Patient Status Engine for wireless monitoring of vital signs

Study size, design and location

Prospective observational study of 155 adults admitted to a labour and delivery suite in the US at the time of labour (more than 35 weeks gestational age and less than 6 cm cervical dilation). This study measured maternal temperature during labour as a risk factor for early onset sepsis in newborns.

Intervention and comparator

Intervention: Lifetemp sensors. Data was transmitted every minute.

Comparator: Manual intermittent temperature measurements taken every 3 to 6 hours following clinical protocols.

Key outcomes

Over 90% of Lifetemp data was valid in 98 people, and over 75% of data was valid in 127 people. Lifetemp continuous measurements of temperature correlated with manual measurements (r=0.399, p<0.001). Manual temperature measurements missed 32 fevers above 38°C and 13 fevers above 38.5°C that were found by Lifetemp. Review of this data showed these episodes of fever were missed because they happened between manual measurements. Lifetemp missed 7 fevers above 38°C and 5 fevers above 38.5°C that were found by manual measurements. This was likely related to low-grade fevers on the alert threshold. Both Lifetemp and manual measurements detected fever above 38°C in 15 people. Of these, 13 were found earlier by Lifetemp continuous measurements, with 9 detected more than 1 hour earlier compared with manual measurements.

There were no reports of discomfort or adverse events from using Lifetemp.

Strengths and limitations

This study suggests using Lifetemp is feasible in taking continuous temperature measurements during labour. Issues with non-adherence were not reported despite some invalid data capture. It was unclear from the study methods if the healthcare professionals who took the manual temperature measurements were blinded to Lifetemp data. Authors stated the study could not be powered to assess the occurrence or outcome of early onset sepsis in newborns because of the low incidence of this disease. The study only assessed the use of Lifetemp sensors and not the overall PSE system.

Study size, design and location

Prospective observational study of 982 children (birth to 17 years) admitted for elective or emergency care in a specialist paediatric hospital in the UK providing intermittent and continuous monitoring and care.

Intervention and comparator

Intervention: The RAPID Index calculated from the PSE, which used Lifetouch (electrocardiogram [ECG]-derived heart rate and respiratory rate) and Nonin WristOx2 sensors (pulse oximetry and derived pulse rate). Data was transferred every minute.

Comparator: Paediatric Early Warning (PEW) score recorded manually every 1 to 4 hours. The PEW score includes respiratory rate, respiratory distress, pulse oximetry, inspired oxygen, heart rate, systolic blood pressure, and capillary refill time.

Key outcomes

The study showed the feasibility of using the PSE to collect valid clinical data wirelessly for at least 50% of the intended monitoring time. Final data capture as a proportion of intended monitoring time was 93% for Lifetouch and 55% for WristOx2. The final proportion of valid clinical data recorded was 63% (Lifetouch) and 50% (WristOx2).

Findings showed 29 children had 36 significant clinical deteriorations during the study period. The RAPID Index identified the onset of deterioration within 72 hours for 97% (35 of 36) of these events compared with 86% (31 of 36) by the PEW score. The onset of deterioration was detected earlier by the RAPID Index (mean 46.9 hours before significant deterioration) than the PEW score (mean 40.2 hours before significant deterioration).

The RAPID Index demonstrated high sensitivity (97%) and negative predictive value (NPV; 99%) compared with the PEW score (sensitivity 86%, NPV 99%). However, the PEW score had overall better predictive discrimination (specificity 81%, positive predictive value ([PPV] 21%), with the RAPID Index having a high rate of false alarms (specificity 25%, PPV 7%). Clinical review found 14 children who had significant deterioration could have benefited from earlier review. Of these, 4 children had potentially avoidable respiratory and cardiac arrests. Authors suggested these could have been avoided had the PSE and RAPID Index been reviewed.

Strengths and limitations

It is unclear from the study methods whether the nurses who measured clinical observations and PEW scores were blinded to the RAPID Index score. Sensor usability differed between the Lifetouch and WristOx2. The WristOx2 captured and recorded less valid clinical data leading authors to suggest a better and more tolerable device was needed. The RAPID Index was recorded more often than the PEW score. Authors suggested this partly explained the high sensitivity and false alarms of the PSE. Authors concluded that the data was preliminary, and the technology needed more research to establish efficacy and cost effectiveness before it could be adopted.

