itoring;

Interm

ediate-andhigh-risk

patients:continuo

uspu

lse

oxim

etry

andcapn

ograph

ymon

itoring

Allpatients:continuo

uspu

lse

oxim

etry

andcapn

ograph

ymon

itoring

Occurrenceof

‡1

respiratory

depression

episod

e

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

NPatients,b

yPR

ODIG

Yrisk

group

Low

171/835

664/835

171/835

664/835

171/835

664/835

137/835

698/835

Interm

ediate

300/874

574/874

300/874

574/874

240/874

634/874

240/874

634/874

High

442/738

296/738

353/738

385/738

353/738

385/738

353/738

385/738

Cum

ulativedays

inhospital,b

yPR

ODIG

Yrisk

group

Low

1155

3471

1155

3471

1155

3471

925

3649

Interm

ediate

2041

3434

2041

3434

1633

3793

1633

3793

High

3330

1889

2659

2456

2659

2456

2659

2456

Cum

ulativecostof

monitoring,by

PRODIG

Yrisk

group

Low

$116

$452

$116

$452

$116

$452

$7224

$36,806

Interm

ediate

$204

$390

$204

$390

$12,655

$33,431

$12,655

$33,431

High

$301

$201

$18,614

$20,301

$18,614

$20,301

$18,614

$20,301

Cum

ulativeadmission

cost,b

yPR

ODIG

Yrisk

group

Low

$3,815,951

$12,372,026

$3,815,951

$12,372,026

$3,815,951

$12,372,026

$3,057,224

$13,005,534

Interm

ediate

$6,681,450

$11,670,057

$6,681,450

$11,670,057

$5,345,160

$12,889,924

$5,345,160

$12,889,924

High

$11,075,318

$5,508,107

$8,845,220

$7,164,261

$8,845,220

$7,164,261

$8,845,220

$7,164,261

3752 Adv Ther (2021) 38:3745–3759

Table2

continued

Patient

mon

itoring

scenario

Allpatients:standard

ofcare

interm

ittent

pulseoxim

etry

mon

itoring

Low

-andinterm

ediate-risk

patients:standard

ofcare

interm

ittent

mon

itoring;High-

risk

patients:continuo

uspu

lse

oxim

etry

andcapn

ograph

ymon

itoring

Low

-riskpatients:standard

ofcare

interm

ittent

mon

itoring;Interm

ediate-and

high-riskpatients:continuo

uspu

lseoxim

etry

and

capn

ograph

ymon

itoring

Allpatients:continuo

uspu

lse

oxim

etry

andcapn

ograph

ymon

itoring

Occurrenceof

‡1

respiratory

depression

episod

e

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Patientswith

‡1

respiratory

depression

episod

e

Patients

witho

utrespiratory

depression

episod

es

Totaldays

in

hospital

15,320

15,218

15,168

15,117

Totalcost

$51,124,573

$50,589,042

$50,518,110

$50,436,352

Lengthof

stay

savings(days)

Reference

103

152

204

Costsavings

Reference

$535,531

$606,463

$688,221

Percentage

respiratory

depression

reductionneeded

tobreakeven

Reference

1.5%

2.5%

3.5%

Results

arereported

inpatients

receivingopioidson

thegeneralcare

floor

withhigh,high

orinterm

ediate,or

high,interm

ediate,or

low

risk

forrespiratory

depression.M

odelwas

derivedon

thebasisof

theUSPR

ODIG

Ycohortwithcostdata

available,includingoutliers

Adv Ther (2021) 38:3745–3759 3753

savings of $380,405 and $497,734 followingimplementation of continuous respiratorymonitoring on patients with high and inter-mediate risk of respiratory depression, or on allpatients, respectively (Table S3 in Supplemen-tary Material). In these scenarios, the break-even point would occur when respiratorydepression decreases by 4–4.5%, with a lineardecrease in length of stay as respiratory depres-sion decreases (Fig. S2A-B in SupplementaryMaterial).

Probabilistic Sensitivity Analysis

To achieve a C 80% probability of cost savingswhen patients with high respiratory depressionrisk undergo continuous respiratory monitor-ing, a decrease in respiratory depression C 17%would be needed. This increases to a C 96%probability of cost savings if respiratorydepression is decreased by C 30% (Fig. 3). Inscenarios in which continuous monitoring isapplied to high- and intermediate-risk patients,or to all patients, a C 27% and C 31% reductionin respiratory depression would be needed toachieve a C 80% probability of being cost sav-ing, respectively (Fig. 3).

