The economic cost of stroke-associated pneumonia in a UK setting

Abstract

Introduction

Stroke-associated pneumonia (SAP) is common, however, data on the economic impact of SAP are scarce. This study aimed to prospectively evaluate the impact of SAP on acute stroke care costs in a UK setting.

Methods

Prospective cohort study of 213 consecutive patients with stroke (196 ischemic, 17 hemorrhagic) was admitted to a UK hospital over 1 year. Socio demographic and clinical characteristics were recorded along with all treatments and rehabilitation activity. Patients were classified as having SAP if they fulfilled criteria for “probable” or “definite” respiratory tract infection according to the Centres for Disease Control and Prevention definition, within the first seven days following stroke. Resource use was calculated using a “bottom up” approach of cumulative unit costs. Univariate and multivariate regression analyses were used to establish independent predictors of direct costs.

Results

Probable or definite SAP occurred in 13.2% (28/213) of patients. Patients with SAP experienced greater inpatient stays (31 days vs. 9 days, p ≤ 0.001) and higher in-hospital mortality (29.2% vs. 10.2%, p = 0.007). Mean (SD) acute care costs per patient was £7035 (6767), but costs were significantly greater for patients with SAP than without [£14,371 (9484) versus £6,103 (5,735); p ≤ 0.001]. SAP was an independent predictor of costs along with increasing stroke severity (NIHSS) and age. Occurrence of SAP resulted in an adjusted incremental additional cost of £5817 (95% CI 4945–6689; p = 0.001) per patient.

Conclusions

SAP increased acute care costs for stroke by approximately 80%. This provides further impetus for research aimed at reducing SAP, and will inform cost-effectiveness analyses of potential therapeutic strategies.

Keywords:

• Stroke
• pneumonia
• cost
• The UK

Acknowledgments

We are grateful to Caroline Leggott for her assistance with costings for care interventions at the hospital, to Jeanette Harris for assistance with data collection, and to Karen Kilner for assistance with statistical analysis.