Which mechanisms of primary care influence the population outcomes of standardised mortality rates under 65 years and of emergency admissions for ambulatory care sensitive conditions?

Talk Code: 
Richard Baker
L Steve Levene, M John Bankart, Arch G Mainous III, Tim Stokes, Kate Honeyford
Author institutions: 
University of Leicester, University of Florida, University of Otago, Imperial College


It is widely accepted that health systems with stronger primary care tend to deliver better outcomes, but the mechanisms by which primary care achieves this are not fully understood. Starfield (Milbank Quarterly 2005) proposed six mechanisms, including access, focus on prevention, and primary care characteristics. Building on these and on further research, we developed a conceptual model incorporating population characteristics and primary care mechanisms (capacity to meet need, access, disease prevention, detection, and management, consultation performance and continuity) and applied it to a study of all cause premature (aged under 75) mortality at practice level in England (Baker et al, BMJ Open 2016). Although the findings tended to support the model, it needs testing with other outcomes.


Using a cross sectional design, we investigated the all cause standardised mortality ratio (SMR) under aged 65, and emergency admissions for ambulatory care sensitive conditions (NIH, Access to Health Care in America, 1993), using 2009-10 data, the most recent year with mortality data available at practice level. Guided by our model, the selected population variables were deprivation, smoking, diabetes prevalence, ethnicity, and, for admissions, age. The primary care variables were numbers of GPs, patient experience of access using items from the general practice patient survey, and performance indicators for disease detection, prevention and management from the Quality and Outcomes Framework. Statistical analysis involved linear regression (SMR under 65) and negative binomial regression (admissions). All data were publicly available from the Health and Social Care Information Centre.


Of the 8579 practices in England, complete data were available for 7837. In accordance with the conceptual model, population characteristics, including deprivation, were the most powerful predictors of each of the two outcomes. Mean SMR under 65 was 106. In the initial analyses, deprivation, followed levels of smoking, were the most powerful predictors of SMR. Primary care characteristics associated with mortality included numbers of GPs (an increase of one GP per 10,000 patients was associated with reduction in SMR of 0.4), and detection of hypertension (1% increase in people recorded as hypertensive was associated with reduction of SMR of 0.6). Characteristics associated with admissions included access (0.17% decrease in admissions for 1% increase in being able to get an appointment quickly, and 0.14% decrease in admissions for 1% increase in being able to book an appointment in advance).


1.Detection of hypertension by English general practices needs to improve in order to reduce cardiovascular mortality. 2.Plans to increase the capacity of English general practice are supported by the findings. 3.In addition to addressing the needs of individuals, primary care should respond to the needs of populations in order to improve population outcomes.However, longitudinal studies are needed for further evaluation of the model.

Submitted by: 
Richard H. Baker
Funding acknowledgement: 
No external funding