PARTICIPANTS, METHODS, AND RESULTS

The LSR prospectively identified stroke patients admitted to the medical unit of our hospital (excluding those with subarachnoid haemorrhage), and all patients were assessed by a stroke physician. We extracted time of symptom onset and of hospital admission; home postcode; computed tomography (CT) findings; and whether symptoms were first apparent on waking from sleep. Carstairs DepCat scores⁵ were determined from the home postcode and, to maximise statistical power, patients were grouped into approximate quartiles of deprivation (DepCats 1 and 2; 3; 4; and 5–7). Prespecified analyses were the proportion of patients reaching hospital within three hours of symptom onset by deprivation quartile, CT findings, and whether symptoms were present on waking. During the period of this study there were considerable developments in the provision of in-hospital stroke services, and for this reason analysis of the proportion of patients from each deprivation quartile admitted to a stroke unit were though likely to be unreliable.

Data were obtained for Lothian for mean general practitioner list size (1999; Lothian Health Board, personal communication), for mean ambulance response times (2000; Scottish Ambulance Service, personal communication), and for mean distance to hospital (derived from the distance “as the crow flies” from the centre of each postcode sector to the centre of the postcode sector in which the hospital lies, using online mapping tools (http://www.streetmap.co.uk)). The Information and Statistics Department of the Scottish Health Service provided data for hospital admissions and for death by deprivation category for patients with a main diagnosis of stroke (ICD 9 codes (to March 1996) 430–438; ICD 10 codes (from April 1996) I60–I69, G450, G451, G458, G459) for the years 1995 to 1999, and from these, age and sex adjusted rates of admission and death were calculated. This analysis was restricted to persons under 80 years of age because of concerns regarding the accuracy of death certification data among the very elderly.

Of 1927 patients, data on time to hospital and home postcode were available for 1416; patients with incomplete data were evenly distributed between deprivation quartiles (table 1). The number of patients reaching hospital within three hours fell from 105/407 (25.8%) in the most affluent quartile to 57/317 (17.9%) in the most deprived quartile (p = 0.037, χ² on ranks, table 1), most of the observed difference lying between the most affluent quartile and the rest. Among 1316 patients with CT data available, 164 (12.5%) had haemorrhagic strokes, but there was no difference in the proportion reaching hospital within three hours. Of 1374 patients with data available, 428 (31.1%) had their symptoms present on waking, and fewer of these reached hospital within three hours (16.4% v 23.7%, p = 0.002, χ²); 10.5% of such patients from the most deprived quartile arrived at hospital within three hours, compared with 25.4% of those from the most affluent quartile (p = 0.002, χ² on ranks).

There were no significant differences between deprivation quartiles for GP list size, distance to hospital, or ambulance response times (table 1). There were higher rates of hospital admission in patients from more deprived areas, and this reflected higher rates of death from stroke in persons resident in such areas (table 1).

COMMENT

Patients from affluent areas get to hospital sooner following stroke. This does not reflect differences in the distance to hospital, ambulance response times, or GP list size. Public recognition of stroke symptoms and understanding of the importance of early hospital admission may be greater in those living in affluent areas. Efforts to reduce time to hospital following stroke should include specific attention to public education targeted at those living in deprived areas.