Objectives: To identify the relationship between 'inappropriate' attendance at an accident and emergency department (AED) by adults registered with local general practices and their use of primary care.

Methods: A case-control study matched for age, sex, distance from the AED, social class and registered general practice and set in a single AED and two health centres in South Essex. The participants were a total of 452 patients over 15 years old from the two health centres classified as having attended the AED 'inappropriately' in 1997 as identified by a modified Sheffield process method, and 452 controls. The predictive variables were measures of utilisation in the year 1997, including number of contacts in primary care, referral and investigation costs. Measures of morbidity were collected as potential confounders. These included a recorded history of anxiety or depression in the year 1997, or being in receipt of repeat prescriptions in that year.

Results: The rate of 'inappropriate' attendance was 16.8% {95% confidence interval (CI): 15.7-18.0}. All measures of utilisation and markers of anxiety and depression were significantly positively associated with 'inappropriate' attendance, but there was no association with markers of chronic morbidity. Only the number of general practitioner (GP) appointments (P < 0.0001) and out-of-hours advice calls (P < 0.0001) were independently correlated with 'inappropriate' attendance in a conditional logistic regression. 'Inappropriate' attendees had approximately twice as many GP appointments and 10 times as many out-of-hours telephone contacts with the GP.

Conclusions: GP-registered, 'inappropriate' attendees at AEDs utilise primary care services more than matched controls; this pattern of service utilisation appears to be unrelated to chronic physical illness. Thus, simply providing new, directly accessible primary health care services may not significantly reduce AED use.