Probabilistic sensitivity analysis of the modelusing the PRODIGY cohort excluding outliersprojects that when patients with high, high orintermediate, or any respiratory depression risk

undergo continuous monitoring, a respiratorydepression decrease of C 27%, C 35%, andC 35%, respectively, would be needed toachieve a C 80% probability of being cost sav-ing (Fig. S3 in Supplementary Material).

Deterministic Sensitivity Analysis

Model drivers were evaluated using one-waydeterministic sensitivity analysis. For the modelbased on the PRODIGY cohort with outliers,admission cost of high- and intermediate-riskpatients with respiratory depression had thestrongest influence on model outcomes(Fig. S4A in Supplementary Material). Otherdrivers included the admission cost of high- andintermediate-risk patients without respiratorydepression, the cost of low-risk patients withrespiratory depression, and the number of gen-eral care floor patients receiving opioids peryear. For the model excluding patient outliers,the main model drivers were similar (Fig. S4B inSupplementary Material). Monitoring devicecosts had a minimal influence on results.

DISCUSSION

Our analysis suggests that a reduction ofC 1.5%, C 2.5%, and C 3.5% in respiratorydepression episodes would justify the

Fig. 2 a Annual cost savings (US dollars) and b length ofstay reduction predicted following implementation ofcontinuous pulse oximetry and capnography monitoringon patients with high (blue line), high or intermediate (red

line), or high, intermediate, or low (green line) risk forrespiratory depression. Model was derived the on basis ofthe US PRODIGY cohort with cost data available,including outliers

3754 Adv Ther (2021) 38:3745–3759

investment for continuous pulse oximetry andcapnography monitoring when continuouslymonitoring high, high and intermediate, or allpatients, respectively. Projected annual costsavings, assuming a 20% reduction in respira-tory depression, would be $535,531 to$688,221, and reduction in cumulative lengthof stay would be 103–204 days annually,depending on the PRODIGY risk group(s) mon-itored. Furthermore, implementation of con-tinuous respiratory monitoring has C 80%probability of being cost saving if respiratorydepression is reduced by C 17%, C 27%, andC 31% if implemented on high-risk, high- andintermediate-risk, and all patients, respectively.Importantly, implementation of continuousmonitoring alone will not decrease respiratorydepression or adverse events, but can alertbedside providers to respiratory depression andfacilitate early intervention. A true decrease inrespiratory depression is dependent on howbedside providers respond to these alerts.

Our analysis showed that the top modeldrivers include the cost burden of respiratory

depression and general care floor volume, notthe cost of monitoring equipment. This may beexplained by the high cost of respiratorydepression events. Specifically, monitoringequipment per patient may cost less to hospitalsthan a respiratory depression event, whichsometimes leads to rapid response team activa-tion and ICU transfer [6, 16]. The cost burden ofsurgical patients with ORADEs are reported tobe $4350–8225 [10, 21, 27, 28], representing a27–47% increase in admission cost. This isconsistent with the cost burden of respiratorydepression observed in the PRODIGY trial,which were inputs in this model [6].

We also evaluated the effect of patient out-liers on modeling results. Model outcomes wereconsistent across both analyses, with a lowbreak-even point for implementation of con-tinuous respiratory monitoring and decreasedannual cost and length of stay followingreduction in respiratory depression.

This model has important implications forUS hospital administrators, value and analysiscommittees, and clinical practice stakeholders

Fig. 3 Probability of cost savings following implementa-tion of continuous pulse oximetry and capnographymonitoring on patients with high (blue line), high orintermediate (red line), or high, intermediate, or low

(green line) risk for respiratory depression. Model wasderived on the basis of the US PRODIGY cohort withcost data available, including outliers

Adv Ther (2021) 38:3745–3759 3755

who are considering (1) whether to continu-ously monitor hospital general care floorpatients receiving opioids, as well as (2) if themonitoring equipment is worth the investmentfor a subset of this patient population. A criticaldecision-making element for hospital adminis-trators is identifying appropriate patients tomonitor, to balance therapeutic benefit whilelimiting expenditure. Importantly, sensitivityanalysis indicated that across the modeled sce-narios, there is a high probability of continuousmonitoring being cost saving, with the highestlikelihood when monitoring high-risk patients.This analysis may be of particular value to cash-strapped hospitals needing to minimize equip-ment expenditure while maximizing patientsafety.

