Diabetes has been described as a modern epidemic within Australia and the developing world. General practice is at the coalface of diabetes management, yet little is known about its impact on this disease.
In 1999, the Australian Government introduced the Enhanced Primary Care (EPC) Program, designed to encourage general practitioners to participate in multidisciplinary care.1 The program enabled funding of structured care and coordinated multidisciplinary support for patients with chronic or complex conditions. While there has been evidence of population-based improvements in health outcomes through a region-wide coordinated care approach,2 there have been no reports on the effect of a structured practice-based intervention.
In 2002, the Condamine Medical Centre moved to formalise diabetic management by adopting a program of structured care for patients with diabetes. This involved care planning, regular recall, and multidisciplinary review. We report a quality-of-care audit to determine the clinical impact of this change.
Any patient with type 2 diabetes who consented to be involved in the multidisciplinary care process could participate in the program. Enrolments occurred via self-presentation to the practice. The records of patients enrolled from July 2002 to December 2003 were used in the audit.
A pre-formatted care plan was generated using an electronic patient management system (Medical Director, Health Communication Network, Sydney, NSW) containing the required Health Insurance Commission administrative details, a clinical section with parameters set by the National Diabetes Goals,3 and an area for individual clinical notes.
Two practice nurses attended to all patients and were responsible for managing the recall system, ensuring pathology tests were ordered (serum lipid levels, glycated haemoglobin [HbA1c] concentration, and annual urinary microalbumin estimation), measuring and recording the patient’s weight, waist circumference, blood pressure, and visual acuity, and for performing foot screening before every general practice visit.
A retrospective audit of the electronic records was performed to evaluate the clinical impact of the program on the patients. The program was funded by payments for a combination of Enhanced Primary Care item numbers and normal consultation item numbers.
The impact of the program was measured by:
Change in known cardiovascular disease (CVD) risk factors: weight, waist circumference, body mass index (BMI), serum lipid levels, systolic and diastolic blood pressure.
Change in indicators of risks associated with poor control of diabetes: HbA1c concentration, presence of foot lesions, clinically significant hypoglycaemia in the past 12 months.
Analysis by CVD risk groups: 5-year CVD risk was calculated using the United Kingdom Prospective Diabetes Study (UKPDS) risk formulas for coronary heart disease4 and stroke5 for each patient. Risk allocations were low (5-year CVD risk, < 10%), moderate (5-year CVD risk, 10%–15%) or high (5-year CVD risk, > 15%).
Analysis of 5-year CVD risk: controlling for non-modifiable risk factors of age, sex and duration of diabetes, again using the UKPDS risk formulas for coronary heart disease and stroke at both time points.
the risk for each patient was calculated at 2 years assuming that the level of modifiable risk factors at baseline remained unchanged.
the “ideal” 5-year risk for each patient was calculated assuming all modifiable risk factors achieved optimal levels (HbA1c concentration, 7%; systolic blood pressure, 130 mmHg; non-smoker; and ratio of total cholesterol level : high-density lipoprotein [HDL] cholesterol level, 3.5).
Both measures were compared against the results actually achieved at 2 years.
Comparing our results against international benchmark data — the intensive treatment arm of the UKPDS,4,5 and the Steno 2 trial for patients with microalbuminuria.6,7
Analysis was by intention to treat. For patients who withdrew, the last complete set of data from an intermediate check was carried forward to complete the 2-year dataset. Significance was calculated using McNemar’s test for paired categorical variables, and two-tailed paired sample t tests for continuous variables. The χ2 test was used to test for correlation of categorical variables at one time point, and analysis of variance was used as a test of correlation of continuous variables at one time point. Significance was set at P < 0.05. Calculations were performed using SPSS, version 11 (SPSS Inc, Chicago, Ill, USA).
Four hundred and four consecutive patients with type 2 diabetes were enrolled (70% of the patients with type 2 diabetes attending the practice at the time). During the study period, 67 people (16.6% of 404) withdrew from the program or died: 26 patients (39% of 67) left the practice when a nearby town’s general practice re-opened; 30 patients (45%) left the program voluntarily (16 [24%] because their newly diagnosed diabetes was well controlled by diet modification; and 14 patients [21%] for other reasons); and there were 11 deaths (16%).
There were equal numbers of men and women in the program, with similar age distributions (mean, 64 years) and duration of diabetes (mean, 6 years; 5.4 years for women and 6.5 years for men). Pensioners comprised 71% of the patients with type 2 diabetes. Forty per cent of the patients in the program had a calculated risk of an adverse CVD event within 5 years of < 10% (53% of women, 22% of men) and 29% had a risk of ≥ 15% (25% of women, 58% of men) (P < 0.001 for both comparisons).
Box 1 presents changes in CVD risk factors from enrolment to audit at 2 years. There was a significant decrease in mean waist circumference, and a non-significant mean weight reduction. Significant improvements occurred in systolic and diastolic blood pressure and in all serum lipid levels except HDL cholesterol.
