Karen Cairns

The Northern Ireland Public Access Defibrillation (NIPAD) study: effectiveness in urban and rural populations

Objective: To assess the impact of mobile automated external defibrillators (AEDs) on out-of-hospital cardiac arrests (OHCAs) in urban and rural populations. Design: Prospective before and after intervention, population study. Setting: Urban and rural areas of 160 000 each. Patients, interventions and main outcome measures: In 2004–6 the demographics of OHCAs were assessed. In 2005–6 AEDs were deployed (29 urban, 53 rural): 335 urban first responders (FRs) and 493 rural FRs were trained in AED use and dispatched to OHCAs. Call-to-response interval (CRI), resuscitation and survival-to-discharge rates for OHCA were compared. Results: In 2004 there were 163 urban OHCAs and the emergency medical services (EMS) attended 158 (ventricular fibrillation (VF) 27/158 (17.1%)). In 2005–6 there were 226 OHCAs, EMS attended 216 (VF 30/216 (13.9%)). In 2005–6 FRs were paged to 128 OHCAs (56.6%), FRs attended 88/128 (68.8%): 18/128 (14.1%) reached before the EMS. The best combined FR/EMS mean (SD) CRI in 2005–6 (5 min 56 s (4)) was better than the EMS alone in 2004 (7 min (3); p = 0.002). Survival rate was 5.1% in 2004, 1.4% in 2005–6 (p = NS). In 2004 there were 131 rural OHCAs, EMS attended 121 (VF 19/121 (15.7%)). In 2005–6 there were 122 OHCAs, EMS attended 114 (VF 19/114 (16.7%)). In 2005–6 FRs were paged to 49 OHCAs, FRs attended 42/49 (85.7%): 23/49 (46.9%) reached before the EMS. The best combined FR/EMS mean (SD) CRI in 2005–6 (9 min 22 s (6)) was better than the EMS alone in 2004 (11 min 2 s (6); p = 0.018). Survival rate was 2.5% in 2004, 3.5% in 2005–6 (p = NS). Conclusions: Despite improvement in CRI there was no impact on survival (witnessed arrest 32.8%, VF 15.6%). Trial registration number: ISRCTN07286796.

The obstacles to maximising the impact of public access defibrillation: an assessment of the dispatch mechanism for out-of-hospital cardiac arrest

Objectives: To determine the diagnostic accuracy of advanced medical priority dispatch system (AMPDS) software used to dispatch public access defibrillation first responders to out-of-hospital cardiac arrests (OHCA). Design: All true OHCA events in North and West Belfast in 2004 were prospectively collated. This was achieved by a comprehensive search of all manually completed Patient Report Forms compiled by paramedics, together with autopsy reports, death certificates and medical records. The dispatch coding of all emergency calls by AMPDS software was also obtained for the same time period and region, and a comparison was made between these two datasets. Setting: A single urban ambulance control centre in Northern Ireland. Population: All 238 individuals with a presumed or actual OHCA in the North and West Belfast Health and Social Services Trust population of 138 591 (2001 Census), as defined by the Utstein Criteria. Main outcome measures: The accurate dispatch of an emergency ambulance to a true OHCA. Results: The sensitivity of the dispatch mechanism for detecting OHCA was 68.9% (115/167, 95% confidence interval (CI) 61.3% to 75.8%). However, the sensitivity for arrests with ventricular fibrillation (VF) was 44.4% (12/27) with sensitivity for witnessed VF of 47.1% (8/17). The positive predictive value was 63.5% (115/181, 95% CI 56.1% to 70.6%). Conclusions: The sensitivity of this dispatch process for cardiac arrest is moderate and will constrain the effectiveness of Public Access Defibrillation (PAD) schemes which utilise it. Trial registration: controlled-trials.com ISRCTN 07286796.

The value of positron emission tomography in patients with non-small cell lung cancer

BACKGROUND Pre-operative assessment of non-small cell lung cancer (NSCLC) is a major application of positron emission tomography (FDG-PET). Despite substantial evidence of diagnostic accuracy, relatively little attention has been paid to its effects on patient outcomes. This paper addresses this by extending an existing decision model to include patient-elicited utilities. PATIENTS AND METHODS A decision-tree model of the effect of FDG-PET on pre-operative staging was converted to a Markov model. Utilities for futile and appropriate thoracotomy were elicited from 75 patients undergoing staging investigation for NSCLC. The decision model was then used to estimate the expected value of perfect information (EVPI) associated with three sources of uncertainty-the accuracy of PET, the accuracy of CT and the patient related utility of a futile thoracotomy. RESULTS The model confirmed the apparent cost-effectiveness of FDG-PET and indicated that the EVPI associated with the utility of futile thoracotomy considerably exceeds that associated with measures of accuracy. CONCLUSION The study highlights the importance of patient related utilities in assessing the cost-effectiveness of diagnostic technologies. In the specific case of PET for pre-operative staging of NSCLC, future research effort should focus on such elicitation, rather than further refinement of accuracy estimates.

