Peter Watkinson

A randomised controlled trial of the effect of continuous electronic physiological monitoring on the adverse event rate in high risk medical and surgical patients.

We conducted a randomised controlled trial of mandated five‐channel physiological monitoring vs standard care, in acute medical and surgical wards in a single UK teaching hospital. In all, 402 high‐risk medical and surgical patients were studied. The primary outcome was the proportion of patients experiencing one or more major adverse events, including urgent staff calls, changes to higher care levels, cardiac arrests or death, in 96 h following randomisation. Secondary outcomes were the proportion of patients requiring acute treatment changes, and the 30‐day and hospital mortality. In the 96 h following randomisation, 113 (56%) patients in the monitored arm and 116 (58%) in the control arm (OR 0.94, 95% CI 0.63–1.40, p = 0.76) had a major event. An acute change in treatment was necessary in 107 (53%) monitored patients and 101 (50%) control patients (OR 0.55, 95% CI 0.87–1.29). Thirty‐four (17%) monitored patients and 35 (17%) control patients died within 30 days. Thirteen patients in the control group received full five‐channel monitoring at the request of the ward staff. We conclude that mandated electronic vital signs monitoring in high risk medical and surgical patients has no effect on adverse events or mortality.

Centile-based early warning scores derived from statistical distributions of vital signs

AIM OF STUDY To develop an early warning score (EWS) system based on the statistical properties of the vital signs in at-risk hospitalised patients. MATERIALS AND METHODS A large dataset comprising 64,622 h of vital-sign data, acquired from 863 acutely ill in-hospital patients using bedside monitors, was used to investigate the statistical properties of the four main vital signs. Normalised histograms and cumulative distribution functions were plotted for each of the four variables. A centile-based alerting system was modelled using the aggregated database. RESULTS The means and standard deviations of our populations vital signs are very similar to those published in previous studies. When compared with EWS systems based on a future outcome, the cut-off values in our system are most different for respiratory rate and systolic blood pressure. With four-hourly observations in a 12-h shift, about 1 in 8 at-risk patients would trigger our alerting system during the shift. CONCLUSIONS A centile-based EWS system will identify patients with abnormal vital signs regardless of their eventual outcome and might therefore be more likely to generate an alert when presented with patients with redeemable morbidity or avoidable mortality. We are about to start a stepped-wedge clinical trial gradually introducing an electronic version of our EWS system on the trauma wards in a teaching hospital.

Gaussian Processes for Personalized e-Health Monitoring With Wearable Sensors

Advances in wearable sensing and communications infrastructure have allowed the widespread development of prototype medical devices for patient monitoring. However, such devices have not penetrated into clinical practice, primarily due to a lack of research into “intelligent” analysis methods that are sufficiently robust to support large-scale deployment. Existing systems are typically plagued by large false-alarm rates, and an inability to cope with sensor artifact in a principled manner. This paper has two aims: 1) proposal of a novel, patient-personalized system for analysis and inference in the presence of data uncertainty, typically caused by sensor artifact and data incompleteness; 2) demonstration of the method using a large-scale clinical study in which 200 patients have been monitored using the proposed system. This latter provides much-needed evidence that personalized e-health monitoring is feasible within an actual clinical environment, at scale, and that the method is capable of improving patient outcomes via personalized healthcare.

Predictive Monitoring of Mobile Patients by Combining Clinical Observations With Data From Wearable Sensors

The majority of patients in the hospital are ambulatory and would benefit significantly from predictive and personalized monitoring systems. Such patients are well suited to having their physiological condition monitored using low-power, minimally intrusive wearable sensors. Despite data-collection systems now being manufactured commercially, allowing physiological data to be acquired from mobile patients, little work has been undertaken on the use of the resultant data in a principled manner for robust patient care, including predictive monitoring. Most current devices generate so many false-positive alerts that devices cannot be used for routine clinical practice. This paper explores principled machine learning approaches to interpreting large quantities of continuously acquired, multivariate physiological data, using wearable patient monitors, where the goal is to provide early warning of serious physiological determination, such that a degree of predictive care may be provided. We adopt a one-class support vector machine formulation, proposing a formulation for determining the free parameters of the model using partial area under the ROC curve, a method arising from the unique requirements of performing online analysis with data from patient-worn sensors. There are few clinical evaluations of machine learning techniques in the literature, so we present results from a study at the Oxford University Hospitals NHS Trust devised to investigate the large-scale clinical use of patient-worn sensors for predictive monitoring in a ward with a high incidence of patient mortality. We show that our system can combine routine manual observations made by clinical staff with the continuous data acquired from wearable sensors. Practical considerations and recommendations based on our experiences of this clinical study are discussed, in the context of a framework for personalized monitoring.

