HM Medical Clinic

 

Diabetes in cmdhb and northern region: estimation using routinely collected data

Diabetes in CMDHB and northern
region: Estimation using
routinely collected data
Counties Manukau District Health Board This report is available in pdf format on www.cmdhb.org.nz Information within the report may be freely used provided the source is acknowledged. Every effort has been made to ensure that the information in this report is correct. Counties Manukau District Health Board and the authors will not accept any responsibility for information which is incorrect, or any actions taken as a result of the information in this report. Published in May 2008 By Counties Manukau District Health Board Private Bag 94052 South Auckland Mail Centre ISBN 978-0-9582842-8-8 Key points

• The aim of this report was to provide a timely estimate of the prevalence of
diabetes in the populations of CMDHB and the northern region. This enables better understanding of the diabetes epidemic in CMDHB and the northern region more generally, and informs decisions around future diabetes service provision. • Retrospective, cross-sectional analyses used data from three different routinely collected administrative data sources to evaluate the magnitude of diabetes in CMDHB and three other district health boards (DHB's) in the northern region and to examine patterns of pharmaceutical, laboratory test and medical/surgical inpatient service utilisation • Almost 27,000 people in CMDHB were identified as diabetes cases in 2006-2007, out of a reconstructed population of around 427,000 people for CMDHB. Within the entire northern region (made up of CMDHB, Northland DHB, Auckland DHB and Waitemata DHB), 78,000 diabetes cases were identified in a reconstructed population of almost 1.4 million people • The age- and sex-standardised prevalence of diabetes in CMDHB was 7.1% in 2006-2007, compared with an age- and sex-standardised prevalence of 5.2% in the remaining three northern DHB's. The difference of 1.9% was statistically significant. Age- and sex-standardised prevalence estimates were similar using both the reconstructed population and census 2006 population estimates as denominators • Consistent with other studies demonstrating inequity in the distribution of diabetes by ethnicity in New Zealand, the prevalence of diabetes was highest in Maaori and Pacific CMDHB residents. Pacific women had the highest prevalence of diabetes of any group in CMDHB, with an age-standardised prevalence of 15.0%. Women of Other ethnicity had the lowest prevalence of any group, with an age-standardised prevalence of 4.0% • Of the 27,000 diabetes cases in CMDHB, 83% had at least two HbA1c monitoring tests in 2006-2007 (52% had four or more), 92% had at least one lipid monitoring test (and 81% had two or more), while 78% had claims for at least one urinary microalbumin test in the two-year period • Sixty-one percent of diabetes cases had regular subsidy claims in CMDHB for drugs that affect the renin-angiotensin system in 2006-2007, 64% had regular claims for lipid modifying agents, 18% had regular claims for insulin and 56% had regular claims for the drug metformin • 11,800 medical/surgical hospital discharges in 2007 had principal or secondary diagnosis codes for diabetes and this constituted 17% of discharges in CMDHB. Average length of stay for discharges with diabetes diagnosis codes was about 50% longer than those without diabetes codes. The hospital discharge rate for diabetes cases was 2.5 times that of the total reconstructed CMDHB population Table of contents


