Article Text


Characteristics of general practices associated with emergency-department attendance rates: a cross-sectional study
  1. R Baker1,
  2. M J Bankart1,
  3. A Rashid2,
  4. J Banerjee3,
  5. S Conroy1,
  6. M Habiba4,
  7. R Hsu5,
  8. A Wilson6,
  9. S Agarwal1,
  10. J Camosso-Stefinovic1
  1. 1Department of Health Sciences, NIHR CLAHRC for LNR, University of Leicester, Leicester, UK
  2. 2Leicestershire County and Rutland PCT, Leicester, UK
  3. 3University Hospitals of Leicester NHS Trust and Acute Care Lead, NHS Leicester Leicestershire & Rutland, Leicester, UK
  4. 4Cancer Studies, University of Leicester, and Associate Medical Director, Leicestershire County and Rutland PCT, Leicester, UK
  5. 5Department of Medical and Social Care Education, University of Leicester, and Leicestershire County and Rutland PCT, Leicester, UK
  6. 6Department of Health Sciences, University of Leicester, Leicester, UK
  1. Correspondence to Professor Richard Baker, Department of Health Sciences, University of Leicester, 22–28 Princess Rd West, Leicester LE1 6TP, UK; rb14{at}


Background Strategies are needed to contain emergency-department attendance. Quality of care in general practice might influence the use of emergency departments, including management of patients with chronic conditions and access to consultations.

Aim The aim was to determine whether emergency‐department attendance rates are lower for practices with higher quality and outcomes framework performance and lower for practices with better patient reported access.

Design A cross-sectional study.

Setting Two English primary-care trusts, Leicester City and Leicestershire County and Rutland, with 145 general practices.

Method Using data on attendances at emergency departments in 2006/2007 and 2007/2008, a practice attendance rate was calculated for each practice. In a hierarchical negative binomial regression model, practice population characteristics (deprivation, proportion of patients aged 65 or over, ethnicity, gender) and practice characteristics (total list size, distance from the emergency department, quality and outcomes framework points, and variables measuring satisfaction with access) were included as potential explanatory variables.

Results In both years, greater deprivation, shorter distance from the central emergency department, lower practice list size, white ethnicity and lower satisfaction with practice telephone access were associated with higher emergency-department attendance rates.

Conclusions Performance as indicated by the quality and outcomes framework did not predict rates of attendance at emergency departments, but satisfaction with telephone access did. Consideration should be given to improving access to some general practices to contain the use of emergency departments.

  • Primary care
  • emergency department
  • utilisation
  • general practice
  • quality of care

Statistics from


Emergency departments in England, in common with some other countries, are under pressure because of rising demand, the total number of attendances increasing from 14.6 million in 2002/2003 to 19.6 million in 2008/2009.1 2 There are variations in emergency-department use between general practices, but the characteristics of practices, including quality of care, have not been found to explain these variations. Patient variables including levels of deprivation, ethnicity and numbers of pensioners living alone were the explanatory factors in a study in London.3 People who live closer to emergency departments tend to make greater use of them,3–5 and practices with high emergency-department attendance rates may also have high demands on out-of-hours care.6 Initiatives to retain patients in primary care or direct them to alternative services such as walk-in centres or emergency care practitioners do not appear to have affected emergency-department attendance rates.7–10 Nevertheless, patients' ability to obtain care in general practice may be a factor that influences their decisions on the use of emergency departments.4

Quality of care has been defined as having two principal dimensions, effectiveness and access.11 Improvements in the primary-care management of people with chronic conditions (an aspect of effectiveness) may be expected to alleviate the pressure on emergency services. The quality and outcomes framework is an incentive scheme with indicators for clinical care, practice organisation and patient experience.12 The clinical indicators address common chronic conditions—for example, hypertension and diabetes. Improved access to a doctor or nurse in general practice may also be hypothesised to reduce demand on emergency services. For example, changes in out-of-hours primary-care services in 2004 were associated with an increase in the use of emergency departments for non-traumatic conditions.13 Initiatives to improve access to general practice have included, from 2006/2007, a financial reward scheme dependent on the results of a survey of patient experience of access.14 To date, there has been no evidence on the impact of these initiatives on use of emergency departments. Therefore, we undertook a study to investigate whether the measures of practice performance as described by the quality and outcomes framework and access survey are associated with emergency-department attendance rates. Our hypotheses were that attendance rates would be lower for the patients of practices with higher quality and outcomes framework performance, and lower for practices with better reported access.



