Early Detection of Deterioration in COVID-19 Patients by Continuous Ward Respiratory Rate Monitoring: A Pilot Prospective Cohort Study

doi:10.21203/rs.3.rs-2460282/v1

Download PDF

Research Article

Early Detection of Deterioration in COVID-19 Patients by Continuous Ward Respiratory Rate Monitoring: A Pilot Prospective Cohort Study

https://doi.org/10.21203/rs.3.rs-2460282/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Tachypnea is among the earliest signs of pulmonary decompensation. Contactless continuously respiratory rate monitoring might beuseful in isolated COVID-19 patients admitted in wards. We aim to determine whether continuous untethered ward respiratory rate patterns in hospitalized COVID-19 identify patients who require escalation of pulmonary management 24 hours ahead of time.

Methods

Single-center pilot prospective cohort study in COVID-19 patients who were cared for in routine wards. COVID-19 patients who has at least one escalation of pulmonary management were matched to 3 non-escalated patients. Contactless Breathing Monitoring was instituted after patients enrolled, and continued for 15 days unless hospital discharge, initiation of invasive mechanical ventilation, or death occurred. Respiratory rate data from the continuous monitor was not available to clinicians. The exposures were respiratory features over rolling periods of 30 min, 24 hours, and 72 hours before respiratory care escalation. The primary outcome was escalation in the pulmonary care beyond Venturi-mask.

Results

Among 125 included patients, 13 exhibited at least one escalation and were each matched to 3 non-escalated patients. A total of 28 escalation events were matched to 84 non-escalation episodes. The 30-min mean respiratory rate in escalated patients was 23 breaths per minute (bpm) ranging from 13 to 40 bpm, similar to the 22 bpm in non-escalated patients, although with less variability (range 14 to 31 bpm). However, higher respiratory rate variability, especially skewness over 1 day, was associated with higher incidence of an escalation event. Our overall model, based on continuous data, had a moderate accuracy with an AUC 0.81 (95%CI:0.73,0.88) and a good specificity 0.93 (95%CI:0.87,0.99).

Conclusions

Our pilot observational study suggests that continuous respiratory monitoring and respiratory rate variability are associated with the need for care escalation 24 hours in advance. ur results suggest that continuous respiratory monitoring is a valuable increment over intermittent monitoring.

The COVID-19 pandemic has affected almost 600 million people and is on track to kill 6.5 million people worldwide.[1] During the first wave, most patients experienced mild-moderate symptoms, but up to 10% required hospitalization.[2, 3] Among those admitted to the hospital, depending on various comorbidities, a substantial fraction progressed to respiratory failure requiring ventilatory support.[4]

Early detection of deteriorating patients may improve management and outcomes. Tachypnoea is among the earliest signs of pulmonary decompensation. For example, respiratory rates exceeding 35 breaths/min predict life-threatening events within 8 hours (odds ratio [OR], 31; 95%CI: 8, 130).[5] And in COVID-19 patients, tachypnoea has been associated with progression to mechanical ventilation,[6] ICU admission,[7] and mortality.[2]

Respiratory depression[8]and hypoxemia[9]are common among hospitalized patients, even those without COVID-19. Furthermore, nurses assessing vital signs at 4-hour intervals in hospital wards miss half of all serious and prolonged hypotensive episodes,[10] more than 90% of serious and prolonged desaturation episodes.[9, 10] An additional consideration is that healthcare professionals’ manually-obtained respiratory rate measurements are often inaccurate,[11, 12] and recorded values often default to either 18 or 20 breaths per minute (bpm).[13] Automated continuous monitoring tools, although not immune to artefacts, have well-established accuracy and artefact-caused errors can be averaged out through continuous monitoring.