List of tables


List of figures


List of abbreviations

ACE
Albumin-creatinine Auckland District Health Board Australian Refined Diagnosis Related Group Counties Manukau District Health Board General Transaction Processing System ICD-10-AM International Classification of Diseases, 10th edition, Australian Ischaemic heart disease Northland District Health Board Northern DHB Support Agency New Zealand Index of Deprivation (2006) New Zealand Guidelines Group New Zealand Health Information Service New Zealand Health Survey Waitemata District Health Board Introduction
Diabetes is a common metabolic disorder, characterised by hyperglycaemia 1. It is
associated with a range of complications, including macrovascular disease (such as
coronary heart disease, stroke and peripheral vascular disease) and microvascular
disease (such as retinopathy [eye disease], kidney disease and peripheral vascular
disease) 2. Diabetes is a leading cause of morbidity and mortality in New Zealand 3 4.
In the New Zealand Health Strategy, diabetes is one of three disease priority areas and
one of 13 population health objectives 5.
Diabetes is common in the population served by Counties Manukau District Health
Board (CMDHB) and it places particular health burden on Maaori and Pacific
communities. The 2002/03 New Zealand Health Survey (NZHS) estimated the age-
standardised prevalence of self-reported diabetes in those aged ≥ 15 years in CMDHB
to be 5.0%, compared with a national self-reported prevalence of 4.1% 6. The
prevalence of diabetes was 9.5% in those of Maaori ethnicity and 7.2% in those of
Pacific ethnicities surveyed in the NZHS. Initial results from the 2006/07 NZHS
indicate that around 26,400 adults in CMDHB have diabetes (8.2% crude prevalence
for adults aged ≥ 15 years) 7. The Let's Beat Diabetes (LBD) 2006-2007 benchmark
survey of 2,520 people in CMDHB found an age-standardised, self-reported diabetes
prevalence of 7.0% in those aged ≥ 16 years 8. In this LBD study, the age-
standardised prevalence for Maaori was 6.2%, while for Pacific people it was 14.6%.
Aims and objectives
The aim of the analyses described in this report is to provide a timely estimate of the
prevalence of diabetes in CMDHB and three other northern district health boards
(DHB's) using routinely collected health care data. This will enable better
understanding of the diabetes epidemic in CMDHB and the northern region more
generally and inform decisions around future diabetes service provision.
The aim of this report is addressed through the following research objectives:
• Describe the prevalence of diabetes in CMDHB and northern region and describe
the diabetes population according to socio-demographic variables such as age, sex, ethnicity and deprivation • Explore and describe the laboratory monitoring of individuals identified as having diabetes (hereafter described as ‘diabetes cases') and identify any inequities (by ethnicity and deprivation) that may exist in laboratory monitoring • Examine and describe dispensing patterns amongst diabetes cases and identify any inequities in claims for subsidised medications, by ethnicity and deprivation • Review hospital service utilisation amongst diabetes cases in the study period Rationale for study
The Known Diabetes project undertaken in CMDHB towards the end of 2007
identified a ‘super set' of CMDHB residents with diabetes who had accessed services
such as hospital care, retinopathy screening and chronic care management (CCM) 9.
Around 23,000 people with diabetes had been identified within this database by May
2008 10. While the Known Diabetes database has an important role in understanding
health care utilisation within DHB services, it also has certain draw-backs. Cases in
the Known Diabetes database are collected from patient data that is up to ten years
old, meaning that a proportion of individuals in the database may no longer reside within CMDHB (‘residential churn'). Also, the Known Diabetes database gives only limited insight into care of diabetes in the community setting, outside the limits of formal DHB data collection. The data used in the current report is recent (covering calendar years 2006 and 2007) and has a community focus, utilising subsidy claims for community laboratory testing and claims for retail pharmaceutical dispensing. This allows a timely, ‘whole of community' perspective to analysis of diabetes in CMDHB and the northern region which is not available from other sources of routinely collected data. The reconstructed population
The reconstructed population is discussed further in the Methods section. Records of
subsidy claims for pharmaceuticals dispensed at community pharmacies were
combined with records of claims for laboratory investigations and with data from
hospital events in the National Minimum Data Set (NMDS) to create a ‘reconstructed'
set of data related to around 1,390,000 people in the four DHB's of the northern
region – Northland DHB (NDHB), Waitemata DHB (WDHB), Auckland DHB
(ADHB) and CMDHB. Individuals were included in the reconstructed population for
analysis if they had a health event recorded in the two years from January 2006 to
December 2007. Data was not available for those who did not have hospital events
recorded in NMDS, or did not have claims made for subsidised pharmaceuticals or
laboratory tests (with National Health Index [NHI] numbers documented) during the
study period. The reconstructed study population formed the denominator for most of
the analyses described in this report.
Comments on ethnicity and deprivation
A detailed summary of the requirements for collection, recording and output of
ethnicity data by the health and disability sector is provided by the Ministry of Health
document Ethnicity Data Protocols for the Health and Disability Sector 11. Each
individual in the reconstructed population had up to three ethnicity codes recorded at
Statistics New Zealand Level 2, consistent with recording requirements for the health
and disability sector. In this report, ethnicity is grouped into four categories – Maaori,
Pacific, Asian and Other – formerly Statistics New Zealand Level 0. Pacific ethnicity
includes all of the Polynesian and Melanesian Pacific ethnicities (but Fijian Indians).
South, East and Southeast Asian ethnicities are included in the Asian category. In
keeping with New Zealand health data conventions, Arab (and other Middle Eastern)
ethnicities, Afghani ethnicities and ethnicities of former Soviet Union countries are
not included in the Asian category. The Other category is mainly made up of
individuals of European ethnicities, although it does include all other ethnicities, for
example African and South American ethnicities.
This report presents ethnicity data by prioritised ethnicity, whereby individuals are