The clinical utility of implementing contin-uous respiratory monitoring using pulseoximetry and capnography is supported bymultiple studies. In one study, all postoperativepatients who had respiratory depressionrequiring intervention were recognized bycapnography, not by pulse oximetry [29].Additionally, a systematic review and meta-analysis reported that compared to intermittentspot-check monitoring, continuous pulseoximetry decreased the risk of ICU transfer 34%,and SpO2\ 90%[1 h was 15 times more likelyto be detected [24]. With continuous capnog-raphy monitoring, 8.6% more respiratorydepression events were detected compared topulse oximetry monitoring alone, and the oddsof respiratory depression detection were 5.83times higher using capnography vs pulseoximetry [24].

The top strengths of this analysis include itsreal-world utility to hospital budget holders, itsnovelty in the literature, and its use of clinicaltrial data as the basis for the economic model.The hospital budget holder perspective, alongwith modeling a range of scenarios to minimizebudget impact while maximizing patient safety,is pertinent since hospitals are likely to be infinancial distress due to the Covid-19 pandemic.This work also adds novelty to the literature, inwhich we are not aware of any comparablemodels for respiratory depression. Finally, thismodel was created on the basis of PRODIGYtrial data, which was the largest prospective

observational trial of continuous multiparame-ter monitoring, where waveform data was col-lected in a blinded manner and subsequentlyevaluated by an independent adjudicationcommittee.

Our work has some limitations. Ourassumption of the number of patients per hos-pital does not apply to all hospitals. However,we used the Premier� database, which is one ofthe largest hospital-based discharge databases inthe country, to determine the modeled cohortsize and represent a median-sized US hospital.While this approach does represent all hospi-tals, a PRODIGY cost savings calculator to tailorthis model to individual hospitals may behelpful but was beyond the scope of this anal-ysis. Second, a 20% reduction in respiratorydepression in the base case is supported bylimited literature demonstrating reducedadverse events following implementation ofcontinuous respiratory monitoring. Impor-tantly, our sensitivity analysis eliminated thisassumption and evaluated the likelihood of costsavings when all possible scenarios of respira-tory depression reduction were modeled from0% to 100%. Third, we assume that bedsideproviders will respond to continuous monitor-ing alerts, resulting in decreased incidence ofrespiratory depression, though data on responseto continuous monitoring alerts is limited. Thismodel did not include the cost of early inter-vention to prevent respiratory depression.Finally, the model is based on US PRODIGYpatients only, which may limit the applicabilityof this model internationally, where opioidadministration, standard of care practices, andhospital costs vary significantly [30–33].

CONCLUSION

Although intermittent monitoring has beenstandard of care for decades, continuous pulseoximetry and capnography monitoring detectmore deviations of vital signs and has thepotential to increase patient safety on the gen-eral care floor [1, 2, 24]. We assessed theinvestment cost–benefit ratio, and found thatcompared to intermittent pulse oximetry mon-itoring, continuous pulse oximetry and

3756 Adv Ther (2021) 38:3745–3759

capnography monitoring of general care floorpatients receiving opioids may be a cost-effec-tive and worthwhile investment from the UShospital budget holder perspective. Our resultssuggest that limiting continuous monitoring tohigh-risk patients may be the least impactful tobudgets, though expansion to intermediate-and low-risk patients may offer good valueconsidering the likely reduction in respiratorydepression and associated costs.

ACKNOWLEDGEMENTS

We thank the participants of the PRediction ofOpioid-induced respiratory Depression Inpatients monitored by capnoGraphY (PROD-IGY) trial for their valuable contributions.

Funding. Medtronic (Boulder, CO) spon-sored the PRODIGY trial and this post hocanalysis, and funded Rapid Service Fees forpublication in Advances in Therapy. The authorshad final responsibility for the decision to sub-mit for publication. The authors were not paidto write this article by the sponsor or any otheragency.