Recent advances in sensor technologies such as radar have made it possible to obtain respiratory rate without direct attachment to patients or their beds. Such contactless monitoring technologies, by virtue of not requiring effort from patient or healthcare professionals, facilitate continuous vital sign monitoring. Use of continuous respiratory rate data has two additional advantages over infrequent spot-check data for prediction of clinical events. Firstly, naturally occurring short term variation in breathing rate is fully sampled and can thus be averaged out, resulting in a more precise estimate of mean respiratory rate.[14] Secondly, variation of respiratory rate over time may contain clinically relevant information. For example, trends and variations in respiratory rate across various intervals predict adverse events.[15–17]

Undetected vital sign abnormalities may be especially likely in COVID-19 wards because patient isolation and the need of personal protection equipment preclude facile clinical assessments. Automated and continuous monitoring of respiratory function may thus be especially helpful in COVID-19 patients sick enough to require hospitalization. We therefore conducted a single-centre pilot prospective cohort study in hospitalized COVID-19 patients. Our primary goal was to assess the extent to which continuous untethered ward respiratory rate patterns identify patients who require escalation of pulmonary management 24 hours ahead of time, defined by progression to high-flow oxygen, non-invasive mechanical ventilation, invasive mechanical ventilation, or death. Specifically, we sought to identify whether identifiable respiratory rate patterns are meaningfully associated with ventilatory decompensation in COVID-19 patients.

Our prospective observational pilot study was conducted with approval from the Hospital Clinic of Barcelona IRB (Ethics Committee's approval/ID: HCB 2020/0666). Written informed consent was obtained from participating patients. The study was performed in accordance with relevant guidelines and regulations.

Subject selection

We enrolled adults of any race and ethnicity who were admitted to a routine nursing ward with a diagnosis of SARS-CoV-2 infection, confirmed by a reverse-transcription polymerase chain reaction test. We excluded patients younger than 18 years old and pregnant women. Because we targeted patients at risk of clinical deterioration who were eligible for non-invasive and invasive mechanical ventilation support, we also excluded patients in whom escalation to life-sustaining respiratory support was not anticipated because treatment was unlikely to prove helpful.

Protocol

We included patients who had COVID-19 and were hypoxemic, defined as arterial pressure of oxygen (PaO₂) < 80 mmHg or pulse oximeter saturation (SpO₂) < 90% while breathing ambient air. All were hospitalized for treatment and required supplemental oxygen, to ensure a SpO₂ over 91%.

Patients were considered to be in respiratory failure and candidates for escalation to non-invasive ventilation when they developed moderate-to-severe shortness-of-breath and tachypnoea of ≥ 30 bpm; when the PaO₂/Inspired fraction of oxygen (FiO₂) ratio was ≤ 200; if a FiO₂ > 0.4 was needed to maintain a SpO₂ > 92%; or when pH was < 7.35 accompanied by a PaCO₂ exceeding 45 mmHg.

High-flow nasal cannula oxygen (HFNCO) was generally the initial treatment, followed by Non-Invasive Mechanical Ventilation (NIMV). Patients were escalated to invasive mechanical ventilation if they did not improve and had a ROX index [(SpO₂/FiO₂) / respiratory rate] < 3 within 2 hours, < 3.5 within 6 hours, or < 4 within 12 hours after HFNCO and NIVM initiation.

Measurements

Demographic, morphometric, and clinical characteristics, as well as nurse-recorded vital signs including blood pressure, heart rate, respiratory rate, oxygen saturation, ventilation mode, supplemental oxygen concentration and delivery system were recorded.

Pulmonary management was recorded, including use of HFNCO, NIMV, invasive mechanical ventilation, and clinically detected deterioration. We also recorded episodes when nurses requested non-routine physician or ICU team evaluation, transfer to a higher-acuity unit, and discharge disposition including vital status. Arterial blood gas analysis was not mandated per protocol, but available values were extracted from medical records.

The Circadia Contactless Breathing Monitor (model C100, Circadia Technologies, Ltd., London, United Kingdom) is an FDA-cleared non-contact device that uses radar to measure patient respiratory rate. Accuracy of the Circadia System has been validated by direct comparison against manually scored reference end-tidal CO₂ capnography and polygraphy ventilatory effort, both during spot measurements and continuous monitoring in awake and sleeping patients.[18]

We began Circadia monitoring as soon as practical after patients consented to participate in our study. Circadia Monitors were positioned about 1.5 meters from patients. Monitoring continued for 15 days, unless aborted for hospital discharge, initiation of invasive mechanical ventilation, or death. Respiratory rate data from the Circadia Monitor were transmitted wirelessly to cloud-based storage for offline analysis and was not available to clinicians who managed participating patients. Respiratory rate from the Circadia Monitor was recorded at 3-second intervals while patients were within the detection range of the Circadia Monitor (typically while in bed).