categorised into only one ethnic group, according to a prioritised schedule. The idea
behind this system is that there are instances where individuals need to be allocated to
only one ethnic group in analysis of socio-demographic data. Where this need exists
it is important to identify groups of policy importance and ensure that groups of small
size are not lost in amongst the dominant NZ European ethnic group. Consistent with
the standard prioritisation protocol recommended by the Ministry of Health 11,
ethnicity is prioritised in the following order: Maaori, Pacific, Asian, Other.
Domicile codes in the reconstructed population are linked to the Census Area Unit (CAU) associated with an individual case (rather than to the individual meshblock in which the case resides). Average NZDep2006 scores for the CAU in which the case resides are applied to each individual as a measure of deprivation. This is only a crude indicator of deprivation, as the score given to an individual is frequently the average for an area that may contain several thousand people (as opposed to meshblocks, which contain a median of 87 people). Such population-weighted scores will frequently disguise heterogeneity of deprivation within CAU's. Note on diabetes categorisation
Within the available data (except for NMDS), it was not possible to categorise
diabetes further into groups such as type I and type II diabetes (although it is
recognised that this distinction is important at an individual level). For practical
purposes therefore, all types of diabetes mellitus were aggregated together in the
single category ‘diabetes'.
In this report ‘diabetes' refers to diabetes mellitus in all its acknowledged forms.
Data
The ‘reconstructed population' referred to above, was composed of three primary sets
of routinely collected data – pharmaceutical claims data, laboratory claims data and
data from the National Minimum Data Set (NMDS). NMDS data extended back to
1990, while the pharmaceutical and laboratory claims data was collected over the 30
month period from July 2005 through to December 2007. A data collection for all
four DHB's in the northern region was created by selecting all individuals in the three
sets of routinely collected data who had health care events recorded in the two-year
period January 2006 to December 2007 (inclusive). In other words, individuals
appeared in the final reconstructed data set if they:
• Were dispensed a pharmaceutical product on the New Zealand Pharmaceutical Schedule 12 13 for which a reimbursement claim was made (and an NHI number was recorded for the claim - around 94% of laboratory and pharmaceutical claims had NHI numbers recorded) • Undertook a community laboratory investigation for which a reimbursement claim was made (and an NHI number was recorded), or • Appeared in NMDS through experience of a hospital event during the two Available mortality data was used to remove deceased people from the reconstructed population. Inclusion in the reconstructed population required documentation of an NHI number in claims records and residence within the geographic boundaries of the four northern DHB's at the time of the most recent recorded health event. Pharmaceutical reimbursement claims data was extracted from Pharmhouse, the national pharmaceutical subsidy data collection held by the New Zealand Health Information Service (HZHIS) and Pharmac. The Pharmhouse data warehouse contains claim and payment data from pharmacists for the dispensing of subsidised prescriptions that have been processed within the HealthPAC General Transaction Processing System (GTPS) 14. Pharmaceutical claims data for this analysis were obtained by the Regional Decision Support Team at NDSA (Northern DHB Support Agency) and passed on to CMDHB. Like the pharmaceutical claims data, laboratory claims data were also sourced from NZHIS, via NDSA. This data came from the Laboratory Claims Data Warehouse (Labs), rather than Pharmhouse. The purpose of the Labs database is to allow the Ministry of Health and DHB's to monitor primary care test subsidies 15. Labs also contains information from claims and payments processed by the HealthPAC GTPS 15. NMDS is a national collection of discharge information from public and private hospitals 16. NZHIS has provided CMDHB with NMDS data for the northern region. Analysis of discharge data in this report generally refers to medical and surgical inpatient discharges, rather than from other services such as psychiatric services. Finally, the NZHIS Mortality Collection is a complete set of national data, in which the underlying cause of death for all deaths registered in New Zealand is classified according to ICD-10-AM criteria 17. This data was used to remove deceased individuals from the reconstructed populations. Privacy
NHI key codes (also known as HCU codes) for all individuals in the reconstructed
population were encrypted by the Analytical Services team at NZHIS. The encryption
process is designed to maintain the anonymity of individuals within routinely collected data used for epidemiological analyses, by de-identifying unit record data 18. Only aggregated results are reported in this document and no contact with individuals was undertaken. Ethical approval for this analysis was therefore not required. Generalisability of estimates in CMDHB
The reconstructed CMDHB population contained 427,000 individuals, all of whom
had some form of contact with the health care system recorded in one or more of the
contributing data sets between January 2006 and December 2007. By way of
comparison, the official Counties Manukau population estimate for the March 2006
national census was 455,000. The missing 28,000 people (6.2% of census population)
probably consisted of individuals who either had no NHI numbers recorded in
pharmaceutical or laboratory claims data or who had no encounter with the health
system during the two years. The age, sex and ethnicity characteristics of the
reconstructed CMDHB population differed somewhat from those found in 2006
national census estimates. However, there was sufficient similarity between the two
populations for the analyses in this report to be generalised to the broader population.
The age distribution of the reconstructed CMDHB population followed a generally
similar trend to the census population, although notably fewer individuals in younger
age groups were identified in the reconstructed population (Figure 1). In the older age
groups (55 years or older), the two populations tracked reasonably closely together.
Figure 1: Age distribution of 2006-2007 reconstructed CMDHB population compared with age
distribution in 2006 official census estimate
ber 50000
Age group (years)