Medical Writing and/or Editorial Assis-tance. Biostatistical analysis was supported byPaola Di Stefano, MS (Medtronic, Rome, Italy).Rhodri Saunders, PhD of Coreva Scientific(Konigswinter, North Rhine-Westphalia Ger-many) contributed to economic modeling.Medical writing support was provided byKatherine E. Liu, PhD (Medtronic, Minneapolis,MN, US) and supported by Medtronic.

Authorship. All named authors meet theInternational Committee of Medical JournalEditors (ICMJE) criteria for authorship for thisarticle, take responsibility for the integrity ofthe work as a whole, and have given theirapproval for this version to be published.

Authorship Contributions. Concept anddesign of work: Ashish K. Khanna, WolfgangBuhre, Fabio Di Piazza, Leif Saager. Acquisition,analysis, or interpretation of data for the work:

Ashish K. Khanna, Carla R. Jungquist, WolfgangBuhre, Roy Soto, Fabio Di Piazza, Leif Saager.Drafting of the manuscript: Ashish K. Khanna,Fabio Di Piazza. Critical revision of the manu-script for important intellectual content: AshishK. Khanna, Carla R. Jungquist, Wolfgang Buhre,Roy Soto, Fabio Di Piazza, Leif Saager. Approvalof the final version: Ashish K. Khanna, Carla R.Jungquist, Wolfgang Buhre, Roy Soto, Fabio DiPiazza, Leif Saager.

List of Investigators. PRODIGY investiga-tors included: Ashish K. Khanna, MD; Sergio D.Bergese, MD; Carla R. Jungquist, NP, PhD; Hir-oshi Morimatsu, MD, PhD; Shoichi Uezono,MD; Simon Lee, MD; Lian Kah Ti, MBBS, MMed;Richard D. Urman, MD; Robert McIntyre Jr,MD; Carlos Tornero, MD, PhD; Albert Dahan,MD, PhD; Leif Saager, Dr.Med; Toby N. Wein-garten, MD; Maria Wittmann, MD; DennisAuckley, MD; Luca Brazzi, MD, PhD; Morgan LeGuen, MD, PhD; Roy Soto, MD; FrankSchramm, MD; Wolfgang Buhre, MD; and FrankJ. Overdyk, MD.

Prior Presentation. A portion of this workwas presented as an abstract at the Society forTechnology in Anesthesia 2021 Virtual AnnualMeeting.

Disclosures. Ashish K. Khanna, Carla R.Jungquist, Wolfgang Buhre, Roy Soto, and LeifSaager (or their institutions) received researchsupport from Medtronic to conduct the PROD-IGY trial. In addition, Ashish K. Khanna reportsconsulting fees from Medtronic, Edwards Life-sciences, and Philips North America and isfunded with a Clinical and Translational Sci-ence Institute (CTSI) NIH/NCTAS KL2TR001421 award for a trial on continuouspostoperative hemodynamic and saturationmonitoring; Carla R. Jungquist reports partici-pation in the Medtronic Nurse Advisory Group;Wolfgang Buhre grants from the EuropeanUnion and Interreg Consortium, and personalfees from European Society of Anaesthesiologystudies (PHOENICS and TETHYS) supported byB Braun Medical and Fresenius Medical Care,and from Medtronic; Fabio Di Piazza reportsemployment with Medtronic; and Leif Saager

Adv Ther (2021) 38:3745–3759 3757

reports a grant from Merck & Co. Inc. andconsultant fees from Merck & Co. Inc, The 37Company, and Ferrer International. Ashish K.Khanna is currently affiliated with the Depart-ment of Anesthesiology, Section on CriticalCare Medicine, Wake Forest School of Medicine(Winston-Salem, North Carolina, US), and LeifSaager is currently affiliated with the Depart-ment of Anesthesiology, University MedicalCenter Goettingen (Goettingen, Germany).

Compliance with Ethics Guidelines. ThePRODIGY trial was conducted following insti-tutional review board or ethics committeeapproval for each trial site, and in accordancewith the Helsinki Declaration of 1964 and itslater amendments. The names of all reviewboards and ethics committees are provided inTable S4 of the Supplementary Material. Writ-ten informed consent was collected from allpatients before participation in the trial.