Exposure and outcomes

Our primary outcome was pulmonary decompensation, defined as an escalation in the pulmonary care beyond Venturi-mask support. The following levels of respiratory support were defined, from less to more severe: 1) ambient air; 2) nasal cannula; 3) Venti-mask; 4) HFNCO; 5) NIMV; 6) invasive mechanical ventilation; and 7) death. Our primary aim was to assess whether the information extracted from respiratory rate continuous monitoring is associated with respiratory care escalation 24 hours in advance in hospitalized COVID-19 patients.

The exposure was short-term and long-term respiratory patterns associated with respiratory care escalation. Respiratory patterns (variability and fluctuations over time) were captured through features, which included respiratory rate mean, respiratory rate standard deviation, kurtosis of the respiratory rate, skewness of the respiratory rate, and trend (extracted from the slope of a linear regression model), all computed over three rolling periods, 30 min, 24 hours and 72 hours in duration, occurring 24 hours prior to respiratory care escalation (Supplemental Table 1).

Data analysis

Analysis was restricted to subjects who had successful Circadia monitoring for at least half of their hospitalization. We considered patients to have escalated respiratory care when they required an increment in their ventilatory support category, with others falling into the non-escalating category. Patients could experience multiple escalation episodes within a day. However, we considered only the initial escalation within each 48-hour period to avoid situations where a prior de-escalation of treatment was quickly unsuccessful, rather than representing a true new worsening of the underlying pulmonary function. We also excluded patients who de-escalated ventilatory support after joining the study. We considered a patient to be de-escalating if they required HFNCO or NIMV at study inclusion and thereafter progressively improved (Supplemental Fig. 1).

Patients who required ventilatory support escalation were matched to three non-escalating patients on baseline variables (Table 1) using minimum Euclidean distance. Escalation episodes in each patient were matched to a similar point in the three matched non-escalation patients based on time since admission. When the exact corresponding time was unavailable because the non-escalation patient was already discharged at this time point, the last available time window was used. A non-escalated patient could be matched with more than 1 escalated patient (Supplemental Table 2).

Table 1

Patient demographic, morphometric, and clinical characteristics
	Before Matching			After Matching
	Escalation			Escalation
	YES (N = 13)	NO (N = 112)	P-value	YES (N = 13)	NO (N = 26)	P-value
Age, years	67 ± 13	61 ± 13	0.06	67 ± 17	63 ± 12	0.322
Female, n (%)	8 (32)	50 (36)	0.049	3 (23)	6 (23)	1
BMI, kg/m²	31 ± 6	28 ± 5	< 0.01	31 ± 5	29 ± 4	0.111
Medical History
Arterial hypertension, n (%)	16 (10)	68 (48)	0.058	8 (61)	19 (73)	0.713
Diabetes mellitus, n (%)	10 (6)	22 (16)	0.021	3 (23)	6 (23)	1
Ischemic cardiac disease, n (%)	2 (1)	9 (6)	0.487	1 (8)	2 (8)	1
Chronic heart failure, n (%)	2 (1)	2 (1)	0.318	2 (15)	1 (4)	0.399
COPD, n (%)	2 (1)	11 (8)	0.275	3(2)	11(23)	1
Asthma, n (%)	1 (1)	11 (8)	0.206	1 (8)	1(4)	1
Neurological disease, n (%)	1 (1)	11 (8)	0.588	1 (8)	1(4)	1
Variables were summarized as means ± SDs and N (%). BMI, body mass index; COPD, Chronic Obstructive Pulmonary Disease. T-tests were used for comparisons. P < 0.05 was considered significant

Respiratory rate was sampled at 3-second intervals. Using a 30-minute rolling analysis window, 30-minute respiratory features were obtained through various aggregation methods, including the mean. The mean respiratory rate, extracted over 30 minutes, was further aggregated into windows of 24 and 72 hours, at 30-minute increments, to obtain the 24-hour and 72-hour features. Extracted respiratory features included mean, standard deviation, minimum, maximum, skewness, kurtosis, and trend.