While the number of females in the reconstructed CMDHB population was similar to
that found in official estimates from the 2006 census, there were fewer males
identified in the reconstructed population (Figure 2).


Figure 2: Comparison of gender between reconstructed 2006-2007 CMDHB population and
official 2006 census estimate

Maaori, Pacific and Asian ethnicities were all under-represented in the reconstructed
CMDHB population, while those of Other ethnicity were slightly over-represented
(Figure 3), possibly due to differences in how individuals approach health services
compared with the census, or differences in how data is captured by the two systems.
Figure 3: Distribution of prioritised ethnicity in 2006-2007 CMDHB reconstructed population
compared with 2006 official census population estimate
er 150000
Decision rules
The following rules were used to identify individuals with diabetes in CMDHB and
the northern region (very similar rules were originally developed by Mr Craig Wright,
Senior Advisor (statistics and epidemiology), Public Health Intelligence, Ministry of
Health):
• Three or more HbA1c (or fructosamine) test claims within the two-year consecutive period January 2006 – December 2007 • Two or more community pharmaceutical dispensing claims for therapeutic group (TG) level 2 categories ‘Diabetes' and ‘Diabetes management' between January 2006 and December 2007 o For those aged < 25 years, only claims from category ‘Diabetes' were • Principal or secondary diagnosis of ICD-10-AM E10-E14 ‘Impaired glucose regulation and diabetes mellitus', or the codes O24.0 to O24.3 which cover pre-existing diabetes (type I, type II and unknown type) in pregnancy, from 1990 onwards (with health event of any kind identified in data in 2006-2007) • Any individual with the following AR-DRGs: K60A & B (Diabetes with and w/o catastrophic or severe complications), and K01Z (Diabetic foot procedures) , from 1990 onwards (with health event of any kind identified by DHB of residence in data in 2006-2007) Routinely collected administrative data sets have previously been used to understand quality of care at a population level 19. However, no examples were found in the published literature of decision rules which related to the identification of individuals with diabetes in routinely collected administrative data. Three methods were therefore used to validate the decision rules: • Literature review examining laboratory tests and diabetes products (in 2006 New Zealand Pharmaceutical Schedule) to establish the scope of use of identified tests and pharmaceuticals beyond their use in diabetes and the likely frequency of laboratory test monitoring in diabetes in the northern region • Consultation took place with seven experts in the areas of diabetes/ endocrinology, epidemiology, primary care and clinical coding to gain different perspectives on the suitability of the decision rules • Sensitivity analyses were undertaken to explore changes in rates of detection of diabetes cases with different laboratory testing frequencies and different frequency thresholds for pharmaceutical dispensing Detail on each of these three validation processes is available in the technical companion document that accompanies this report. Statistical analyses
Analyses presented in this report were undertaken using Microsoft Excel™, SPSS®
(version 13.0) and SAS®.
Unless otherwise stated, reconstructed populations for CMDHB and the northern
region constituted the denominators for the analyses in this report. Point estimates are
reported for descriptive statistics and 95% confidence intervals are included where
appropriate. In the case of calculating 95% confidence limits for proportions, an
assumption was made that the Normal approximation to the binomial distribution
applied in this data collection, as the number of cases in calculations was generally
large.
Standardisation for age and sex used Statistics New Zealand national population
estimates for 2006 and 2007. Estimates for these years by age and sex were averaged
to give the reference population for standardisation.
Results
In total, 26,961 diabetes cases were identified using the decision rules within the
reconstructed CMDHB population of 427, 404 people. Absolute numbers for
identified diabetes cases are presented in terms of age, sex and prioritised ethnicity in
Table 1.
Table 1 Age in years, sex and ethnicity of diabetes cases identified within the reconstructed
CMDHB population for 2006-2007
By way of comparison, 8,448 diabetes cases were identified within the reconstructed Northland DHB population of 145,250 people, 21,056 diabetes cases were identified in the reconstructed Waitemata DHB population of 439,628 people and 21,679 diabetes cases were found in the 381,458 individuals who made up the Auckland DHB reconstructed population. Results of analyses of identified diabetes cases in CMDHB and the northern region are presented in this section. Initially, crude prevalence and age- and sex-standardised prevalence of diabetes in CMDHB and in each of the other three DHB's in the northern region are described. Age, sex, ethnicity and deprivation distributions are explored and compared. Utilisation of laboratory tests and pharmaceuticals is then described. Finally, a review of hospital service utilisation by diabetes cases in NMDS is undertaken. Crude prevalence estimates
The crude prevalence of diabetes cases (i.e. people identified within the reconstructed
population using the decision rules) for all of CMDHB was 6.3% (95% CI 6.2% to
6.4%), as seen in Table 2. This was the highest crude prevalence of any of the four
DHB's in the northern region.
Table 2: Crude prevalence of diabetes in reconstructed northern DHB populations in 2006-2007
95% CI for adult prevalence (%) prevalence (%) prevalence (%) prevalence (%)
In comparison, average estimated DHB population numbers for the calendar years
2006-2007 (based on the March 2006 national census estimate for each DHB) were
used as a denominator for further calculation of crude prevalence (Table 3). The
prevalence estimates were similar to those calculated using the reconstructed
populations, although in all cases were lower due to the larger DHB populations
found in census estimates.
Table 3: Crude prevalence of diabetes in northern DHB populations using identified diabetes
cases and averaged census denominator of 2006-2007 calendar years
95% CI for adult prevalence (%) prevalence (%) prevalence (%) prevalence (%) Across the three other northern DHB's (using the reconstructed denominator) the crude prevalence of diabetes was 5.3%, a statistically significant 1.1% lower than the estimate found in CMDHB for 2006-2007 (χ2 = 572.9, 1 df, P < 0.001) in this crude analysis. Standardised prevalence estimates
Table 4 describes the prevalence of diabetes within the reconstructed CMDHB
population for 2006 and 2007, with results stratified by age (in years), sex and
prioritised ethnicity. A high prevalence of diabetes is noted in those of Maaori,
Pacific and Asian ethnicities, especially in those aged 45 years or more.
Table 4: Prevalence of diabetes in CMDHB reconstructed 2006-2007 population, stratified by
age (in years), sex and ethnicity

By the age of 55 years, 40% of all Pacific people in CMDHB have diabetes, by age 65
years nearly half of all Pacific people have diabetes. Note also that the category
‘Asian' contains much heterogeneity. For example, diabetes prevalence in individuals
of Indian ethnicity has been noted at more than three times that of Chinese people in
CMDHB and both of these ethnic groups contribute to the overall ‘Asian' category 20.
An average New Zealand population for the years 2006 and 2007 (using estimates
developed from the March 2006 national census) was used to standardise the
prevalence of diabetes in the four northern DHB's by age and sex, where the
reconstructed population was used as denominator. The age- and sex-standardised
prevalence of diabetes in CMDHB was found to be 7.1% (95% CI 7.0% to 7.2%), the
highest of any of the four DHB's in the northern region (Table 5).
Table 5: Age- and sex-standardised prevalence of diabetes in four DHB's of northern region,
using 2006-2007 reconstructed population denominator
Adult (15+ years) 95% CI for adult The age- and sex-standardised prevalence for the three other northern DHB's was 5.2% (95% CI 5.1% to 5.2%). The difference in prevalence of 1.9% between CMDHB and the other three northern DHB's was statistically significant (χ2 = 2086.3, 1 df, P < 0.001). By way of comparison, population estimates from the 2006 national census for CMDHB and the other three northern DHB's were used as denominators in the standardisation process, replacing the reconstructed population denominators. An age- and sex-standardised prevalence of 7.0 (95% CI 7.0 to 7.1) was found in CMDHB when census 2006 CMDHB population estimates were used for the denominator groups in direct standardisation (Table 6). This was very close to the estimated prevalence of 7.1% found using the reconstructed denominator. This similarity appears to be due to the better coverage of reconstructed population
estimates in age groups that have the greatest numbers of diabetes cases.
Table 6: Age- and sex-standardised prevalence of diabetes in four DHB's of northern region,
using census 2006 estimate denominator
Adult (15+ years) 95% CI for adult
Age-standardised prevalence estimates for each ethnicity within CMDHB are found in
Table 7. A marked disparity is noted between the prevalence estimates for Maaori,
Pacific and Asian, and the estimates for those of Other ethnicity.