Data Availability. All PRODIGY data gen-erated or analyzed during this study are inclu-ded in this published article/as supplementaryinformation files. The data analyzed from thePremier� Healthcare Database were used underlicense for the current study and are not pub-licly available, but data may be available onreasonable request and with the permission ofPremier�.

Open Access. This article is licensed under aCreative Commons Attribution-NonCommer-cial 4.0 International License, which permitsany non-commercial use, sharing, adaptation,distribution and reproduction in any mediumor format, as long as you give appropriate creditto the original author(s) and the source, providea link to the Creative Commons licence, andindicate if changes were made. The images orother third party material in this article areincluded in the article’s Creative Commonslicence, unless indicated otherwise in a creditline to the material. If material is not includedin the article’s Creative Commons licence andyour intended use is not permitted by statutoryregulation or exceeds the permitted use, youwill need to obtain permission directly from thecopyright holder. To view a copy of this licence,

visit http://creativecommons.org/licenses/by-nc/4.0/.

REFERENCES

1. Sun Z, Sessler DI, Dalton JE, et al. Postoperativehypoxemia is common and persistent: a prospec-tive blinded observational study. Anesth Analg.2015;121(3):709–15.

2. Khanna AK, Bergese SD, Jungquist CR, et al. Pre-diction of opioid-induced respiratory depression oninpatient wards using continuous capnography andoximetry: an international prospective, observa-tional trial. Anesth Analg. 2020;131(4):1012–24.

3. Gupta K, Prasad A, Nagappa M, Wong J, Abra-hamyan L, Chung FF. Risk factors for opioid-in-duced respiratory depression and failure to rescue: areview. Curr Opin Anaesthesiol. 2018;31(1):110–9.

4. Turan A, Chang C, Cohen B, et al. Incidence,severity, and detection of blood pressure perturba-tions after abdominal surgery: a prospective blindedobservational study. Anesthesiology. 2019;130(4):550–9.

5. Overdyk FJ, Carter R, Maddox RR, Callura J, HerrinAE, Henriquez C. Continuous oximetry/capnome-try monitoring reveals frequent desaturation andbradypnea during patient-controlled analgesia.Anesth Analg. 2007;105(2):412–8.

6. Khanna AK, Saager L, Bergese SD, et al. Opioid-in-duced respiratory depression increases hospital costsand length of stay in patients recovering on thegeneral care floor. BMC Anesthesiol. 2021;21(88):1.

7. Lee LA, Caplan RA, Stephens LS, et al. Postoperativeopioid-induced respiratory depression: a closedclaims analysis. Anesthesiology. 2015;122(3):659–65.

8. Oderda GM, Evans RS, Lloyd J, et al. Cost of opioid-related adverse drug events in surgical patients.J Pain Symptom Manage. 2003;25(3):276–83.

9. Kane-Gill SL, Rubin EC, Smithburger PL, BuckleyMS, Dasta JF. The cost of opioid-related adversedrug events. J Pain Palliat Care Pharmacother.2014;28(3):282–93.

10. Urman RD, Seger DL, Fiskio JM, et al. The burden ofopioid-related adverse drug events on hospitalizedpreviously opioid-free surgical patients. J PatientSaf. 2021;17(2):e76–83.

11. Taenzer AH, Pyke JB, McGrath SP, Blike GT. Impactof pulse oximetry surveillance on rescue events and

3758 Adv Ther (2021) 38:3745–3759

intensive care unit transfers: a before-and-afterconcurrence study. Anesthesiology. 2010;112(2):282–7.

12. Fischer JP, Shang EK, Butler CE, et al. Validatedmodel for predicting postoperative respiratory fail-ure: analysis of 1706 abdominal wall reconstruc-tions. Plast Reconstr Surg. 2013;132(5):826e-e835.

13. Cozowicz C, Olson A, Poeran J, et al. Opioid pre-scription levels and postoperative outcomes inorthopedic surgery. Pain. 2017;158(12):2422–30.

14. Dunwoody D, Jungquist CR. Monitoring for opioid-induced advancing sedation and respiratorydepression: ASPMN membership survey of currentpractice 2019. 2019 ASPMN Annual Conference,Portland, Oregon. 2019.

15. Jungquist CR, Quinlan-Colwell A, Vallerand A,et al. American Society for Pain Managementnursing guidelines on monitoring for opioid-in-duced advancing sedation and respiratory depres-sion: revisions. Pain Manag Nurs. 2020;21(1):7–25.