Study size, design and location

Prospective observational study of 30 adults with acute exacerbations of chronic obstructive pulmonary disease admitted to a single-centre hospital in Denmark.

Intervention and comparator

Intervention: Continuous monitoring with Lifetouch (heart rate and respiratory rate), Nonin WristOx2 3150 (arterial oxygen saturation), and Meditech BlueBP‑05 (intermittent blood pressure). Data was transmitted every minute, except for blood pressure which was recorded every 15 to 30 minutes.

Comparator: Early warning score (EWS) done with routine equipment. Measurements taken every 12 hours with increased monitoring with worsening EWS.

Key outcomes

Continuous monitoring using Lifetouch and Nonin WristOx2 3150 sensors detected more abnormal vital signs than EWS. Continuous monitoring with the Nonin WristOx2 3150 found moderate oxygen desaturation events in 27 people (n=30) compared with 4 with EWS (p<0.001). Only 1 person had a moderate desaturation event detected by EWS that was not found by continuous monitoring. This was due to missing data. Continuous monitoring found severe oxygen desaturation events in 19 people. No severe desaturation events were found by EWS.

Lifetouch sensors found tachycardic events in 15 people compared with 4 reported with EWS (p=0.005). All tachycardic events reported by EWS were also found by the Lifetouch sensors. Lifetouch sensors detected tachypnoeic events in 17 people compared with 7 with EWS (p=0.02). EWS found tachypnoeic events in 3 people that were not detected by Lifetouch. Continuous monitoring with Lifetouch found bradypnoea in 16 people, with no episodes reported by EWS (p<0.001).

Meditech BlueBP-05 found hypotension in 7 people during the study period compared with 2 people by EWS (p=0.15). EWS recorded 3 hypotensive events in the same person, which was not detected by wireless monitoring.

Strengths and limitations

The small sample size of 30 people reflected time limitations and the pilot study design. The study was not powered to find differences in clinical outcomes. Healthcare professionals and patients were blinded to vital signs data from the continuous monitoring devices. Researchers saw patients daily to encourage adherence and to change device batteries as needed. They hoped this would ensure high-quality data. However, the study reported some missing data from continuous monitoring because of technical difficulties (battery power, Bluetooth connectivity issues, and the bedside gateway being off) and non-adherence. Authors cautioned that further research was needed to determine the clinical significance of abnormal physiological values and to identify which vitals and thresholds are most predictive of adverse outcomes.

Study size, design and location

Prospective observational study assessing heart rate variability (HRV) in 119 adults at risk of cirrhosis decompensation. The study included 49 people presenting with acute decompensation admitted to hospital in the UK and 70 people with stable cirrhosis having community care in Germany.

Intervention and comparator

Intervention: Lifetouch system used by inpatients in the UK.

Comparator: Holter ECG recording used by outpatients in Germany.

Key outcomes

Continuous remote monitoring of HRV using Lifetouch was feasible in people with cirrhosis. HRV was measured using standard deviation of beat-to-beat intervals (SDNN). HRV was analysed for all inpatients monitored using Lifetouch (n=49) compared with 89% (62 of 70) of outpatients with Holter ECG monitoring. There were no statistically significant differences in SDNN between Lifetouch and Holter when controlling for disease severity.

The study reported several clinical outcomes related to the use of SDNN in cirrhosis. SDNN was inversely related to validated clinical scores and was the only independent predictor of 90‑day mortality. These patient clinical outcomes were unrelated to the performance of the PSE so are not described further. Authors suggested remote monitoring of SDNN may detect inflammatory activity in people at risk of acute decompensation. This could lead to closer clinical review or early intervention before organ failure.

Strengths and limitations

The study was a multicentre study including a liver and transplantation unit in the UK. It included 2 patient groups with differing disease severity. Only the inpatient group in the UK were offered the Lifetouch system. People were recruited sequentially to this group based on the availability of the technology. The limited number of Lifetouch monitors meant only a small sample of people had repeated measurements beyond baseline SDNN. It is unclear how many people had continuous monitoring with Lifetouch beyond baseline. Authors acknowledged that the study was underpowered for data analysis on long-term SDNN monitoring.