Since some patients had multiple escalation events, and to adjust for confounders, we estimated the standard error and confidence intervals using bootstrap resampling of the logistic regression model (10,000 runs, sampling episode ID with replacement and using the percentile method). We also used the variance inflation factor (VIF) to estimate collinearity between respiratory features. Higher values of VIF (above 10) indicate the existence of multicollinearity, meaning that such variables are highly correlated and should not simultaneously be included in the model.[19] The model was fitted using the glm package in R (version 3.6.0).

We used a multivariable logistic regression model to assess the degree to which various respiratory rate features were associated with respiratory care escalation occurring 24 h later. The criterion for a predictor to be significant was a p-value < 0.05 in the multivariable logistic regression model, uncorrected for multiple comparisons. Performance of the model was measured using traditional measures of diagnostic accuracy, specifically c-statistic from the logistic regression model (area under the receiver operating characteristic curve (ROC-AUC)), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). We assumed a decision threshold > 0.5 from the logistic regression model predicted escalation.

Sample size considerations

There was no a priori sample size estimate for our pilot study. Instead, we enrolled all qualifying and consenting patients over the study period. Given the observed confidence intervals, our actual sample size appears sufficient to evaluate early prediction of respiratory escalation.

Between January 15th, 2021 and May 29th, 2021, we enrolled 166 patients. We excluded 14 patients due to data quality issues (10 because of missing demographic characteristics and 4 because of Circadia recordings failures). We also excluded 27 subjects who were in the process of de-escalating ventilatory support when they joined the study. We finally included 125 subjects in the study (Fig. 1). Thirteen patients exhibited at least one escalation, and each was matched to 3 non-escalated patients. Six of the patients who required care escalations experienced 2–5 escalation episodes. There was thus a total of 28 escalation episodes that were matched to 84 non-escalation episodes (Supplementary Table 2). Demographic and clinical characteristics before and after patients matching are shown in Table 1.

The 30 min window analysis showed that the mean respiratory rate 24 hours prior to an escalation episode was 23 bpm, with a wide range from 7 to 40 bpm, while in the non-escalation patients the mean respiratory rate was 22 bpm but the range was narrower, from 14 to 31 bpm. Escalated patients showed significantly more variability in the respiratory rate than non-escalated patients with higher respiratory rate standard deviation and skewness (Table 2).

Table 2

Respiratory rate features in escalated and non-escalated patients
	Escalated (28 events)			Non-escalation (84 events)			Mean Difference (95% CI)	p-value
Mean, 30 minutes	22.5	±	1.2	22.0	±	0.5	-0.50 (-3.10, 2.11)	0.702
SD, 30 minutes	2.1	±	0.2	1.6	±	0.1	-0.57 (-1.04, -07)	0.022
Skew, 30 minutes	0.5	±	0.2	-0.1	±	0.1	-0.54 (-0.98, -0.086)	0.020
Kurtosis, 30 minutes	1.0	±	0.5	0.8	±	0.3	-0.27 (-1.60, 1.06)	0.689
Mean, 1 day	23.0	±	0.1	21.9	±	0.4	-1.16 (-3.31, 0.99)	0.280
SD, 1 day	2.9	±	0.3	2.6	±	0.1	-0.26 (-0.80, 0.28)	0.341
Skew, 1 day	0.4	±	0.1	0.1	±	0.1	-0.31 (-0.54, -0.08)	0.010
Kurtosis, 1 day	0.0	±	0.3	-0.3	±	0.1	-0.33 (-0.84, 0,18)	0.202
Trend, 1 day	-0.7	±	2.2	-1.9	±	0.5	-1.18 (-5.74, 3.38)	0.600
Mean, 3 days	23.6	±	1.0	22.0	±	0.4	-1.52 (-3.48, 0.45	0.127
SD, 3 days	3.4	±	0.3	2.7	±	0.1	-0.62 (-1.17, -0.07)	0.027
Skew, 3 days	0.3	±	0.1	0.1	±	0.5	-0.21 (-0.41, -0.01)	0.037
Kurtosis, 3 days	-0.1	±	0.2	-0.2	±	0.1	-0.11 (-0.44, 0.23)	0.526
Trend, 3 days	-1.5	±	0.9	-0.6	±	0.2	0.89 (-1.09, 2.88)	0.364
Data expressed as means ± standard errors. All units are in breaths per minute, except for Trend (breaths per minute per day). T-tests were used for comparisons. P < 0.05 was considered a significant difference.