16. Stites M, Surprise J, McNiel J, Northrop D, De RuyterM. Continuous capnography reduces the incidenceof opioid-induced respiratory rescue by hospitalrapid resuscitation team. J Patient Saf. 2017. https://doi.org/10.1097/PTS.0000000000000408.

17. Husereau D, Drummond M, Petrou S, et al. Con-solidated health economic evaluation reportingstandards (CHEERS)–explanation and elaboration: areport of the ISPOR health economic evaluationpublication guidelines good reporting practices taskforce. Value Health. 2013;16(2):231–50.

18. Khanna AK, Overdyk FJ, Greening C, Di Stefano P,Buhre WF. Respiratory depression in low acuityhospital settings—seeking answers from thePRODIGY trial. J Crit Care. 2018;47:80–7.

19. Overdyk FJ, Dowling O, Marino J, et al. Associationof opioids and sedatives with increased risk of in-hospital cardiopulmonary arrest from an adminis-trative database. PLoS One. 2016;11(2):e0150214.

20. Herzig SJ, Rothberg MB, Cheung M, Ngo LH, Mar-cantonio ER. Opioid utilization and opioid-relatedadverse events in nonsurgical patients in US hos-pitals. J Hosp Med. 2014;9(2):73–81.

21. Shafi S, Collinsworth AW, Copeland LA, et al.Association of opioid-related adverse drug eventswith clinical and cost outcomes among surgicalpatients in a large integrated health care deliverysystem. JAMA Surg. 2018;153(8):757–63.

22. Donohue JM, Kennedy JN, Seymour CW, et al.Patterns of opioid administration among opioid-naive inpatients and associations with

postdischarge opioid use: a cohort study. AnnIntern Med. 2019;171(2):81–90.

23. Niroula A, Garvia V, Rives-Sanchez M, et al. Opiateuse and escalation of care in hospitalized adultswith acute heart failure and sleep-disorderedbreathing (OpiatesHF study). Ann Am Thorac Soc.2019;16(9):1165–70.

24. Lam T, Nagappa M, Wong J, Singh M, Wong D,Chung F. Continuous pulse oximetry and capnog-raphy monitoring for postoperative respiratorydepression and adverse events: a systematic reviewand meta-analysis. Anesth Analg. 2017;125(6):2019–29.

25. Steele T, Eidem L, Bond J. Impact of adoption ofsmart pump system with continuous capnographymonitoring on opioid-related adverse event rates:experience from a tertiary care hospital. J PatientSaf. 2019;16:e194.

26. Briggs A, Sculpher M, Claxton K. Decision mod-elling for health economic evaluation. Oxford:OUP; 2006.

27. Oderda GM, Gan TJ, Johnson BH, Robinson SB.Effect of opioid-related adverse events on outcomesin selected surgical patients. J Pain Palliat CarePharmacother. 2013;27(1):62–70.

28. Kessler ER, Shah M, Gurschkus SK, Raju A. Cost andquality implications of opioid-based postsurgicalpain control using administrative claims data froma large health system: opioid-related adverse eventsand their impact on clinical and economic out-comes. Pharmacotherapy. 2013;33(4):383–91.

29. McCarter T, Shaik Z, Scarfo K, Thompson LJ.Capnography monitoring enhances safety of post-operative patient-controlled analgesia. Am HealthDrug Benefits. 2008;1(5):28–35.

30. Berterame S, Erthal J, Thomas J, et al. Use of andbarriers to access to opioid analgesics: a worldwide,regional, and national study. Lancet.2016;387(10028):1644–56.

31. Burden M, Keniston A, Wallace MA, et al. Opioidutilization and perception of pain control in hos-pitalized patients: a cross-sectional study of 11 sitesin 8 countries. J Hosp Med. 2019;14(12):737–45.

32. Ladha KS, Neuman MD, Broms G, et al. Opioidprescribing after surgery in the United States,Canada, and Sweden. JAMA Netw Open. 2019;2(9):e1910734.

33. Papanicolas I, Woskie LR, Jha AK. Health carespending in the united states and other high-in-come countries. JAMA. 2018;319(10):1024–39.

Adv Ther (2021) 38:3745–3759 3759