A Monte Carlo Simulation Model to Assess Volunteer Response Times in a Public Access Defibrillation Scheme in Northern Ireland

This paper describes the development of a model to assess the distribution of response times for mobile volunteers of a public access defibrillation (PAD) scheme in Northern Ireland. Using parameters based on a trial period, the model predicts that a PAD volunteer would arrive before the emergency medical services (EMS) to 18.8% of events to which they are paged in a given year period. This is in agreement with what has actually been observed during the trial period (where volunteers have actually reached 15% of events before the EMS), and thus assisting validation of the model. Results from this model illustrate how ongoing volunteer commitment is key to the success of the scheme

A public access defibrillation trial in urban and rural communities in Northern Ireland: developing the roster model

This paper introduces a special computer-based roster scheme developed to allocate and manage volunteers working as part of a public access defibrillation trial. The roster scheme, developed for the urban region, is rooted on population statistics and demographics for that area and utilizes geographical mapping software and spatial modelling techniques to subdivide the geographical location into appropriate paging zones. The central location for zones was constrained to be within a reasonable travelling time for each volunteer. By estimating sudden cardiac arrest occurrences using a Poisson process, the model, together with road network information, selects a roster which minimizes volunteer response time.

Using simulation to assess cardiac first-responder schemes exhibiting stochastic and spatial complexities

A Monte-Carlo simulation-based model has been constructed to assess a public health scheme involving mobile-volunteer cardiac First-Responders. The scheme being assessed aims to improve survival of Sudden-Cardiac-Arrest (SCA) patients, through reducing the time until administration of life-saving defibrillation treatment, with volunteers being paged to respond to possible SCA incidents alongside the Emergency Medical Services. The need for a model, for example, to assess the impact of the scheme in different geographical regions, was apparent upon collection of observational trial data (given it exhibited stochastic and spatial complexities). The simulation-based model developed has been validated and then used to assess the schemes benefits in an alternative rural region (not a part of the original trial). These illustrative results conclude that the scheme may not be the most efficient use of National Health Service resources in this geographical region, thus demonstrating the importance and usefulness of simulation modelling in aiding decision making.

Modelling the development of late onset sepsis and length of stay using discrete conditional survival models with a classification tree component

This paper introduces a discrete conditional survival model (DC-S) with a classification component for predicting patient outcome and survival component for predicting length of stay in hospital. The DC-S model consists of two components; the conditional component which utilises a classification tree and the survival component which models the survival distribution. The survival component of the model is conditioned on the discrete conditional component, the classification tree. The DC-S model with classification tree is applied to a healthcare scenario where the length of stay of babies in neonatal wards in Northern Ireland (United Kingdom) is modelled using the baby characteristics known on the first day of admission. The resulting model can accurately predict length of stay of babies and thus has the potential to be used in bed planning. Hospitals could use such good estimates for the length of stay of patients (determined on the day of arrival) to plan ahead to make the correct provisions available during their stay. Not only does this have resource implications, it can also help patient families. The resulting model can also predict the occurrence (or otherwise) of late onset sepsis, which has implications on a patients stay.

Can More High-Risk Cardiovascular Patients be Identified Using Novel Biomarkers? Extending the msm Package to Measure Reclassification

Numerous novel biomarkers are being considered as tools to enhance cardiovascular risk estimation. Statistical models can indicate how a biomarker influences movements between health states, and clinicians are interested in the impact they may have compared to traditional risk factors alone. Net reclassification indices (NRIs) have recently become popular statistics for measuring the prediction increment of new biomarkers. The msm package within R can fit multi-state models to longitudinal data, giving output for all permitted state-to-state transitions within the one calculation. However its usefulness could be further enhanced through its extension to include NRIs. This paper focuses on extending the msm package to enable NRIs to be calculated. The results are demonstrated for cardiovascular data, with the inclusion of a panel of novel biomarkers (CRP, NT-pro BNP, and troponin I) giving rise to a NRI of 0.2089 for males and 0.1678 for females.

Assessing length of stay and associated characteristics of geriatric patients in Northern Ireland

The effective provision of care for the elderly is becoming increasingly more difficult. This is due to the rising proportion of elderly in the population, increasing demands placed on the health services and the financial strain placed on an already stretched economy. The research presented in this paper uses three different models to represent the length of stay distribution of geriatric patients admitted to one of the six key acute hospitals in Northern Ireland and various patient characteristics associated with their respective length of stay. The accurate modelling of bed usage within wards would enable hospital managers to prepare patient discharge packages and rehabilitation services in advance. The models presented within the paper include a Cox proportional hazards model, a Bayesian network with a discrete variable to represent length of stay and a special conditional phase-type model (C-Ph) with a connecting outcome node. This research demonstrates the new efficient fitting algorithm employed for Coxian phase-type distributions while updating C-Ph models for recent elderly patient data.

The Effects of Anti-Hypertensive Drugs Evaluated Using Markov Modelling for Northern Ireland Chronic Kidney Disease Patients

The aim of this paper is to use Markov modelling to investigate survival for particular types of kidney patients in relation to their exposure to anti-hypertensive treatment drugs. In order to monitor kidney function an intuitive three point assessment is proposed through the collection of blood samples in relation to chronic kidney disease for Northern Ireland patients. A five state Markov model was devised using specific transition probabilities for males and females over all age groups. These transition probabilities were then adjusted appropriately using relative risk scores for the event death for different subgroups of patients. The model was built using TreeAge software package in order to explore the effects of anti-hypertensive drugs on patients.