The variance inflation factor for the mean of the respiratory rate over 1 day was 20.4, while all inflation factors were < 5. Therefore, mean respiratory rate over 1 day was removed, and our final model included 13 respiratory features. The probability that the respiratory rate features were associated with an escalation event 24 hours later is shown in Fig. 2. When assessing the 30 min window, no significant association was found between increased mean respiratory rate and risk of escalation, despite higher respiratory rate found prior to escalation events. The most remarkable result was that features describing continuous respiratory rate distribution (in particular standard deviation and skewness) were most strongly associated with higher incidence of an escalation event (Fig. 2).

Our overall model had a moderate accuracy with an AUC 0.81 (95% CI: 0.73, 0.88) and a good specificity 0.93 (95% CI: 0.87, 0.99) being able to correctly detect 93% of patients who did not escalate (Table 3, Fig. 3).

Table 3

Performance of the full logistic regression model assessing the association between all respiratory features and subsequent respiratory care escalation within 24 hours
Metric	Estimate (95% CI)
AUC	\(0.81\) (0.73, 0.88)
Sensitivity	\(0.42\) (0.26, 0.57)
Specificity	\(0.93\)(0.87, 0.99)
PPV	\(0.85\) (0.67, 0.99)
NPV	\(0.62\)(0.52, 0.72)
AUC: Area Under the Curve, PPV: Positive predictive value, NPV: Negative predictive value. All metrics were calculated after a complete fit of the model. The decision threshold was 0.5.

Caring for COVID-19 patients is challenging because they are fragile and their ventilatory status often degrades over time. For example, 13 of 125 (10%) of our ward patients required 28 meaningful enhancements in ventilatory support such as transition to HFNCO or ICU admission. The clinical challenge, of course, is to identify patients at risk as early as practical so that resources and attention can be wisely allocated.

Early detection of respiratory failure and prompt initiation of ventilatory support reduces COVID-19 complications and mortality.[2] Tachypnoea predicts the need of mechanical ventilation[6]and ICU admission[7]in COVID-19 patients. Moreover, tachypnoea with more than 24 breaths per minute is associated with a 5-fold increase in mortality.[2] Accurate and frequent assessments of respiratory rate are thus an important part of COVID-19 care,[20] and presumably also are for other serious progressive respiratory infections.

Vital signs are normally assessed at 4-8-hour intervals on hospital wards, a frequency that has not much increased in the last half-century, although hospitalized patients are now far sicker than previously. Because of their special ventilatory risk, COVID-19 patients seem likely to benefit from more frequent monitoring but may actually get less because isolation precludes easy access. The Circadia Contactless Breathing Monitor we used is an alternative approach that provides a continuous estimate of respiratory rate.

An advantage of continuous monitoring is that in addition to respiratory rate per se, we were able to simultaneously evaluate various measures of respiratory rate variation over time. Features describing the distribution of consecutive respiratory rate measurements (skew and standard deviation in particular) were most strongly associated with escalation episodes. Analysis of these features requires continuous monitoring as they cannot be estimated from intermittent spot measurements alone. Our results therefore suggest that continuous monitoring may allow for improved detection of ventilatory deterioration.

Our model, based on 24-hour pre-escalation periods, had an area under the receiver operating characteristics curve of 0.81 which indicates moderate predictive accuracy and potential clinical utility. The high specificity gives clinicians the possibility of accurately identify patients at low risk of respiratory de-compensation and thus focus efforts on patients most likely to deteriorate and require enhanced ventilatory support.

A limitation of our analysis is that it lacked internal or external validation. Initial models often over-fit data and prove less reliable when applied to out-of-population datasets. The model is nonetheless likely to remain useful, even if somewhat less predictive in other COVID populations. Practical implementation of our model will require software built-into the Circadia system or medical record systems. Our analysis was restricted to patients with proven SARS-CoV-2 infections. Results will likely differ with other respiratory viral infections, and of course other respiratory diseases. Nonetheless the general concept that continuous respiratory rate monitoring is prognostic for ventilatory decompensation over a range of conditions is probably valid — although model specifics and coefficients will presumably differ for each.

