Biman Chakraborty

Utility of respiratory ward-based NIV in acidotic hypercapnic respiratory failure

We sought to elicit predictors of in‐hospital mortality for first and subsequent admissions with acidotic hypercapnic respiratory failure (AHRF) in a cohort of chronic obstructive pulmonary disease patients who have undergone ward‐based non‐invasive ventilation (NIV), and identify features associated with long‐term survival.

On Some Nonparametric Classifiers Based on Distribution Functions of Multivariate Ranks

Over the last two decades, multivariate sign and rank based methods have become popular in analysing multivariate data. In this paper, we propose a classification methodology based on the distribution of multivariate rank functions. The proposed method is fully nonparametric in nature. Initially, we consider a theoretical version of the classifier for K populations and show that it is equivalent to the Bayes rule for spherically symmetric distributions with a location shift. Then we present the empirical version of that and show that the apparent misclassification rate of the empirical version of the classifier converges asymptotically to the Bayes risk. We also present an affine invariant version of the classifier and its optimality for elliptically symmetric distributions. We illustrate the performance in comparison with some other depth based classifiers using simulated and real data sets.

Determining the Number of Clusters Using Multivariate Ranks

Determining number of clusters in a multivariate data has become one of the most important issues in very diversified areas of scientific disciplines. The forward search algorithm is a graphical approach that helps us in this task. The traditional forward search approach based on Mahalanobis distances has been introduced by Hadi (1992), Atkinson (1994), while Atkinson et al. (2004) used it as a clustering method. But like many other Mahalanobis distance-based methods, it cannot be correctly applied to asymmetric distributions and more generally, to distributions which depart from the elliptical symmetry assumption. We propose a new forward search methodology based on spatial ranks, where clusters are grown with one data point at a time sequentially, using spatial ranks with respect to the points already in the subsample. The algorithm starts from a randomly chosen initial subsample. We illustrate with simulated data that the proposed algorithm is robust to the choice of initial subsample and it performs well in different mixture multivariate distributions. We also propose a modified algorithm based on the volume of central rank regions. Our numerical examples show that it produces the best results under elliptic symmetry.

P221 Temporal Trends in Severity and In-Hospital Mortality in Acute Hypercapnic Respiratory Failure (AHRF) at a Respiratory Ward-Based Non-Invasive Ventilation (NIV) Unit

Introduction Use of NIV for AHRF in COPD, obesity related morbidity, chest wall and neuro muscular conditions have increased significantly over the past decade – there has been a documented 462%increase in acute NIV use in COPD (Chandra D et al. AJRCCM 2011) over 11 years in the United States with similar changes noted in smaller surveys in the UK. In the UK, this has led to the movement of NIV service provision out of critical care at ward-based NIV units. We felt it necessary to analyse temporal trends in the severity and outcomes of ward-based NIV practises. Methods Comparison of the in-house NIV registry data 01/08/2004 –31/01/2006 (Period 1) with 01/01/2011 – 30/06/2012(Period 2) at an 11-bedded ward-based NIV unit within a 1000-bedded hospital Trust in central England, looking at mortality, length(duration) of NIV and initial arterial blood pH, the latter being widely accepted as a marker of AHRF severity. Results There were 281 episodes of AHRF treated in Period 1 and 240 in Period 2 with similar distribution of gender (non-significant increase in the number of women); acute exacerbations of COPD constituted similar proportion (about 70%)of dominant diagnosis behind AHRF in both periods (associated risk factor documentation, e.g. kyphoscoliosis not analysed); the initial arterial blood pH was significantly lower (median initial pH 7.280 vs 7.261; Wilcoxonrank sum test: p=0.03134; pH significantly lower in Period 2); the mean length (duration) of NIV was significantly higher (median length of NIV 4.0 days vs 6.0 days; Wilcoxon rank sum test: p=0.0000018; Length of NIV is significantly higher in period 2), whilst in-hospital mortality was similar (21.6% vs. 22.7%). Discussion Our data confirm the clinical surmise that over time, our ward-based NIV unit is treating more severely ill patients with AHRF who are spending longer periods under acute NIV with no significant change in mortality. Further analysis of population characteristics, co-morbid risk factors for respiratory failure and Domiciliary NIV/Home Mechanical Ventilation practises as well as national trends in NIV use are needed to inform health policy/strategies to deal with long term respiratory conditions. Abstract P221 Figure 1 Wilcoxon rank sum test: p=0.03134; pH signifi cantly lower in Period 2 Wilcoxon rank sum test: p=0.0000018; Length of NIV is signifi cantly higher in period 2