Strict glucose control in the critically ill.

May not be such a good thing for all critically ill patients

An assessment of algorithms to estimate respiratory rate from the electrocardiogram and photoplethysmogram.

Abstract Over 100 algorithms have been proposed to estimate respiratory rate (RR) from the electrocardiogram (ECG) and photoplethysmogram (PPG). As they have never been compared systematically it is unclear which algorithm performs the best. Our primary aim was to determine how closely algorithms agreed with a gold standard RR measure when operating under ideal conditions. Secondary aims were: (i) to compare algorithm performance with IP, the clinical standard for continuous respiratory rate measurement in spontaneously breathing patients; (ii) to compare algorithm performance when using ECG and PPG; and (iii) to provide a toolbox of algorithms and data to allow future researchers to conduct reproducible comparisons of algorithms. Algorithms were divided into three stages: extraction of respiratory signals, estimation of RR, and fusion of estimates. Several interchangeable techniques were implemented for each stage. Algorithms were assembled using all possible combinations of techniques, many of which were novel. After verification on simulated data, algorithms were tested on data from healthy participants. RRs derived from ECG, PPG and IP were compared to reference RRs obtained using a nasal-oral pressure sensor using the limits of agreement (LOA) technique. 314 algorithms were assessed. Of these, 270 could operate on either ECG or PPG, and 44 on only ECG. The best algorithm had 95% LOAs of  −4.7 to 4.7 bpm and a bias of 0.0 bpm when using the ECG, and  −5.1 to 7.2 bpm and 1.0 bpm when using PPG. IP had 95% LOAs of  −5.6 to 5.2 bpm and a bias of  −0.2 bpm. Four algorithms operating on ECG performed better than IP. All high-performing algorithms consisted of novel combinations of time domain RR estimation and modulation fusion techniques. Algorithms performed better when using ECG than PPG. The toolbox of algorithms and data used in this study are publicly available.

Probabilistic Novelty Detection With Support Vector Machines

Novelty detection, or one-class classification, is of particular use in the analysis of high-integrity systems, in which examples of failure are rare in comparison with the number of examples of stable behaviour, such that a conventional multi-class classification approach cannot be taken. Support Vector Machines (SVMs) are a popular means of performing novelty detection, and it is conventional practice to use a train-validate-test approach, often involving cross-validation, to train the one-class SVM, and then select appropriate values for its parameters. An alternative method, used with multi-class SVMs, is to calibrate the SVM output into conditional class probabilities. A probabilistic approach offers many advantages over the conventional method, including the facility to select automatically a probabilistic novelty threshold. The contributions of this paper are (i) the development of a probabilistic calibration technique for one-class SVMs, such that on-line novelty detection may be performed in a probabilistic manner; and (ii) the demonstration of the advantages of the proposed method (in comparison to the conventional one-class SVM methodology) using case studies, in which one-class probabilistic SVMs are used to perform condition monitoring of a high-integrity industrial combustion plant, and in detecting deterioration in patient physiological condition during patient vital-sign monitoring.

Modelling physiological deterioration in post-operative patient vital-sign data

Patients who undergo upper-gastrointestinal surgery have a high incidence of post-operative complications, often requiring admission to the intensive care unit several days after surgery. A dataset comprising observational vital-sign data from 171 post-operative patients taking part in a two-phase clinical trial at the Oxford Cancer Centre, was used to explore the trajectory of patients’ vital-sign changes during their stay in the post-operative ward using both univariate and multivariate analyses. A model of normality based vital-sign data from patients who had a “normal” recovery was constructed using a kernel density estimate, and tested with “abnormal” data from patients who deteriorated sufficiently to be re-admitted to the intensive care unit. The vital-sign distributions from “normal” patients were found to vary over time from admission to the post-operative ward to their discharge home, but no significant changes in their distributions were observed from halfway through their stay on the ward to the time of discharge. The model of normality identified patient deterioration when tested with unseen “abnormal” data, suggesting that such techniques may be used to provide early warning of adverse physiological events.