The prevalence of foot lesions declined, but there was no improvement in the prevalence of foot ulcers.
CVD risk analysis by risk group (Box 2) revealed that the greatest falls in systolic blood pressure occurred in the high-risk group. All risk groups experienced falls in diastolic blood pressure of around 7 mmHg. The high-risk group achieved higher absolute reductions in mean total cholesterol and low-density lipoprotein (LDL) cholesterol levels than those in the moderate- and low-risk groups. Prescribing of statins increased across all risk groups within the cohort.
The mean HbA1c concentration remained almost identical over the 2 years (7.4%, 7.3%), while that of the UKPDS intervention group, with the same average duration of diabetes (6–8 years), rose from 7.1% to 7.5%.
The proportion of patients tested for microalbuminuria in the previous 12 months rose from 19.8% of the population to 75.2%. Patients with microalbuminuria who were the same age as those in the Steno 2 trial showed similar improvements in risk factors at 2 years to those seen at 4 years in the Steno trial (the first reported endpoint in that trial) (Box 3). The results for patients with microalbuminuria of all ages showed similar changes.
Our audit revealed improvements in known CVD risk factors, and in outcomes indicative of control of diabetes. While direct comparisons are not possible, benchmarking against internationally published best practice intensive treatment suggests that structured diabetes care produced broadly similar results to those achieved by intensive therapy. This simple intervention may have been effective in reducing CVD risk in patients with diabetes.
Management priorities in the care of patients with type 2 diabetes have undergone a substantial change in the past decade. Optimal care places increased emphasis on reducing cardiovascular risk, in addition to the processes of care and glycaemic control.8
Care systems that provide one or more of the elements of chronic disease management: delivery system design (ie, team care, coordination and recall systems), decision support, patient self-management, clinical information systems, and use of community resources, have been shown to improve health outcomes.9 These elements are at the heart of current general practice incentives and the Enhanced Primary Care program.
There was no improvement in HbA1c concentrations or in the 5-year CVD risk. However, HbA1c concentrations and CVD risk in a population are influenced by non-modifiable factors.10 Adjusting for non-modifiable risk factors showed more accurately the probable effect of the intervention. We believe that controlling for non-modifiable risk should be more widely used in assessing the value of interventions.
The increased prescribing of statins in patients at low risk, and in particular in women at low risk, suggests that current lipid management guidelines may be too inclusive.11 These guidelines refer to all patients with diabetes as being “high” risk and requiring treatment, whereas individual risk calculations in our patients show that this may not be the case.
Our study used baseline data as a control device, and hence does not have the strength of evidence of a randomised controlled trial. A multisite randomised controlled trial of integrated care will be required to confirm what appear to be effective changes in known risk factors for CVD in diabetic patients.
This audit is an example of a general practice which has evaluated and monitored its care outcomes. The key to being able to implement such programs is financing systems and infrastructure support. We hope this report encourages GPs to utilise the abundance of health data they have stored electronically on their desktop.
1 Changes in cardiovascular disease risk factors after structured diabetes care
2 Changes in lipid levels and blood pressure (BP) control after structured diabetes care, by 5-year cardiovascular disease (CVD) risk at enrolment
Abstract
Objective: To assess the impact of structured diabetes care in a rural general practice.
Design and setting: A cohort study of structured diabetes care (care plans, multidisciplinary involvement and regular patient recall) in a large general practice in a medium-sized Australian rural town. Medical care followed each doctor’s usual practice.
Participants: The first 404 consecutive patients with type 2 diabetes who consented to take part in the program were evaluated 24 months after enrolment in July 2002 to December 2003.
Main outcome measures: Change in cardiovascular disease risk factors (waist circumference, body mass index, serum lipid levels, blood pressure); change in indicators of risks associated with poorly controlled diabetes (glycated haemoglobin [HbA1c] concentration, foot lesions, clinically significant hypoglycaemia); change in 5-year cardiovascular disease risk.
Results: Women had a lower 5-year risk of a cardiovascular event at enrolment than men. Structured care was associated with statistically significant reductions in mean cardiovascular disease risk factors (waist circumference, − 2.6 cm; blood pressure [systolic, − 3 mmHg; diastolic − 7 mmHg]; and serum lipid levels [total cholesterol, − 0.5 mmol/L; HDL cholesterol, 0.02 mmol/L; LDL cholesterol, − 0.4 mmol/L; triglycerides, − 0.3 mmol/L]); and improvements in indicators of diabetic control (proportion with severe hypoglycaemic events, − 2.2%; proportion with foot lesions, − 14%). The greatest improvements in risk factors occurred in patients with the highest calculated cardiovascular risk. There was a statistically significant increase in the proportion of patients with “ideal” blood pressure (systolic, < 130 mmHg; diastolic, < 80 mmHg) and LDL cholesterol level (< 2.5 mmol/L) of 6.4% and 20.5%, respectively.
Conclusions: Implementing structured care in this rural general practice coincided with improved risk factor management, and may have contributed to the improvement. The greatest benefits were in patients with high cardiovascular risk.