Table 7: Age-standardised prevalence of diabetes in CMDHB by ethnicity and sex, 2006-2007
Social and demographic characteristics
Age distribution
The age distribution for those in the CMDHB population identified as diabetes cases
is compared against that for the whole reconstructed CMDHB population in Figure 4.
Diabetes cases have an age distribution that is weighted towards the older end of the
age spectrum. The peak of the distribution is found in the 55-64 year age group. In
contrast, the age distribution for the whole reconstructed CMDHB population is
dominated by younger age groups, with the four groups made up by those aged 55
years or older contributing to only a small proportion of the population.
Figure 4: Age distributions of diabetes cases in CMDHB and of the entire reconstructed
CMDHB population for 2006-2007
tage 15%
ercen 10%
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group (years)

The difference between the age distribution of diabetes cases and that of the broader
reconstructed CMDHB population is consistent with the difference found between age
distributions for the other three northern DHB's (not including CMDHB) and diabetes
cases within those DHB's, as seen in Figure 5. Again, those identified as having
diabetes are spread towards the older end of the age spectrum, with the peak of the
distribution for diabetes cases again in the 55-74 years age group.
Figure 5: Age distributions of diabetes cases in the northern region (ex. CMDHB) and of the
reconstructed northern region (ex. CMDHB) population for 2006-2007
tage
en 15%
erc 10%
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group (years)
Nothern DHBs (ex CMDHB) Sex
As can be seen in Figure 6, the proportion of females amongst diabetes cases is only
slightly less than the proportion of females in the reconstructed CMDHB population.
This trend is consistent across all four of the northern DHB's, where the proportions
of females amongst diabetes cases are fairly similar to the proportions of females in
the reconstructed DHB populations.
Figure 6: Proportion of females in diabetes cases identified within each DHB, compared with
proportion of females in reconstructed DHB populations for 2006-2007
Proportion of each group who are female
Reconst DHB populations Distribution of ethnicity
As mentioned in the introductory section, prioritised ethnicity is used in this report.
While prioritisation greatly simplifies the analysis and presentation of data for
individuals who have recorded more than one ethnicity, it is not without limitation. It
is contradictory to the concept of self-identification in ethnicity, some groups become
over-represented at the expense of others and it places people in specific ethnic groups
which although simplifying results may also introduce bias. These limitations should
be acknowledged when interpreting the data presented below.
The composition of ethnicity for diabetes cases in the northern DHB's follows the
general pattern of ethnicity distribution within the reconstructed populations of each
of the four DHB's (Figure 7). Those of Pacific, Maaori and Asian ethnicities tended
to be over-represented in the diabetes groups for all four DHB's, while the proportion
of people of Other ethnicity in reconstructed populations for all four DHB's exceeds
their proportional representation in diabetes groups in all cases.

Figure 7: Ethnic composition of the diabetes case group and reconstructed 2006-2007 population
within each of the four northern DHB's
Deprivation
As discussed in the introductory section of this report, level of deprivation for
individuals within the reconstructed data set is approximated by matching individual
domicile codes with average NZDep2006 deciles by census area unit (CAU). Figure
8 compares the proportion of diabetes cases in each NZDep2006 decile against the
approximated distribution of deprivation for the reconstructed CMDHB population.
Likewise, Figure 9 presents the same comparison for the wider northern region (ex.
CMDHB).
Figure 8: Mean NZDep2006 decile of CAU for CMDHB residents identified as diabetes cases
compared with CMDHB residents in 2006-2007 reconstructed population
Mean NZDep decile for CAU
Reconst CMDHB population
Figure 9: Mean NZDep2006 decile of CAU for northern region (ex. CMDHB) residents
identified as diabetes cases compared with general northern regional reconstructed population in
2006-2007
Mean NZDep decile for CAU
Reconst northern population In both Figures 8 and 9 the diabetes case populations are characterised by a shift in the distribution of deprivation towards the lower NZDep2006 deciles in comparison to the reconstructed populations within CMDHB and the northern region (ex. CMDHB) respectively. Laboratory utilisation
Electrolytes, glucose and creatinine
Within the reconstructed CMDHB population, 79.0% of diabetes cases were found to
have had at least one test for ‘BE3 – Sodium and potassium, serum' and 92.4% had at
least one test for ‘BR1 – Creatinine, serum' in the community during the two-year
study period. These results were very similar to those found in the (ex. CMDHB)
northern region. 82.3% of identified CMDHB diabetes cases had the test ‘BG5 –
Serum glucose' in the study period, compared with 83.4% of diabetes cases in the
remaining northern region. Figure 10 shows the frequency of electrolyte testing
amongst CMDHB diabetes cases over the two-year period, while Figure 11 shows
frequency of testing for serum creatinine in the diabetes group.

