A prospective evaluation of the community assessment tools (CATs) in children and adults with acute lower respiratory tract infections: an external validation study

Armon, Ruby (2017) A prospective evaluation of the community assessment tools (CATs) in children and adults with acute lower respiratory tract infections: an external validation study. PhD thesis, University of Nottingham.

[img] PDF (Thesis - as examined) - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (5MB)


Background: CATs is a hospital admission triage tool for influenza and pneumonia, and comprises of 7 criteria (6 objective and 1 subjective); any one of which is expected to trigger hospital admission in adults and children. As the CATs was only recently developed, its performance has only been assessed in the United Kingdom using data from the 2009 A (H1N1) flu pandemic. An external validation study is thus required to assess the generalisability of the CATs in a new patient population before it is recommended for adoption in practice.

Aim: The primary aim of this thesis was to prospectively evaluate the predictive performance of the Community Assessment Tools across a range of acute lower respiratory tract infections in a new geographical setting.

Five studies were included in this thesis, each addressing a specific objective in order to achieve the primary aim of this research:

• The first study examined the predictive performance of individual CATs criteria with regard to the need for hospital based care and mortality,

• The second study examined the performance of CATs as unweighted and weighted scoring indexes for assessing these outcomes in adults and children with ALRTI,

• The third study compared the performance of the unweighted CATs scoring index and the well-established pneumonia severity score –CRB65 in two separate sensitivity analyses involving cases of suspected ALRTI and pneumonia in adults.

• The fourth study sought to investigate the value of clinical intuition in medical decision-making by comparing the performance of the CATs as an unweighted scoring index with or without the inclusion of its subjective criterion (causing other clinical concern).

• The fifth study was a qualitative exploration of critical care practices in Nigerian hospitals, conducted to provide contextual interpretation of findings related to the performance of the CATs in predicting the need for critical care and mechanical ventilation.

Methods: Data specific to the CATs and CRB65 were recorded at the initial consultation of patients with symptoms suggestive of an acute lower respiratory tract infection. The performance of the tools with regard to predicting the need for hospital admission , acute care interventions (supplemental oxygen, IV antibiotics and IV fluids), or any deaths on admission were assessed using specificity, sensitivity, negative and positive predictive values, Area under Receiver Operating Characteristic Curves (AUROC) with 95% confidence intervals and Hosmer-Lemeshow goodness-of-fit test. The qualitative study was conducted within a pragmatic paradigm and a semi-structured guide was used to for individual interviews with key hospital staff. Thematic analysis was used for analysing interview data.

Results: Data were obtained for 1016 (809 children <16years) consecutive patients from four hospitals in Lagos Nigeria. In the univariate analysis each CATs criterion significantly predicted an outcome which indicated the need for hospital care or mortality, also none of these criteria were redundant in the multivariable analysis. Criterion C representing low oxygen saturation levels of 92% or less and Criterion G –clinicians’ intuition or gut feelings ,were the best predictors of the need for hospital based care and risk of death in both children and adults. The predictive accuracy of the individual criteria was generally low in terms of AUROC values, but this improved when they were combined to create scoring indexes. An unweighted CATs threshold score of three points or more was appropriate for identifying both adults and children who would potentially benefit from hospital based care and therapeutic interventions.

In the comparison between CRB65 and the unweighted CATs score, the latter was seen to have better discrimination and calibration qualities for predicting all outcomes in adults with ALRTI, and although both the unweighted CATs and CRB65 scores demonstrated good predictive ability for the outcome ‘in-hospital mortality’ in adults with pneumonia, the unweighted CATs score outperformed the CRB65 in terms of AUROC values, however, the difference was marginal and statistically insignificant.

This thesis provided evidence to support considering clinical intuition or gut feelings in the assessment of patients with ALRTI, as the predictive performance of the unweighted CATs score was significantly better with the inclusion of subjective criterion G (causing other clinical concern) than when this criterion was omitted. Due to insufficient data it was not possible to assess the performance of the CATs in regard to the outcome ‘need for critical care’. Findings from the qualitative study revealed that ALRTI was not a common indication for ICU admission in these hospitals, also that clinical predictions tools were not generally used in this regard, rather decisions regarding ICU admissions were determined by the ability to pay and bed availability.

Conclusion: the CATs criteria showed geographical generalisability despite differences in case-mix and geographical variations. When combined to create an unweighted scoring system, CATs showed good potential for guiding decisions to admit adults and children with ALRTI to hospital for further care. Although CATs was developed for use during an influenza pandemic, the unweighted CATs score could be a useful tool for guiding admission decisions in adults with interpandemic pneumonia, as it performed reasonably well when compared with a widely validated and established pneumonia severity score.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Myles, Puja
Nguyen-Van-Tam, Jonathan
Keywords: Triage; Patient admission; Hospital admission; Predictive performance; Community Assessment Tools
Subjects: W Medicine and related subjects (NLM Classification) > WX Hospitals and other health facilities
Faculties/Schools: UK Campuses > Faculty of Medicine and Health Sciences > School of Medicine
Item ID: 45134
Depositing User: ARMON, RUBY
Date Deposited: 24 Apr 2018 13:43
Last Modified: 25 Apr 2018 15:53
URI: http://eprints.nottingham.ac.uk/id/eprint/45134

Actions (Archive Staff Only)

Edit View Edit View