Last updated 28/11/2016

Diabetes single map Diabetes double map
Consumer summary (PDF, 43KB)

Note: If you have trouble opening the maps, please try a different browser such as Google chrome or Mozilla firefox instead.

The goal of this Atlas of Healthcare Variation domain is to investigate the quality of care provided to people with diabetes. These data are not intended for judgement or definitive statements of quality, rather they are intended to raise questions about potential areas for quality improvement. The indicators were developed with the assistance of an expert advisory group.

Update to include 2015 data
(Updated 25 November 2016)

This Atlas has been updated with data from 2013–15.

Key messages

  • In 2015, diabetes prevalence varied more than two-fold by district health board (DHB), from between 10 and 27 percent of a DHB population aged 65–74.
  • The regular use of medicines for glycaemic control (insulin or metformin) varied 1.8-fold by DHB, from 34 to 61 percent of those with diabetes regularly receiving insulin or metformin.
  • The proportion of bed-days occupied by people with diabetes for any reason increased in 2015. This varied four-fold between DHBs. The percent of bed-days occupied by people with diabetes aged 6574 variation ranged from 10 to 40 percent.

Key findings

Diabetes prevalence is increasing faster in some DHBs than others

The prevalence of diabetes increased significantly with age, from a mean of 0.38 percent in those aged 0–24 years to 19 percent in people aged 75 years and older.

Pacific peoples had a significantly higher prevalence of diabetes than all other ethnic groups, while those identifying as European/Other had significantly lower rates of diabetes. People of Indian ethnicity are not presented separately here; however, rates of diabetes in Indian populations are similar to those observed in Pacific peoples.

Table 1: Diabetes prevalence by age and ethnicity (percent), 2014

Age (years): 0–24 25–44 45–64 65–74 75+ Total
Māori 0.35 3.7 14.2 27.9 29.6 5.3
Pacific peoples 0.36 7.0 29.7 52.5 48.8 10.5
Asian 0.14 3.1 13.5 29.7 34.4 6.0
European/Other 0.44 2.0 6.1 13.6 17.9 5.3
Total 0.37 2.8 8.9 16.8 19.7 5.7
Good glycaemic control is important for outcomes

If HbA1c does not meet an agreed target with dietary and lifestyle changes, drug therapy is recommended. Insulin and metformin are well proven. Evidence suggests good glycaemic control benefits microvascular outcomes and, if started early enough, on long-term macrovascular outcomes.[1]

Insulin and metformin use varied

Given both the lack of data on clinical condition and the inability to split by the type of diabetes, these indicators of medication use are not intended to suggest an ‘ideal’ rate of use, rather they provide a high level view. Wide variation may raise questions such as do DHBs with lower than average rates of medication use have lower or higher rates of diabetes complications?

Nationally, 54 percent of people with diabetes aged 25 years and over regularly received either insulin or metformin, with use highest in those aged 65–74. This varied 1.8-fold by DHB, ranging from 34 to 61 percent.

Metformin is the standard initial drug treatment for type 2 diabetes. Use increased with age. On average, 21 percent of people with diabetes aged 25–44 regularly received metformin compared with 52 percent of those aged 65–74. Pacific peoples and Asian ethnic groups appeared to have higher rates of dispensing, than Māori or European/Other groups. The significance is difficult to interpret in the absence of clinical information.

On average, 19 percent of people with diabetes regularly received insulin. People of Asian ethnicity received significantly less insulin than people of all oter ethnic groups. As might be expected, insulin use was highest in the 0–24-year age group with diabetes, with 53 percent of these regularly dispensed insulin.

Intensive management of high blood pressure and microalbuminuria

On average, 45 percent of people with diabetes received angiotensin-converting enzyme inhibitor (ACEI) or angiotensin II receptor blockers (ARB) medicines. Use increased significantly with age up to 74 years. There was limited variation between DHBs. Intensive management of blood pressure and microalbuminuria is recommended to prevent progression of renal disease in diabetes. ACEI and ARB are first-line treatments for raised blood pressure and/or microalbuminuria.

The impact of diabetes on life expectancy is magnified by a younger age at diagnosis, highlighting the importance of glycaemic control, blood pressure management and prevention of kidney disease in younger people with diabetes.

Complications – admissions for diabetic ketoacidosis and hypoglycaemia varied by age

Note data from 2011 is presented for these indicators as they were not affected by the update to the VDR.

Nationally, admissions to hospital for diabetic ketoacidosis were ten-fold higher in those aged 0–24 years. In contrast, admissions for hypoglycaemia were highest in those aged 65 years and over. The numbers in each age group were too low to report by DHB.

Complications – lower limb amputations increase with age

Lower limb amputation rates increased significantly with age, with 95 percent of amputations occurring in those aged 45 years and over. This rare complication affected only 0.19 percent of the diabetes population in a year.

There was no consistent significant variation by DHB in lower limb amputation rates or admissions for hypoglycaemia in people with diabetes. There was limited variation between DHBs in admissions for diabetic ketoacidosis, although two DHBs had consistently lower rates.