Figure 10: Frequency of electrolyte testing amongst diabetes cases in CMDHB reconstructed
population, 2006-2007
Five or more tests Figure 11: Frequency of serum creatinine testing amongst diabetes cases in CMDHB
reconstructed population, 2006-2007
Five or more tests Glycosylated haemoglobin (HbA1c)
Frequency of glycosylated haemoglobin (HbA1c) testing was included in the decision
rules for identification of individuals with diabetes in the reconstructed population
(methods section). Sensitivity analysis was performed around the frequency of
HbA1c testing within the decision rules. Around 50,000 people in the reconstructed
CMDHB population had a laboratory claim for at least one HbA1c test during the
two-year period, 25,000 people had two or more HbA1c claims during the study
period and 18,000 people satisfied the inclusion criteria of three or more HbA1c tests
during the two-year period. About 4,000 diabetes cases (out of the 18,000 with three
or more tests) were exclusively identified by way of HbA1c testing (and not by
pharmaceutical or NMDS decision rules).
Across the four northern DHB's for the 2006-2007 period, over 6,000 diabetes cases
identified using pharmaceutical and NMDS decision rules did not have subsidy claims
for HbA1c testing. Figure 12 describes the cumulative frequency of HbA1c testing
across the four northern DHB's for the remaining 70,000 or so diabetes cases in 2006-
2007.
Figure 12: Cumulative frequency of HbA1c testing over two-year period in diabetes cases in four
northern DHB's, 2006-2007
ber
m 15000
Nu 10000
Cumulative number of HbA1c tests
The Get Checked diabetes programme entitles all people with diabetes in New
Zealand to a free annual GP or GP practice nurse review, including measurement of
HbA1c. In the CMDHB population of diabetes cases identified using all of the
decision rules, 83.3% of cases had two or more HbA1c tests completed in the
communitypared with 82.1% for the
other three northern DHB's across the same period. Figure 13 describes the
proportion of diabetes cases within each DHB who had HbA1c performed at least
twice in the community during the two year period, by ethnicity. While these results
do not suggest that the HbA1c tests detected in laboratory claims data necessarily
related to annual Get Checked reviews, they do indicate that HbA1c testing was
reasonably frequent in the majority of diabetes cases.

Figure 13: Proportion of diabetes cases with two or more HbA1c tests over two-year period,
2006-2007, by ethnicity and DHB (northern region)
Within the reconstructed CMDHB population, a significantly greater proportion of diabetes cases of Other ethnicity were found to have had two or more HbA1c tests within the two-year period (χ2 = 27.1, 1 df, P < 0.001) in univariate analysis. The proportion of individuals of Other ethnicity who had two or more HbA1c tests was 84.7% (95% CI 84.0% to 85.3%), while 82.2% of those of Maaori, Pacific or Asian ethnicities had two or more HbA1c tests during the period (95% CI 81.6% to 82.8%). Current recommendations suggest that individuals with diabetes should have HbA1c monitored three-to-six monthly, depending on stability of glycaemic control 21-24. This frequency of testing would roughly approximate to four or more tests in the two-year period (although it is recognised that some diabetes cases may have had HbA1c tests performed in hospital laboratories and that four, evenly-spaced HbA1c 1 Not including testing undertaken in hospital monitoring tests would not have been performed in many cases). Overall in CMDHB,
just over half (52.3%) of diabetes cases had four or more HbA1c tests during the two-
year period. This was a significantly greater proportion than the 43.9% found in the
other three northern DHB's in univariate analysis (χ2 = 485.3, 1 df, P < 0.001).
Within CMDHB, only 47.5% of diabetes cases of Asian ethnicity had four or more
HbA1c tests in two years, compared with over 50% in those of Maaori, Pacific and
Other ethnicity (Figure 14).
The reasons behind higher HbA1c test frequency in CMDHB compared with the rest
of the northern region are not clear. The higher frequency could be due to closer
monitoring of individuals with diabetes (though initiatives such as CCM).
Alternatively, more people with diabetes in CMDHB may have glycaemia that is
difficult to control, requiring more frequent monitoring.

Figure 14: Proportion of diabetes cases in CMDHB that had ≥ 4 HbA1c tests in two years, by
ethnicity, 2006-2007