On some classifiers based on multivariate ranks

ABSTRACT Non parametric approaches to classification have gained significant attention in the last two decades. In this paper, we propose a classification methodology based on the multivariate rank functions and show that it is a Bayes rule for spherically symmetric distributions with a location shift. We show that a rank-based classifier is equivalent to optimal Bayes rule under suitable conditions. We also present an affine invariant version of the classifier. To accommodate different covariance structures, we construct a classifier based on the central rank region. Asymptotic properties of these classification methods are studied. We illustrate the performance of our proposed methods in comparison to some other depth-based classifiers using simulated and real data sets.

P125 The effect of preventative hydrocolloid nasal dressings in acute non invasive ventilation (niv)-related nasal bridge pressure ulceration

Introduction There are over 4000 acute mask application episodes coded in the treatment of acute respiratory failure in the UK every month according to a 2017 survey (NCEPOD). Most guidelines on acute NIV use suggest good skin care strategies including regular mask pressure relief. However, data on the magnitude of the problem of nasal bridge pressure ulceration and the effect of proactive preventative steps (e.g., hydrocolloid dressings) remains scant. A previous smaller but similar survey in a district general hospital showed a trend in the reduction of Grade2 Pressure ulcer rates following change in practice but fell short of statistical significance (Stygall G, Morley K, Pickup L, et al. Thorax 2016. 71:3. A124–125.). We set out on a quality improvement project and systematically examined the effect of a proactive approach to prevent Grade2 Pressure ulcers in a dedicated ward-based Physiotherapy-led acute NIV service in a teaching hospital serving a population of about 4 00 000. Methods In addition to the routine acute NIV data for the unit, additional data was collected from 30/10/14 to 31/08/2015 on: NIV mask used (model and size), total number of admissions with days of NIV (NIV bed-days) and nasal bridge tissue viability grading. This included a 12 month period before (period1) and a 12 month period after (period2) the introduction of the proactive prevention approach. A pressure ulcer was defined as Grade2 or above. Pearson’s chi-squared test for comparison between groups and Fisher’s exact test were applied to assess significance. Results [See Table] In period1, there were 161 admissions and 9 Grade2 pressure ulcers from 666 NIV bed-days (ulceration rate=9/666); in period2 there were 134 admissions and 0 pressure ulcers from 718 NIV bed-days (ulceration rate=0/718). There was a statistically significant reduction in Grade2 Pressure ulceration rates (Pearson’s chi-square statistic=7.786; p-value=0.0013 in period2 compared to period1). Conclusions Application of an early prophylactic pressure-relieving hydrocolloid nasal dressing reduces the chance of developing Grade2 pressure ulcers in patients using NIV acutely. Further longitudinal studies including data on a preventative approach towards NIV-related nasal bridge pressure ulceration are needed to confirm the utility of this approach. Abstract P125 Table 1 30/10/14 – 29/10/1512mth period – BEFORE preventative strategy introduced(PERIOD1) Commencement date of application of proactive preventative hydrocolloid nasal dressings in acute NIV set ups: 30/10/15 30/10/15-29/10/1612mth period – AFTER preventative strategy introduced(PERIOD2) NIV Admissions 161 134 Total NIV duration (NIV bed-days) 666 718 Grade2 Nasal bridge pressure ulcers 9 0 The Pearson’s chi-square statistic is 7.786 with a p-value of 0.005. Therefore there were significantly fewer Grade2 Nasal bridge pressure ulcers for Period 2. Since the number of Grade2 Nasal bridge pressure ulcers is less than 5 in one of the cells, a Fisher’s exact test was performed, which yields a p-value of 0.0013 indicating highly significantly fewer Grade2 Nasal bridge pressure ulcers for the Period2.