People with diabetes occupied more bed-days for any reason (2013–15 data)

The number of medical and surgical bed-days occupied by people with diabetes was compared with the total occupied medical and surgical bed-days. On average, people with diabetes occupied 18.2 percent of total bed-days, despite a population prevalence of diabetes of 5.7 percent. In 2015, age-specific rates varied up to four-fold.

Age had a significant effect on occupied bed-days. People with diabetes aged 45–64 years occupied 23 percent of total bed-days, compared with a population prevalence of 8.8 percent of people with diabetes in this age group.

People of Māori, Pacific peoples and Asian ethnicities with diabetes occupied significantly more bed-days than those in the European/Other ethnic group – 26 percent on average compared with 16 percent.

Regular laboratory testing was lowest for screening renal disease (ACR)

People with diabetes who received one or more laboratory tests in a year for HbA1c, albumin: creatinine ratio (ACR) and estimated glomerular filtration rate (eGFR) was analysed. On average, 86 percent received an HbA1c test, 65 percent an ACR test and 83 percent an eGFR test.

Rates differed by ethnicity. Pacific peoples and Asian ethnic groups had higher rates of testing than those of Māori and European/Other ethnicities.

Table two shows the percent of people receiving all three tests by age and ethnic group.

Table two: People receiving all three tests

Age (years) 25–44  45–64 65–74 75+ Total
Māori 45.24 64.1 68.3 62.7 61.4
Pacific peoples 57.01 74.8 77.9 68.3 71.6
Asian 42.61 73.7 76.3 69.9 67.8
European/Other 38.45 61.3 67.0 59.5 60.1
Total 43.72 66.0 69.3 61.0 62.8


Questions raised

  • How many of these results can be explained by the predominant type of diabetes?
  • Are rates for specific indicators lower or higher than might be expected?
  • Is there room for improvement in any of these indicators?
  • Do results reflect local differences in care?

Method and data source

This domain draws on data contained in the Virtual Diabetes Registry, which was developed by the Ministry of Health to track the number of people diagnosed with diabetes. The registry combines and filters various sources of health information, including the National Minimum Dataset, the National Non-admitted Patients Collection (outpatients), the Pharmaceutical Collection, the Laboratory Claims Collection and the Primary Health Organisation Enrolment Collection. The registry was used to estimate the incidence of diabetes. These data are based on an algorithm so should be interpreted with some caution. The Atlas does not use any patient-identifiable data.


The Pharmaceutical Collection contains claim and payment information from community pharmacists for subsidised dispensing. This collection does not indicate whether a medicine was taken or whether the dose was effective. Over-the-counter medicines are not included.

In selecting indicators for oral hypoglycaemic medication use, the group decided to focus only on metformin, as the first line agent for people with type 2 diabetes and insulin as the key medication for people with type 1 diabetes. There is no ideal rate of medicine use in people with diabetes as it depends on clinical need, however, wide variation between DHBs or ethnic groups raises questions as to why that might be.

The Laboratory Collection includes tests performed in the community. The exclusion of hospital and point-of-care tests will under-count testing and may affect results more in some DHBs than in others.

Analysis does not split by type of diabetes

There were some limitations as to what measures could be presented here. It was not possible to reliably split people by type of diabetes; hence the indicators represent a combination of those with type 1 and type 2 diabetes. Generally, most people with diabetes aged 0–24 years will have type 1 diabetes, while around 90 percent of those aged 25 years and over are expected to have type 2 diabetes.

The method used in the Virtual Diabetes Register to identify people with diabetes is less accurate at identifying children than adults with diabetes. A recently published survey highlights that local DHB data may be a better source for identifying prevalence in children: Jefferies C, Owens N, Wiltshire E for the Clinical Network for Children with Diabetes in New Zealand, on behalf of the Paediatric Society of New Zealand diabetes clinical network. 2015. Care for children and adolescents with diabetes in New Zealand District Health Boards: Is the clinical resourcing ready for the challenge? NZMJ. 128 (1424) 20-27.

Some important outcome indicators could not be included

Limitations of currently available data mean it was not possible to explore certain outcome indicators, including screening for diabetic retinopathy, retinopathy rates and end-stage renal failure. Outcome indicators, including myocardial infarction rates, stroke and other cardiovascular outcomes, are not included in this iteration of the Atlas, but are likely to be included in future updates. Users are encouraged to investigate local data in relation to these outcomes.

Note: Changes to the Virtual Diabetes Register

The Ministry of Health and sector representatives are working to improve the VDR algorithm. These analyses involve evaluating the algorithm and how it can be best used to identify people with diabetes. One revision to the algorithm has been made due to a technical error within the pharmaceutical data conditions. For this reason, the Atlas presents updated data for 2013 and 2014, as well the new data for 2015.

For more information, please visit the Ministry of Health website:


  1. New Zealand Guidelines Group. 2012. New Zealand Primary Care Handbook 2012. 3rd edition. Wellington: New Zealand Guidelines Group.

Recommended reading/Suggested links