Crude analysis of frequency of HbA1c testing by deprivation shows that the
proportion of diabetes cases in CMDHB who had either two or more HbA1c tests or
four or more HbA1c tests in the two-year period was fairly consistent across the
spectrum of deprivation, as measured by mean NZDep2006 decile of a resident's
CAU (Figure 15). There was no evidence of any gender difference in the frequency
of HbA1c testing amongst diabetes cases in the reconstructed CMDHB population or
the northern region more generally.
Figure 15: Mean NZDep2006 deprivation decile in CAU by frequency of HbA1c testing in
CMDHB diabetes cases, 2006-2007
The lack of correlation between mean NZDep2006 by CAU and HbA1c test frequency was surprising, given that there is usually a socio-economic gradient attached to health care access 25 26. This may indicate effective targeting of initiatives such as CCM. Lipid studies
The NZGGends that cardiovascular risk assessment (including fasting lipid
studies) be performed annually on all people with diabetes from the time of diagnosis
23. This advice is consistent with international recommendations from groups such as
the ADA21.
In diabetes cases identified in the CMDHB reconstructed population for 2006-2007,
91.7% (95% CI 91.4% to 92.1%) had at least one claim for the test ‘BL4 – Fasting
lipid group', and 80.6% (95% CI 80.1% to 81.2%) had claims for at least two fasting
lipid tests during this two-year period. This is higher than 89.8% (95% CI 89.6% to
90.1%) for one lipid test or more and 75.3% (95% CI74.9% to 75.7%) for two or
more lipid tests found in the remaining three DHB's in the northern region.
Slightly fewer Maaori were found to have had either one (or more) or two (or more)
fasting lipid test claims in the two-year period (Table 8). The differences observed
were not statistically significant.
Table 8: Proportion of diabetes cases in CMDHB reconstructed population in 2006-2007 who
had claims for at least either one or two fasting lipid tests
One or more lipid tests Two or more lipid tests
Figure 16: Mean NZDep2006 deprivation decile in CAU by frequency of fasting lipid testing in
CMDHB diabetes cases, 2006-2007
2 New Zealand Guidelines Group 3 American Diabetes Association No significant gender differences in fasting lipid testing were identified either within the CMDHB diabetes cases or diabetes cases identified in reconstructed populations of the remaining three northern DHB's. Likewise, no significant differences in frequency of fasting lipid testing were noted between different groups by level of deprivation in CMDHB (Figure 16). Microalbumin
Urinary microalbumin is a screening test used in people with diabetes to detect early
nephropathy (at a point where it is reversible with good blood pressure control and
management of hyperglycaemia) 27. Results for this test are often expressed in the
form of an albumin-creatinine ration (ACR) 28. In people with diabetes, no
microalbuminuria and normal serum creatinine, the NZGG recommends that urinary
microalbumin be tested annually 23. This recommendation is consistent with
international guidelines generated by organisations such as the ADA 21.
Of the diabetes cases identified in the reconstructed CMDHB population, 77.7% (95%
CI 77.2% to 78.2%) had one claim or more for test ‘BP8 – Microalbumin, early
morning urine', leaving almost 6,000 cases with no claim for microalbumin during the
two-year study period. By comparison, only 68.8% (95% CI 68.4% to 69.2%) of
diabetes cases in the remaining three northern DHB's had one or more claims for
urinary microalbumin during the two years. This difference of almost 9% was
statistically significant in univariate analysis (χ2 = 512.8, 1 df, P < 0.001). When the
threshold was lifted to two or more microalbumin tests within the two-years, 61.8%
(95% CI 61.3% to 62.4%) of CMDHB diabetes cases had at least two such tests while
only 49.2% (95% CI 48.8% to 49.6%) of diabetes cases in the remaining three DHB's
were identified, a statistically significant difference of almost 13% (χ2 = 1098.3, 1 df,
P < 0.001).
The distribution of testing frequency for urinary microalbumin by ethnicity in
CMDHB diabetes cases is presented in Figure 17, while the distribution by
deprivation level is shown in Figure 18.
Figure 17: Proportion of diabetes cases in CMDHB with either 1+ or 2+ urinary microalbumin
screening tests in two years (2006-2007), by ethnicity
One or more tests Two or more tests Figure 18: Mean NZDep2006 deprivation decile in CAU by frequency of urinary microalbumin
testing in CMDHB diabetes cases, 2006-2007
A proportion of diabetes cases have no requirement for microalbumin as a screening tool (as they have known renal disease) and it was difficult to identify these people within the reconstructed populations, given the limitations of the available data. Some insight into this population can be gained by removing individuals with documented ICD-10-AM diagnosis and procedure codes for renal disease in NMDS from the study population. In CMDHB, 12.4% of diabetes cases had either an ICD-10-AM diagnosis code or a procedure code related to renal disease, compared with 11.2% in the remaining three northern DHB's. For the whole northern region (including CMDHB), 9,103 diabetes cases were identified who also had documented ICD-10-AM diagnosis and procedure codes relevant to renal disease. This left a remainder of 23,620 diabetes cases who did not have evidence of renal disease documented in NMDS in the reconstructed CMDHB population and 45,435 diabetes cases without documented renal disease in the remaining northern DHB's. Frequency of microalbumin screening was in fact lower in diabetes cases without documented evidence of renal disease in NMDS. Of CMDHB diabetes cases without documented renal disease, 76.2% (95% CI 75.6% to 76.7%) had claims for at least one microalbumin test in two years and 59.7% (95% CI 59.1% to 60.4%) had claims for at least two tests. In the other three northern DHB's, 66.9% (95% CI 66.5% to 67.4%) of diabetes cases without renal disease had claims for one or more microalbumin test, while 47.4% (95% CI 46.9% to 47.8%) had claims for at least two tests. The reason for the lower frequency of microalbumin screening in those without renal disease is not easily explained in the data. Perhaps this test is also being used for monitoring progression of renal disease in those already recognised as having nephropathy. Cost
The total cost of laboratory tests in the reconstructed population for the whole
northern region (excluding GST) in 2007 was just over $86 million, of which diabetes
cases accounted for around $15 million (18%). Within CMDHB, the total cost of
laboratory claims in the reconstructed population for 2007 was almost $25 million, of
which $5.5 million was attributed to diabetes cases, i.e. 6.3% of the population used
almost one quarter of the laboratory costs for this period.

The unadjusted cost of community laboratory testing claims per person in the
reconstructed population is presented in Table 9 for diabetes cases and those without
diabetes, by DHB in 2007.
Table 9: Claims for community laboratory testing by DHB, per person for diabetes cases and
those without diabetes, by DHB in 2007 (excl. GST)