In summary, our pilot observational study suggests that continuous respiratory monitoring and derived features predict the need for care escalation 24 in advance. Out-of-sample validation remains necessary. But in the meantime, our results suggest that continuous respiratory monitoring is a valuable increment over intermittent monitoring.

Our study was the initial evaluation of Circadia contactless respiratory monitoring in COVID-19 patient who are at special risk of pulmonary deterioration. The major limitation is that the analysis was largely post hoc and thus needs to be confirmed in an out-of-sample population.

Ethics approval and consent to participate: The study was conducted with approval from the Hospital Clinic of Barcelona IRB (Ethics Committee's approval/ID: HCB 2020/0666). Written informed consent was obtained from participating patients.

Consent for publication: Not applicable
Availability of data and materials: The datasets generated and analysed during the current study are not publicly available due tue institutional and commercial regulatory policies but are available from the corresponding author on reasonable request.
Competing interests: ER, ML, JS, EDR, AR, TR, AL, MA, AS, RB, PC, EM, and DIS have no personal financial interest in this work. SK, TL and OA are Circadia employees.
Funding: This study was supported by an investigator-initiated grant from Circadia Technologies, Ltd (London, United Kingdom).
Contributorship statement
- Eva Rivas: This author helped conceive and design the study, collect analyze and interpret the data, write, and critically revise the manuscript.
- Manuel López-Baamonde: This author helped collect and interpret the data, write, and critically revise the manuscript.
- Josep Sanahuja: This author helped collect the data, write, and critically revise the manuscript.
- Elena Del Rio: This author helped collect the data, write, and critically revise the manuscript.
- Tomeu Ramis: This author helped collect the data, write, and critically revise the manuscript.
- Anna Recasens: This author helped collect the data, write, and critically revise the manuscript.
- Antonio López: This author helped collect the data, write, and critically revise the manuscript.
- Marilyn Arias: This author helped collect the data, write, and critically revise the manuscript.
- Stylianos Kampakis This author helped analyze and interpret the data, write, and critically revise the manuscript.
- Timo Lauteslager: This author helped conceive and design the study, analyze and interpret the data, write, and critically revise the manuscript.
- Osama Awara: This author helped analyze and interpret the data, write, and critically revise the manuscript.
- Edward J. Mascha: This author helped conceive and design the study, analyze and interpret the data, write, and critically revise the manuscript.
- Alex Soriano: This author helped collecte the data, write, and critically revise the manuscript.
- Joan Ramon Badía: This author helped collecte the data, write, and critically revise the manuscript.
- PedroCastro: This author helped collecte the data, write, and critically revise the manuscript.
- Daniel I. Sessler: This author helped conceive and design the study, analyze and interpret the data, write, and critically revise the manuscript.
Acknowledgement: Not applicable