The effects of precision, haematocrit, pH and oxygen tension on point-of-care glucose measurement in critically ill patients: a prospective study.

Background Critical care glycaemic control protocols commonly have treatment adjustment (target) ranges spanning ≤2 mmol/L. These require precise point-of-care glucose measurement, unaffected by other variables, to avoid measurement errors increasing glycaemic variability and hypoglycaemic episodes (both strongly associated with mortality in critically ill patients). Methods A prospective 206 intensive care patient study was carried out. Arterial glucose concentrations were measured in duplicate using three point-of-care instruments (MediSense Precision PCχ, HemoCue DM and Radiometer 700), a central laboratory instrument (Siemens ADVIA), and in whole blood and plasma using the Yellow Springs Instruments 2300 instrument. Results Coefficients of variation for the MediSense, HemoCue, Radiometer and Siemens instruments were 5.1%, 2.5%, 2.1% and 2.3%, respectively. Compared with the Siemens instrument, the bias (95% limits of agreement) for the MediSense, HemoCue and Radiometer instruments were 0.0 (−1.4 to 1.4), 0.0 (−1.2 to 1.1) and −0.2 (−0.9 to 0.6) mmol/L, respectively. The whole blood–plasma glucose concentration difference was significantly affected by the haematocrit. MediSense and HemoCue instrument performances were substantially affected by haematocrit. MediSense instrument performance was also affected by pH and PaO2. Radiometer instrument performance was not affected by haematocrit, pH or PaO2. Conclusions The MediSense instrument was too imprecise for use in critically ill patients. The haematocrit range seen is too great to allow fixed-factor conversion between whole blood and plasma values, substantially affecting the accuracy of both glucose meters. However, the Radiometer instrument was unaffected by the haematocrit, pH or pO2, resulting in a performance equivalent to the laboratory method. Instrument performance differences may therefore partially explain the differing results of tight glycaemic control therapy trials.

Mortality risks associated with emergency admissions during weekends and public holidays: an analysis of electronic health records

Summary Background Weekend hospital admission is associated with increased mortality, but the contributions of varying illness severity and admission time to this weekend effect remain unexplored. Methods We analysed unselected emergency admissions to four Oxford University National Health Service hospitals in the UK from Jan 1, 2006, to Dec 31, 2014. The primary outcome was death within 30 days of admission (in or out of hospital), analysed using Cox models measuring time from admission. The primary exposure was day of the week of admission. We adjusted for multiple confounders including demographics, comorbidities, and admission characteristics, incorporating non-linearity and interactions. Models then considered the effect of adjusting for 15 common haematology and biochemistry test results or proxies for hospital workload. Findings 257 596 individuals underwent 503 938 emergency admissions. 18 313 (4·7%) patients admitted as weekday energency admissions and 6070 (5·1%) patients admitted as weekend emergency admissions died within 30 days (p<0·0001). 9347 individuals underwent 9707 emergency admissions on public holidays. 559 (5·8%) died within 30 days (p<0·0001 vs weekday). 15 routine haematology and biochemistry test results were highly prognostic for mortality. In 271 465 (53·9%) admissions with complete data, adjustment for test results explained 33% (95% CI 21 to 70) of the excess mortality associated with emergency admission on Saturdays compared with Wednesdays, 52% (lower 95% CI 34) on Sundays, and 87% (lower 95% CI 45) on public holidays after adjustment for standard patient characteristics. Excess mortality was predominantly restricted to admissions between 1100 h and 1500 h (pinteraction=0·04). No hospital workload measure was independently associated with mortality (all p values >0·06). Interpretation Adjustment for routine test results substantially reduced excess mortality associated with emergency admission at weekends and public holidays. Adjustment for patient-level factors not available in our study might further reduce the residual excess mortality, particularly as this clustered around midday at weekends. Hospital workload was not associated with mortality. Together, these findings suggest that the weekend effect arises from patient-level differences at admission rather than reduced hospital staffing or services. Funding NIHR Oxford Biomedical Research Centre.