P73 Acute Non Invasive Ventilation (NIV)–related nasal bridge pressure ulceration: effect of a proactive prevention approach

Introduction NIV applied via full face masks is increasingly becoming the standard of care in the treatment of acute hypercapnic respiratory failure. Most guidelines suggest good skin care: strategies include regular pressure relief, use of masks with softer cushions/pressure-avoidance masks and application of pressure-relieving dressing to the skin to redistribute pressure and reduce friction. We set out to examine the effect of a systematic proactive prevention approach to prevent Grade 2 Pressure ulcers in a Ward Based Physiotherapy-led acute NIV service in a general hospital serving a population of about 400000. This included (a) prophylactic protective dressing and (b) reactive change to a pressure-avoidance mask for identified Grade 1 pressure sore. Methods Data was collected from 01/05/2014 and 31/08/2015 which included an 8-month period before (period1) and an 8-month period after (period2) introduction of the proactive prevention approach. Five main sets of data were collected; the NIV mask used (model and size), whether the mask was changed, the total number of days using NIV, pressure ulcer grading and the outcome of the NIV admission. A pressure ulcer was defined as Grade 2 or above. Results Grade 2 Pressure ulcer rates showed a trend in reduction by over 50% following the change in practice (but fell short of statistical significance: chi-squared test p-value = 0.3). In period1 there were 11 Grade 2 pressure ulcers from 109 admissions; in period2 there were 5 pressure ulcers from 105 admissions. Benefits of using total face masks for NIV delivery were also noted with those patients who were poorly complaint with the standard NIV full face mask to prevent treatment failure. Conclusions An early prophylactic pressure-relieving dressing and a reactive change to a pressure-avoidance mask for identified Grade 1 pressure sore, can reduce the chance of developing Grade 2 pressure ulcers for patients using NIV acutely. Further studies including longitudinal data on a proactive prevention approach adjusted for acute NIV duration for NIV-related nasal bridge pressure ulceration are needed to confirm the utility of this approach.

P118 Factors affecting concordance with continuous positive airway pressure (CPAP) in obstructive sleep apnoea syndrome (OSAS)

Introduction and objectives The benefits of continuous positive airways pressure in the treatment of obstructive sleep apnoea syndrome have been well established. Despite this, CPAP adherence remains a significant issue resulting in many patients not receiving adequate treatment. A number of variables have been suggested as contributing to non-concordance, however study results have been inconsistent. Studies assessing long term concordance, suggest severity of OSAS and sleepiness to be good predictors of this. This scientific survey looked at the influence of co-morbidity and the severity of OSAS as represented by apnoea hypopnoea index (AHI) at diagnosis on the usage and concordance with CPAP. Methods Data from 230 patients completing annual follow up after initiation of CPAP by 31st December 2014 was collected retrospectively. The presence and severity of co-morbidity was assessed by the Adult Co-morbidity Evaluation- 27 (ACE-27) score. CPAP usage per day was averaged over the preceding year. The association between usage and initial AHI (data available for 207 patients) was analysed by linear regression. The association between usage and ACE-27 score was analysed by ANOVA. Results The regression coefficient for initial AHI against CPAP usage shows a statistically significant effect ([p = 0.00126] fitted equation: concordance = 4.161 + 0.024 × AHI). There was no significant difference in CPAP usage between different ACE-27 groups. Further analysis of individual co-morbidities revealed significance in four categories; cardiac arrhythmia (p = 0.031), coronary artery disease (p = 0.006), congestive heart failure (p = 0.045) and malignancy (p = 0.001). Conclusion AHI at diagnosis remains a strong determinant of CPAP concordance at 1 year. Severity of co-morbidity cannot be conclusively demonstrated to influence usage however further studies into overall and specific co-morbidities are warranted.

