Accessibility and health service utilization for asthma

Vol. 20, No. 3, pp. 312-317 Printed in Great Britain

Journal of Public Health Medicine

Accessibility and health service utilization for asthma in Norfolk, England A. P. Jones, G. Bentham, B. D. W. Harrison, D. Jarvis, R. M. Badminton and N. J. Wareham

Abstract

Methods The study involved analysis by logistic regression of questionnaire responses from a 10 per cent population sample of 9764 adults aged between 20 and 44 years, and resident within the catchment area of Norwich Health Authority. Results Utilization behaviour was associated with the smoking status of respondents, and levels of car ownership in their ward of residence. After controlling for these factors, respondents reporting asthma were less likely to have ever visited a general practitioner if they lived outside a settlement containing a surgery (odds ratio (OR) 3.07, 95 per cent confidence interval (Cl) 1.11-8.48, p = 0.03), and the likelihood of consultation declined with distance from a surgery (OR for a 1 minute increase in travel time 0.79, 95 per cent Cl 0.66-0.94, p < 0.01). Those living further from an acute hospital unit were also less likely to have consulted a hospital doctor in the previous 12 months (OR for a 1 minute increase in travel time 0.95, 95 per cent Cl 0.9-0.99, p = 0.01). Conclusions Our finding of lower levels of health service utilization amongst some self-reported asthmatics living further from health facilities suggests that the condition of some individuals might be poorly treated, which could increase the risk of fatality. Keywords: asthma management. Health Service accessibility

Introduction Despite improvements in available therapies, asthma mortality has only recently begun to show signs of decline. There is undoubtedly scope for reducing morbidity from the disease, and it has been estimated that up to 86 per cent of deaths may have potentially preventable factors given more timely and appropriate interventions. 3

Studies of asthma deaths suggest that social and geographical isolation may be especially important prognostic factors for appropriate asthma management.3'4'6"9 In the United Kingdom, there has been a trend towards the centralization of health services into larger urban units, particularly since the 1960s.10'11 Service centralization may offer benefits, yet may also present problems to some members of the population.12 Mobility may be especially important in determining consultations with a doctor. Less mobile patients may consult less if they need transport to reach the surgery.12 Rural communities may be more insular and reliant on lay treatment compared with urban populations," so that individuals are more reluctant to seek help in controlling their condition. General practitioners (GPs)

School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ. A. P. Jones, Lecturer G. Bentham, Reader Department of Respiratory Medicine, Norfolk and Norwich Hospital, Norwich NR1 3SR. B. D. W. Harrison, Consultant Physician Department of Public Health Medicine, United Medical and Dental School, St Thomas' Hospital, London SE1 7EH. D. Jarvis, Senior Lecturer Department of Public Health Medicine, East Norfolk Health Commission, St Andrew's House, St Andrew's Park, Norwich NR7 0HT R M Badmintoni

consultant in Public Health Medicine

Department of Community Medicine, Institute of Public Health, Cambridge CB2 2SR. N. J. Wareham, Senior Registrar in Public Health Medicine Address correspondence to Dr Andrew Jones.

© Oxford University Press 1998

Downloaded from jpubhealth.oxfordjournals.org by guest on July 13, 2011

Background Good access to health services may be important for effective asthma management amongst patients. Previously, we found elevated asthma mortality in parts of England with poor access to acute hospitals. A possible explanation is that barriers of access to care may lead to residents of more remote areas having a lower propensity to utilize health services, with the result that their asthma is poorly controlled. Here, we examine the relationship between utilization and the geographical accessibility of health services amongst self-reported asthmatics in a rural county of England.

In the predominantly rural county of Norfolk, England, a confidential enquiry has been established into the circumstances surrounding asthma deaths. 4 A recurrent issue was a poor understanding of the condition amongst sufferers, which caused delays in seeking care when required. Another study in Norfolk investigated asthma mortality and health service utilization rates.5 Populations in the most rural areas generally exhibited high mortality from asthma, yet low hospital admission rates. The study concluded that this inverse relationship may be because of an inequality in access to care.

HEALTH SERVICE UTILIZATION FOR ASTHMA

located far from hospitals may also be less likely to refer their patients for out-patient treatment.13 Recently, we studied the relationship between geographical isolation from large acute hospitals and mortality from asthma in local authority districts in England and Wales.14 After controlling for socio-economic factors we found there was a tendency for mortality to rise with increasing distance to hospital, suggesting that inaccessibility of hospital services may be a risk factor for asthma mortality. A possible reason for this is that barriers of access to care may mean that residents of more remote areas might have a lower propensity to utilize health services, which leads to their asthma being less well controlled. Using the results of a questionnaire survey of respiratory morbidity and health service utilization in Norfolk, the relationship between accessibility and health service utilization in a sample of young adults is examined here.

Subjects and methods

The sampling frame was the Norfolk Family Health Services Authority (FHSA) patient register, a record of all Norfolk inhabitants registered with a doctor. The central urban area of Norwich was excluded, as this had been surveyed previously.15 The sampling strategy was the same as used elsewhere.15 Our analysis concentrated on associations between indicators of health service accessibility and responses to questions regarding the use of health services. Particular attention was paid to the relationship between accessibility and whether respondents had consulted their doctor for certain respiratory problems, either in the 12 months before the questionnaire or ever, as well as if they had consulted a hospital doctor in the previous 12 months, and if they had ever visited a hospital casualty department for breathing problems. The survey contained no information on the accessibility of care from where respondents lived, or on their socio-economic status. Consequently, a key factor was the linkage of survey results with such measures using a geographical information system (GIS).16 This was done using respondents' addresses, which were located to an accuracy of 100 m from their postcode.17 Full details of the methods used have been given elsewhere.18 The measures calculated are outlined in Table 1. Based on the residential location of each respondent, three measures of health service accessibility were calculated. These were an indicator of the existence of a main or branch GP surgery in the settlement where the respondent lived, an estimate of the travel time by road to the nearest surgery, and an estimate of the travel time by road to the nearest large hospital offering acute services.