Using Lifetouch for continuous remote monitoring of HRV in the community is not reflected in the findings. The comparison of HRV data captured by Lifetouch compared to Holter is limited by the differences in the healthcare settings where the devices were used. One researcher did all SDNN analyses. They were blind to individual patient data and outcomes and used standard methods in analysing data from both sites. This likely increased the consistency of the analysis but limited interrater reliability.

Study size, design and location

Randomised controlled trial of 195 adults having acute or elective surgery who were expected to be hospitalised more than 24 hours postoperatively at a single-centre site in Norway.

Intervention and comparator

Intervention: PSE in combination with efficacy safety score (ESS, a validated clinical decision-making tool). ESS was done hourly during the first 4 hours after discharge from the post-anaesthesia care unit, and then every 2 hours after.

Comparator: Standard care (paper-based National Early Warning Score [NEWS] at least every 12 hours along with postoperative pain assessment at least every 8 hours).

Key outcomes

Use of the PSE with ESS resulted in earlier postoperative mobilisation (mean 10.1 hours) compared with standard care (mean 14.2 hours, p=0.008; hazard ratio 1.54).

The intervention group received higher opioid doses (mean 25.5 mg) than standard care (mean 15.2 mg, p=0.001), with the former reporting lower average pain intensity (p<0.001) and higher patient satisfaction (p<0.001). Authors suggested that nurses may have felt more comfortable administering higher opioid doses because of their increased attention to and communication with patients. A mean of 6.7 pain evaluations were documented in the intervention group, compared with 1.4 for standard care (p<0.001). Pain was not documented for 17 people in the standard care group. Similarly, more NEWS were recorded for people in the intervention group (mean 8.2) than standard care (mean 3.4, p<0.001). Supplementary oxygen was provided to 26% more people in the intervention group (57 of 96) than standard care (32 of 99; p<0.001).

There were no observed serious complications in the intervention group. There were 2 serious events in the standard care group (1 severe bradycardia and 1 stroke), but the authors cautioned that the study had not been powered to make conclusions about safety issues. Five people in the intervention group were identified as needing additional treatment and follow up: 2 for pain, 2 for hypotension, and 1 for atrial fibrillation. Mean length of hospital stay was similar for both groups (70.9 hours for intervention compared with 76.6 hours for standard care; p=0.58).

Strengths and limitations

Authors reported the study was powered to identify meaningful effects of the intervention on the primary outcome of time to full mobilisation. The study used robust methods to randomise people to the intervention or standard care group, including a random number generator and allocation concealment. Patients and healthcare professionals were not blinded to group allocation at the ward due to the obvious differences between the PSE and standard care.

The care delivered in the standard care group was clearly outlined and followed the hospital's clinical guidelines. The intervention group was assessed significantly more often than people in standard care. As such, it is not clear whether the findings reported are wholly because of the tools used (PSE and ESS) or the increased attention from nurses. The PSE was assessed in combination with ESS. The effects of the PSE alone can therefore not be inferred from these findings.

Study size, design and location

Prospective observational study investigating the feasibility of wireless monitoring to assess autonomic function in 10 newborns during parent holding episodes in a single centre in the UK.

Intervention

Intervention: ECG recorded using the PSE. HRV was calculated before, during, and after parent holding using Kubios HRV software.

Key outcomes

The study showed that wireless ECG monitoring is feasible in newborns and may make parent holding easier because of the lack of wired connections. A total of 140 parent holding episodes were analysed. The PSE was acceptable to parents and healthcare professionals. Data was not analysed in 20 cases because of sensor detachment during holding. There was a significant reduction in newborns' heartrate when they were held by their parents (p<0.001). Parent holding was found to be associated with autonomic stability in newborns.

Strengths and limitations

The study is limited by its small sample but benefits from its repeated measures design to analyse HRV of each newborn before, during, and after parent holding. The study was reported in an abstract which limits details on research methods and quality. The medical conditions of the newborns were unclear as was the length of the parent holding episodes and monitoring periods.