World Health Organization. WHO Coronavirus Disease Dashboard. 2022. https://covid19.who.int/. Accessed 16 Aug 2022.
Zhou F, Yu T, Du R, Fan G, Liu Y, Liu Z, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. The Lancet. 2020;395:1054–62.
Singer AJ, Morley EJ, Meyers K, Fernandes R, Rowe AL, Viccellio P, et al. Cohort of Four Thousand Four Hundred Four Persons Under Investigation for COVID-19 in a New York Hospital and Predictors of ICU Care and Ventilation. Ann Emerg Med. 2020;76:394.
Liu J, Xie W, Wang Y, Xiong Y, Chen S, Han J, et al. A comparative overview of COVID-19, MERS and SARS: Review article. Int J Surg. 2020;81:1.
Smith MEB, Chiovaro JC, O’neil M, Kansagara D, Quiñones AR, Freeman M, et al. SYSTEMATIC REVIEW Early Warning System Scores for Clinical Deterioration in Hospitalized Patients: A Systematic Review. Ann Am Thorac Soc. 2014;11:1454–65.
Kafan S, Vajargah KT, Sheikhvatan M, Tabrizi G, Salimzadeh A, Montazeri M, et al. Predicting Risk Score for Mechanical Ventilation in Hospitalized Adult Patients Suffering from COVID-19. Anesth Pain Med. 2021;11.
Hao B, Sotudian S, Wang T, Xu T, Hu Y, Gaitanidis A, et al. Early prediction of level-of-care requirements in patients with COVID-19. Elife. 2020;9:1–23.
Khanna AK, Bergese SD, Jungquist CR, Morimatsu H, Uezono S, Lee S, et al. Prediction of Opioid-Induced Respiratory Depression on Inpatient Wards Using Continuous Capnography and Oximetry: An International Prospective, Observational Trial. Anesth Analg. 2020;131:1012–24.
Sun Z, Sessler DI, Dalton JE, Devereaux PJ, Shahinyan A, Naylor AJ, et al. Postoperative Hypoxemia Is Common and Persistent: A Prospective Blinded Observational Study. Anesth Analg. 2015;121:709–15.
Turan A, Chang C, Cohen B, Saasouh W, Essber H, Yang D, et al. Incidence, Severity, and Detection of Blood Pressure Perturbations after Abdominal Surgery. Anesthesiology. 2019;130:550–9.
Latten GHP, Spek M, Muris JWM, Cals JWL, Stassen PM. Accuracy and interobserver-agreement of respiratory rate measurements by healthcare professionals, and its effect on the outcomes of clinical prediction/diagnostic rules. PLoS One. 2019;14.
Lim WS, Carty SM, Macfarlane JT, Anthony RE, Christian J, Dakin KS, et al. Respiratory rate measurement in adults--how reliable is it? Respir Med. 2002;96:31–3.
Gupta S, Gennings C, Wenzel RP. R = 20: bias in the reporting of respiratory rates. Am J Emerg Med. 2008;26:237–9.
Drummond GB, Fischer D, Arvind DK. Current clinical methods of measurement of respiratory rate give imprecise values. ERJ Open Res. 2020;6:00023–2020.
Garrido D, Assioun JJ, Keshishyan A, Sanchez-Gonzalez MA, Goubran B. Respiratory Rate Variability as a Prognostic Factor in Hospitalized Patients Transferred to the Intensive Care Unit. Cureus. 2018. https://doi.org/10.7759/CUREUS.2100.
Churpek MM, Adhikari R, Edelson DP. The value of vital sign trends for detecting clinical deterioration on the wards. Resuscitation. 2016;102:1–5.
Forleo GB, Santini L, Campoli M, Malavasi M, Scaccia A, Menichelli M, et al. Long-term monitoring of respiratory rate in patients with heart failure: the Multiparametric Heart Failure Evaluation in Implantable Cardioverter-Defibrillator Patients (MULTITUDE-HF) study. Journal of Interventional Cardiac Electrophysiology. 2015;43:135–44.
Lauteslager T, Maslik M, Siddiqui F, Marfani S, Leschziner GD, Williams AJ. Validation of a New Contactless and Continuous Respiratory Rate Monitoring Device Based on Ultra-Wideband Radar Technology. Sensors . 2021;21.
Thompson CG, Kim RS, Aloe AM, Becker BJ. Extracting the Variance Inflation Factor and Other Multicollinearity Diagnostics from Typical Regression Results. Basic Appl Soc Psych. 2017;39:81–90.
Santos MD, Roman C, Pimentel MAF, Vollam S, Areia C, Young L, et al. A Real-Time Wearable System for Monitoring Vital Signs of COVID-19 Patients in a Hospital Setting. Front Digit Health. 2021;3:630273.

Competing interest reported. Competing interests: ER, ML, JS, EDR, AR, TR, AL, MA, AS, RB, PC, EM, and DIS have no personal financial interest in this work. SK, TL and OA are Circadia employees.

SupplemmentalTablesfigures40.pdf

Download PDF

Version 1

posted

You are reading this latest preprint version

Early Detection of Deterioration in COVID-19 Patients by Continuous Ward Respiratory Rate Monitoring: A Pilot Prospective Cohort Study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Background

Methods

Subject selection

Protocol

Measurements

Exposure and outcomes

Data analysis

Sample size considerations

Results

Discussion

Strengths and limitations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1