The study location was Leicestershire, where there is a central emergency department based in a hospital trust in Leicester city, situated in the centre of the county, surrounded by rural areas served by market towns. There are other emergency departments in cities in neighbouring counties, including those in Nottingham, Coventry, Derby and Kettering, and attendance at these departments may be easier for Leicestershire residents living near the county border. There are two primary-care trusts (Leicester City and Leicestershire County and Rutland), 145 general practices, a walk-in centre in one small town, and several minor injuries units. The total population is approximately 900 000,15 with wide socio-economic diversity, and a large ethnic minority population in Leicester city and three other suburban areas of the county.

Emergency attendance rates

Anonymised attendance data for the 2 years 2006/2007 and 2007/2008 to all emergency departments were available to the primary-care trusts. There were 166 230 attendances in 2006/2007 and 170 050 in 2007/2008 by residents of Leicester City and Leicestershire to any emergency department, and we included all these attendances in the study. The number of emergency-department attendances per practice was divided by the total practice list size to create a rate per person per year for each practice.

Practice characteristics

We obtained information on practice list size including age and gender groupings, and ethnicity of practice populations from the primary-care trusts. We obtained publicly available information on practice performance from the quality and outcomes framework.16 The framework incorporates a series of performance indicators, divided into clinical, organisational, patient experience and additional services domains, with performance expressed in terms of points in each domain. We used the 2006/2007 and 2007/2008 quality and outcomes framework data for the two study years respectively.

The patient access survey is a nationally administered survey of samples of patients aged 18 years or over registered at each practice. In the first 2 years of the survey, the questionnaire included five questions on access—whether the patient was able to get an appointment on the same day or within 2 days, able to book an appointment in advance (more than 2 full days in advance), able to make an appointment with a particular doctor, satisfaction with getting through to the practice on the telephone and satisfaction with the practice opening hours. The 2006/2007 and 2007/2008 surveys were used in this study.14 17 The distance of each practice in miles from the hospital was obtained from a route planner,18 and the practice index of multiple deprivation 2007 was used as the indicator of deprivation for both years.19

Statistical methods

Descriptive statistics and univariable analyses of the attendance rates and candidate predictors were carried out for each year separately. The data were expected to be overdispersed counts, and therefore an appropriate analysis method was negative binomial regression using the log of the practice list size as an offset. If the variance of the data exceeds that of the mean, the negative binomial model makes an adjustment to the log-likelihood function to take this into account. If the data proved to be neither overdispersed nor underdispersed, a Poisson model would be appropriate. There were no missing data for the study variables.

We expected certain variables to predict attendance rates, and therefore used a two-level hierarchical multiple regression model (implementing a backward stepwise procedure in level 2), using data for the 2007/2008 year. Analyses were undertaken in SAS version 9.1. Level 1 included variables assumed to have an impact on emergency-department attendance rates, as indicated by previous studies and informed by the univariable models. The candidate level 1 variables were: deprivation,3 practice size, the proportion of the practice population aged 65 or over, the proportion of the practice population categorised as white, the proportion who were males and the distance of the practice from the hospital.4 5 We intended to use the proportion of people on practice quality and outcomes framework coronary heart disease registers as an indicator of the level of morbidity in practices, but this variable was very highly correlated with the proportion of people aged 65 and over in each year, so we omitted it from our analyses to reduce multicollinearity.

Level 2 variables were chosen to test the hypotheses that (1) practice performance as measured by the quality and outcomes framework and (2) patient reported access would influence attendance rates. The variables chosen were three quality and outcomes framework domains (clinical, organisational and patient experience), and the five questions from the access survey. These level 2 variables were then entered into the model along with all level 1 predictors. A backward stepwise phase was then undertaken in which non-significant level two variables were sequentially removed in order to determine which of the level 2 variables were significant multivariable predictors of the attendance rate.20 21 Level 1 variables were forced into the models at stages 1 and 2.

Internal model validation was undertaken by bootstrapping. The model derived from 2007/2008 was then applied to the previous year's data (2006/2007). The distance of practices from the hospital, the deprivation score and the proportion of Caucasians were recorded using the same variable in each year; otherwise all predictors were specific to each particular year.


The mean attendance rate for people in Leicestershire was approximately 0.18 per person per year (table 1). Approximately 70% of attendances occurred between 08:00 and 20:00. Most practices achieved a high level of points in the quality and outcomes framework, and responses to the access survey were also generally positive.

Table 1

Descriptive statistics (n=145, for all variables)

In the univariable analyses (table 2), the following variables were not significantly associated with practice emergency-department use rates in both years: the percentage of patients identified as having coronary heart disease, gender, and the total quality and outcomes framework clinical points.