Location of GP surgery

Figure 1 Map of the study area. Hospitals offering acute facilities are located in the city of Norwich (+) and the coastal town of Great Yarmouth (*).


The Norfolk Respiratory Health Survey was conducted by the East Norfolk Health Commission, the Department of Respiratory Medicine, Norwich, and the Department of Public Health Medicine, St Thomas' Hospital, London, between 1992 and 1993. It was posted in three waves to a random 10 per cent sample of inhabitants of the East Norfolk Health Commission catchment area aged between 20 and 44 years (Fig. 1). It consisted of 13 main questions covering self-reported symptoms, treatment, health service utilization, and personal details.

313

314

JOURNAL OF PUBLIC HEALTH MEDICINE

Table 1 The independent variables used in the regression models Description

Source

Coded as

Min.

The travel time by road to the nearest hospital offering acute services (minutes)

Arc/Info GIS

Continuous

3

33

The travel time by road to the nearest main or branch GP surgery (minutes)

Arc/Info GIS

Continuous

2

12

An indicator of the existence of a main or branch GP surgery in the settlement of residence of each respondent

Arc/Info GIS

0 = no surgery, 1 = surgery

_

_

Max.

The percentage of households where the head is of social class 4 or 5

1991 Census

Continuous

5

40

The percentage of households renting from a local authority

1991 Census

Continuous

0

34

The percentage of households without access to a car

1991 Census

Continuous

The sex of each respondent

NRHS

0 = male, 1 = female

7 —

38 -

The age of each respondent (years)

NRHS

Continuous

20

44

Whether respondent had ever smoked one or more cigarettes a day for a year or more

NRHS

0 = no, 1 =yes

-

-

Whether respondent had smoked in the last month

NRHS

0 = no, 1 = yes

_

_

NRHS, Norfolk Respiratory Health Survey.

all variables was available for each model. Significance testing was performed by checking that the removal of each variable from thefinalmodel resulted in a statistically significant change in model fit. At each step, categorical variables were tested for interaction effects. To account for any seasonal bias in responses, the wave within which each respondent was mailed was controlled for by fitting a dummy variable.

Results Forty-one per cent- of respondents reporting an asthma attack were male, compared with 44 per cent of those reporting waking with shortness of breath but not asthma (p = 0.04). Amongst those reporting an attack, 37 per cent had smoked compared with 51 per cent of respondents with shortness of breath but not asthma (p < 0.01). The prevalence of smoking within the previous month was also lower in those reporting an attack of asthma; 24 per cent vs 36 per cent (/? < 0.01). For respondents reporting an asthma attack, 397 (69 per cent) had seen a GP in the previous 12 months. The only predictor of consultation during this period was smoking status; those who smoked in the previous month were more likely to have seen a GP than those who had not (odds ratio (OR) = 1.67, p = 0.04). No trend was apparent with access to surgeries. Similarly, no associations were observed when the wider definition of asthma was used. Amongst those reporting an asthma attack, 95 per cent recounted having ever seen a GP for breathing problems. Table 2 shows that consultations were negatively associated with households lacking access to a car in the ward. Respondents were also less likely to have consulted if they lived outside a settlement containing a surgery. When the indicator of the presence of a surgery was replaced with that measuring the


Because no individual information on socio-economic status was available, surrogate measures calculated from 1991 UK Census of population data for the electoral ward were used. Electoral wards are small geographical areas, delineated for the purposes of local government. The measures chosen were the percentage of households where the head is in social class 4 or 5 (i.e. of low socio-economic status), the percentage without access to a car, and the percentage renting from a local authority. There is evidence that the severity of asthma may be higher in the more socio-economically disadvantaged19 and that they may be less likely to seek help in the case of medical problems.20 Furthermore, car ownership serves as more than a simple measure of socio-economic circumstance, as the effects of isolation from care may be amplified by a lack of private transport. These indicators were combined with information on respondent age, sex, and smoking status taken from the survey. The survey included questions on a range of respiratory symptoms. Our analysis focused on the respondents who had reported an attack of asthma in the previous 12 months. This definition may omit some asthmatics who were not aware of their condition, and hence would not recognize their symptoms as being due to an attack of asthma. Therefore, a wider definition of asthma was also adopted which included both respondents reporting an attack of asthma, plus those reporting waking with shortness of breath in the previous 12 months.15 The use of asthma medication was not included in this wider definition, as this may in itself be associated with accessibility. Amongst the 9764 respondents to the survey, 591 reported an attack of asthma, indicating a prevalence of 6.1 per cent. An additional 340 were included in the wider definition. Questionnaire responses were examined by logistic regression21 using the S-Plus package.22 A version of each model was fitted for each definition of asthma. A full dataset of

315

HEALTH SERVICE UTILIZATION FOR ASTHMA Table 2 The factors associated with having ever consulted a GP for breathing problems

Model for respondents reporting an asthma attack in the previous 12 months Explanatory variable

OR

95% Cl

p value

Model for respondents reporting an asthma attack and/or waking with shortness of breath in the previous 12 months OR

95% Cl

p value

When modelling GP accessibility as the existence of a surgery in the residential settlement of the respondent Respondent smoked one or more cigarettes a day for a year or more

*

*

*

0.42 (0.41)

Percentage of households without access to a car in ward

0.91 (0.94)

0.86-0.96 (0.90-0.98)