M21 Comparison of the effect of a ventilation multidisciplinary meeting on utilisation of critical care resources

Introduction and objectives Optimal utilisation of critical care resources requires timely discharge of patients from critical care to appropriate wards. This represents a challenging and high risk transition. Local audits revealed that a few multimorbid patients with difficult respiratory weans accounted for 30% of critical care bed days. A weekly ventilation multidisciplinary team (VMDT) meeting combining respiratory and critical care expertise was established at a 692-bed hospital to improve management and resource use for this patient group. The effect was compared to a 2nd hospital within the same trust without VMDT. Method A retrospective comparison of 6 month periods before (period 1: 1/10/07–31/3/08) and after (period 2: 1/10/12–31/3/13) introducing VMDT was carried out using data collected for Intensive Care National Audit and Research Centre. The same data was collected for a sister hospital, belonging to the same trust, without VMDT. The numbers of discharges to a respiratory ward with non- invasive ventilation (NIV) facilities were compared with Chi-Square test. The numbers of level 1 critical care bed days were compared with T test. Results In period 1, hospital 1 discharged 458 patients from critical care and hospital 2 discharged 456. In period 2 these figures were 494 (p = 0.30) and 495 (p = 0.84) respectively. There was no change to background parameters. The number of discharges to respiratory ward with NIV facilities increased significantly in hospital 1 (36 to 65, p = 0,011) after VMDT. Whilst the number of patients discharged to respiratory ward increased in hospital 2 this was not significant (9 to 19, p = 0.13). The number of level 1 bed days fell significantly (208 to 18, p < 0.0000000001) in hospital 1. Hospital 2 saw an increase in level 1 days over the same period. Conclusion Introduction of VMDT increased the proportion of respiratory patients discharged to a respiratory ward from critical care and reduced level one bed days in hospital 1 by expediting the discharge of complex respiratory wean patients thereby increasing patient flow and liberating critical care resources. The same reduction was not observed in the hospital 2 suggesting this effect was not due to trust wide changes in critical care practice.

P42 Factors affecting the duration of acute non invasive ventilation required in patients with acute hypercapnic respiratory failure

Introduction and objectives Factors predicting the likelihood of failure of NIV, i.e. requirement of intubation or death, have been well documented with low pH shown to be the most important factor. Factors affecting the duration of NIV required in those patients who receive ward base treatment without the need for intubation have not been established. This study aimed to identify factors which influence the duration of NIV required in acute hypercapnic respiratory failure. Methods A retrospective analysis of 123 consecutive episodes of acute hypercapnic respiratory failure requiring NIV between June 2013 and June 2014 was carried out. Correlation between duration of NIV treatment and a number of variables, namely admission creatinine, pH, worst arterial carbon dioxide level (CO2) and presence and severity of chronic kidney disease (CKD) was assessed by simple linear regression. Results There was a statistically significant regression coefficient between worst observed CO2 and the duration of NIV (fitted equation: NIV Duration = 4.281 + 14.357 × Worst CO2, p = 0.019). The plotted linear relationship showed an increase in duration of NIV treatment of 14.35 h for every 1 kPa increase in CO2 above 6 kPa. The admission creatinine and severity of CKD did not significantly alter the duration of NIV required. The presence of acute kidney injury was also not significant. The pH value did not significantly alter the duration of NIV treatment.Abstract P42 Figure 1 NIV duration plotted against worst CO2. Linear regression fitted equation shown (duration NIV (hours) = 4.281 + 14.357 x Worst CO2) Conclusion This survey shows that the level of CO2 influences the duration of acute NIV required, in that for every 1 kPa rise in CO2, the duration of acute NIV treatment rises by 14.35 h. The other studied variables do not correlate with treatment duration.