Table 2

Univariable analysis of predictors of practice attendance rate (n=145) 2006/2007 and 2007/2008

In level 1 of the multivariable model using data from the 2007/2008 year, higher attendance rates were associated with: higher levels of deprivation, shorter distance from the emergency department, lower practice list size and an increasing proportion of white ethnicity (table 3). At level 2, none of the quality and outcomes framework variables predicted attendance rates. However, as the proportion of patients who were satisfied with being able to get through to someone on the telephone at the doctor's surgery increased, the attendance rate decreased. As satisfaction increased by 1 unit (1%), the log of the count decreased by 0.005. The exponentiated β coefficient is 0.995. This represents a decrease in the attendances count of 0.5% for every percentage point increase in access score. A 1% increase in satisfaction thus equates to nine fewer attendances per year for a practice of 10 000 patients (0.5% of 1800) and around 850 (0.5% of 170 000) fewer attendances for Leicestershire as a whole. Bootstrapping supported internal validity (95% bootstrapped CIs for the estimate for satisfaction with opening hours −0.0026 to −0.0008 for 2007/2008 and −0.0024 to −0.0003 for 2006/2007).21 In order to evaluate the model derived from the 2007/2008 data, we fitted the model to the 2006/2007 data. All predictor variables had the same direction of slope, and coefficients were very similar, providing reassurance that the associations identified in the model are stable across at least these 2 years (table 4).

Table 3

Final multivariable model, 2007/2008, using negative binomial regression with log of list size as offset, after backward selection process

Table 4

Final multivariable model, 2006/2007, using negative binomial regression with log of list size as offset, after backward selection process


Summary of findings

The first hypothesis, that a higher performance in the quality and outcomes framework would be associated with lower emergency-department use, was not supported. In contrast, the study provided evidence to support the second hypothesis that reported patient perception of poor quality of access, specifically access by telephone, in general practice is associated with a modest increase in the use of emergency departments. Higher patient satisfaction with telephone access to primary care was associated with lower attendance rates to secondary care.


Several study limitations should be noted. We may have omitted one or more variables that could help explain attendance rates. The study involved only one English county, and the findings may not be directly applicable elsewhere. The access survey is an imperfect measure. First, it is composed of patient reports of access rather than an objective measure. Second, the survey addresses only one aspect of the concept of access, namely the ability of patients to call on services when required.22 23 Third, the response rate to the survey varied between practices and was low (the median practice response rate in 2007/2008 was 47%). However, in a recent study, the response rate to the access survey was found to have little impact on practice questionnaire scores,24 although in our study practices with higher reported access tended to have higher survey response rates. Although the study investigates associations only, the view that patients who perceive their practice as providing poor access will be more frequent users of emergency departments is both plausible and supported by other studies.7 25


The finding that an increasing proportion of ethnic minority patients in the practice population was associated with decreasing emergency-department use may reflect difficulties patients from ethnic minority communities experience in accessing services such as emergency departments, or that, in the South Asian population of Leicester, families tend to place importance on caring for their relatives themselves and are consequently reluctant to resort to the emergency department.26 Although a greater proportion of patients aged 65 years or older has been found to be associated with higher emergency admission rates,27 this did not predict emergency-department attendance rates in the multivariable analysis. The study also suggests that improved access in general practice should be one part of initiatives to manage demand in the prehospital environment and control recourse to emergency departments. However, a focus only on telephone access should be avoided, since patient perceptions of different aspects of access are correlated. Attention should also be given to other aspects of access as well, including the ability to get an appointment with a doctor quickly (within two working days), the ability to consult a particular doctor (one aspect of continuity) and the ability to obtain home visits. Furthermore, although the effect of access is small, unlike some other factors associated with attendance rates, access can be improved through service interventions, and so it warrants attention as a means to reduce attendance rates.

If improved access is to be achieved, a better understanding of the needs of the patients of different practices is required. In some practices, demand for care may be higher; in others, the ability of some patients to negotiate with practice staff to obtain appointments may be poor.28 For these reasons, a uniform strategy to improve access is unlikely to be successful.


View Abstract


  • Funding National Institute for Health Research. The views expressed in this paper do not necessarily reflect those of the NIHR or the Department of Health.

  • Competing interests (1) AR and MH are employees of Leicestershire County and Rutland Trust, and MH and JB are employees of University Hospitals of Leicester Trust that might have an interest in the submitted work in the previous 3 years; (2) AR's spouse is an employee of Leicestershire County and Rutland Trust; and (3) RB, MJB, SC, RH, AW, SA and JC‐S have no non-financial interests that may be relevant to the submitted work.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.