Clinical Practice Variation in Diabetes Screening and Monitoring in Auckland Primary Care

Abstract

Introduction: Diabetes is an increasingly important health issue in New Zealand and is estimated to affect 270,000 people with an additional 100,000 with undiagnosed diabetes. It is a major cause of death, disability and reduced quality of life. National guidelines for diabetes screening and monitoring have been developed for GPs targeting high risk population groups. However, very little is known about the current extent of diabetes screening and monitoring practice in routine primary care in New Zealand. The aim of this research was to firstly investigate general practice screening via HbA1c, Oral Glucose Tolerance Tests or serum glucose level tests for people without diabetes, and glycaemic control monitoring via HbA1c testing for patients with diagnosed diabetes. Secondly this research aimed to investigate the equity of these care processes by socioeconomic determinants (age, gender, deprivation level, and ethnicity) and clinically relevant factors such as prior cardiovascular disease (CVD), CVD risk score or history of smoking. Methods: Data were obtained from PREDICT -a clinical decision support program for CVD risk assessment and management used by general practitioners (GPs) and practice nurses in routine consultations in the Auckland Region from August 2002 until May 2011. As PREDICT is web-based, it simultaneously stores patient profiles and thus generates a research cohort. With permission from primary healthcare organisations (PHOs) this data were linked anonymously via encrypted National Health Index (NHI) numbers to the National Health Board NHI dataset, and regional laboratory results held by Diagnostic MedLab Ltd (2002-September 2009). A participant was considered ‘screened’ for diabetes if they did not have diabetes recorded by the GP or practice nurse at the time of first PREDICT assessment and if they had a recorded result for an Oral Glucose Tolerance Test (OGTT), HbA1c, or blood lucose test either five years prior or two weeks after the first PREDICT assessment. A participant was considered eligible for screening if they met age group and risk factor screening criteria based on national guideline recommendations. A participant was considered ‘monitored’ for diabetes if they were identified as having diabetes by their GP or practice nurse, and if they had one or more HbA1c test recorded after entry into the PREDICT cohort. Univariate, stratified and multivariate analyses (adjusting for sex, ethnicity, age, New Zealand Deprivation Index [NZDep01] quintile, history of CVD, New Zealand-adjusted Framingham CVD risk score and smoking) were conducted for screening and monitoring of diabetes using STATA version 8.2. Results: As of May 2011, a total of 166,996 patients (132,023 patients without diabetes and 34,973 patients with diabetes) had been risk assessed and managed using PREDICT in the Auckland Region. At the time of the first CVD risk assessment, 78,238 patients (59%) had laboratory records indicating screening for diabetes. However, taking into account only those patients (n=68,035) for whom complete laboratory records were available (2003-2008), 70% had been screened for diabetes. Eligibility criteria for screening according to the 2003 CVD guidelines were applied in those patients with complete laboratory data, and found that 75% (n=44,342) of the eligible patients and 68% (n=6,156) of the potentially ineligible patients by age, gender, ethnicity and some clinical criteria had been screened for diabetes. Overall screening was equitable by gender. Compared to European and Other ethnic group, all other ethnicities were more likely to be screened; Pacific people were 30% more likely (adjusted RR 1.30[95% CI 1.28, 1.31]), Indian 23% (adjusted RR 1.23 [95% CI 1.21, 1.25]), Maori 14% (adjusted RR 1.14 [95% CI 1.13, 1.16]) and Asian were 7% (adjusted RR 1.07 [95% CI 1.05, 1.09]) more likely. Higher levels of screening were observed in those most deprived (NZDep01 quintile 5) compared to those resident in the least deprived areas (adjusted RR 1.06 [95% CI 1.05, 1.08]), and those in 55-64 age group compared to all other age groups. Having a prior history of CVD, the NZ-adjusted Framingham CVD risk score, and smoking were not associated with an increased likelihood of screening. Of those with a history of prior CVD or with a high CVD risk (NZ adjusted Framingham CVD score > 15%), only 62% and 57% had undergone diabetes screening respectively. Of the 34,973 participants with diabetes, only 5,397 were able to be included for follow-up monitoring analysis due to unavailability of recent laboratory data (September 2009- May 2011). As a conservative estimate to determine annual or near annual glycaemic monitoring, a patient with diabetes who had been followed for 2.5 to 3.5 years should have had at least two tests, 3.6-4.5 years of follow-up should have at least three tests, 4.6-5.5 years of follow-up should have at least four tests and greater than 5.6 years of follow-up should have at least five follow-up tests. Approximately 83% of the patients met these HbA1c monitoring criteria. The likelihood of monitoring for the 5,397 people with diabetes did not vary by gender, ethnicity, age group, deprivation level, history of prior CVD or CVD risk score or smoking status. Conclusion: While it is possible that people in the PREDICT cohort could have had laboratory testing in other regions of New Zealand or in hospital if admitted during the time periods of interest, the findings demonstrate an evidence practice gap for general practice diabetes screening and monitoring. About one in four of those eligible had not been screened. There is a positive bias towards screening those most at risk by sociodemographic factors especially age and high risk ethnicity group but not by history of important relevant CVD factors. For those patients with diabetes for whom annual monitoring could be estimated, the gap was smaller (about 1 in 6 had not been monitored appropriately) with no differences found by sociodemographic or clinical risk factors. To improve the level of screening and monitoring, a multidisciplinary and intersectoral approach is necessary. Population health measures supporting healthy eating and healthy activity (such as applying a “fat tax”, reducing costs of fruit and vegetables, safe walkways and bike paths) are important to reduce the incidence of diabetes. Raising community awareness of diabetes screening and monitoring through media campaigns and education programmes could be implemented targeting those with known clinical risk factors (such as prior CVD) as well as targeting by increasing age and high risk ethnicity groups. At a provider level, target goals could be set for PHOs and DHBs, and performance incentives provided for meeting targets. Additional GP and patient reminders could be incorporated to increase screening proportions among patients with clinical risk factors. A limitation of this research was the incomplete laboratory records available for patients assessed after 2009. Further research using the new TestSafe regional laboratory repository (hospital and community laboratory test) is recommended and/or using the National laboratory claims dataset (national community laboratory claims). Measuring the extent and equity of care processes for diabetes is an important audit required to further improve the quality of care and service delivery in order to reduce the burden of diabetes in New Zealand.