District Health Board
Cost/person ($) – Cost/person ($) – no Counties Manukau DHB
While CMDHB spent the about the same amount as the other three DHB's on
community laboratory testing per (average) resident overall, it spent the most (on
average) on diabetes cases. The peak age group for community laboratory claims (in
absolute terms) was the 55-64 year age group, while the proportional contribution of
diabetes cases to total laboratory claims increased steadily until the 65-74 year age
group and then declined slightly amongst those aged 75 years or more (Figure 19).
Figure 19: Distribution of dollar value of laboratory claims by age group for diabetes cases and
those without diabetes in reconstructed CMDHB population, 2007
'000) 2500
st ($ 2000
Co 1500
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group
Pharmaceutical utilisation
As was the case with decision rules for identification of diabetes cases, the decision
rule for identification of pharmaceutical utilisation in the following analyses consisted
of two or more dispensing claims for a drug category within calendar years 2006 and
2007.
The analysis looked at claims for medications considered relevant to the diabetes
population. It used the therapeutic groupings for pharmaceutical claims found in the
New Zealand Pharmaceutical Schedule at around the mid-point of the analysis period,
in August 2006 12. The specific therapeutic groupings analysed are found in Table 10.
Table 10: Specific therapeutic group categories and chemical names of medications for which
analysis of pharmaceutical utilisation was undertaken
TG Level 2
TG Level 3
Insulin - short-, intermediate-, long-acting preparations and rapid acting insulin analogues Agents affecting renin- angiotensin system Angiotensin II antagonists AntidepressantsLipid modifying agents HMG Co A reductase inhibitors Beta adrenoceptor blockersCalcium channel blockersThiazide and related diuretics Antiplatelet agents Diabetes medications
The therapeutic group (TG) level 2 category ‘Diabetes' contained 15 TG level 3
classifications which in turn contained a variety of different chemical preparations 12.
In this analysis, short-, intermediate-, and long-acting insulin preparations, plus rapid-
acting insulin analogues were aggregated into the single category ‘insulin'. Oral
diabetes agents were analysed at the level of chemical name.
Almost 5,000 diabetes cases had two or more claims for insulin in 2006 and 2007 in
CMDHB, 18.0% of the diabetes population. In the remaining three northern DHB's,
17.5% of diabetes cases had claims for regular insulin dispensing. These figures
appear to be consistent with the frequency of insulin use amongst people with diabetes
nationally. In the 2006/07 NZHS, 19.4% of adults with diabetes reported daily insulin
injections (with or without concurrent oral diabetes agents) 7. The distribution of
insulin claims in diabetes cases by ethnicity is given in Figure 24. Notably, the
proportion of diabetes cases of Asian ethnicity with regular claims for insulin was
much lower than those of other ethnicities. Proportionally more Maaori diabetes
cases in CMDHB had claims for insulin compared to the remainder of the region,
while the opposite was found for diabetes cases of Pacific ethnicity in CMDHB.
Note that some insulin prescriptions and monitoring equipment for diabetes cases in
CMDHB were obtained from Diabetes New Zealand in Oamaru, rather than from
local pharmacies. Because individuals are identified by residential address in this data
set, northern region diabetes cases who purchased diabetes supplies in Oamaru have
been accounted for according to their DHB of residence.
Figure 20: Proportion of diabetes cases with two or more claims for insulin in two years in
CMDHB and northern region, by ethnicity
Northern region (ex. CMDHB)
No significant gender differences in claims for insulin were noted either amongst
diabetes cases in CMDHB or those in the other three DHB's. The age distribution for
diabetes cases with two or more insulin prescription claims in 2006-2007 is presented
in Figure 25. In keeping with the distribution of diabetes cases more generally, the
peak of insulin claims amongst northern region diabetes cases is in the 55-64 year age
group.
Figure 21: Absolute numbers of diabetes cases with regular insulin claims in CMDHB and
northern region by age group
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group
Northern region (ex. CMDHB) When proportion of diabetes cases with regular prescription claims for insulin was compared across the deprivation spectrum and between DHB's, no observable difference was found in claim patterns. Around 16% to 18% of diabetes cases in each deprivation decile had regular claims for insulin and this pattern was similar between CMDHB and the remaining three DHB's in the northern region. Oral diabetes medications are an important part of the management of hyperglycaemia in individuals with type 2 diabetes 29. In the 2006/07 NZHS, 52.4% of adults with diabetes in the national sample reported taking only oral diabetes medication 7. In CMDHB, 56.1% of diabetes cases had community pharmaceutical claims for two or more metformin scripts in two years and over 80% of these people were aged 45 years or more. In comparison, only 46.7% of diabetes cases in the other three northern DHB's had two or more claims for metformin. The proportion of diabetes cases with
claims for metformin was highest amongst those of Pacific ethnicity and lowest
amongst those of Other ethnicity (Figure 26).
Figure 22: Proportion of diabetes cases with two or more claims for metformin in two years in
CMDHB and northern region, by ethnicity
Northern region (ex. CMDHB)
An association was noted between level of deprivation and proportion of diabetes
cases with two or more claims for metformin, with proportion of metformin claimants
increasing with greater average NZDep2006 decile for CAU (Figure 27). The
regression coefficient for this trend in the northern region was statistically significant
in simple (unadjusted) linear regression (t = 24.5, 8 df, P < 0.001).
Figure 23: Proportion of diabetes cases with two or more claims for metformin in two years by
mean NZDep2006 decile for area of residence
Northern region (ex. CMDHB) Pharmaceutical claims for sulfonylurea oral hypoglycaemic agents were common in both CMDHB and in the remainder of the northern region in 2006-2007. Table 13 documents the proportion of diabetes cases in CMDHB and in the other three northern DHB's who had two or more claims for sulfonylurea agents during these two years. Table 11: Proportion of diabetes cases in CMDHB and three other DHB's in northern region
with 2+ sulfonylurea claims in two years
Proportion of CMDHB Proportion of northern region diabetes cases Pioglitazone is a drug in the thiazolidinedione class which increases the sensitivity of fat, muscle and liver to insulin 29. It is used infrequently in New Zealand in combination with other diabetes medications or as monotherapy where other oral agents are not tolerated. It requires special authority approval for reimbursement in New Zealand 12 13. Around 1,500 diabetes cases in the northern region were regularly prescribed pioglitazone in 2006 and 2007, of which 450 people were CMDHB residents. Likewise, the drug acarbose is an alpha-glucosidase inhibitor which reduces the digestion of complex sugars in the intestine. It too is used rarely in New Zealand and requires special authority approval for reimbursement. Only 185 diabetes cases in CMDHB (less than one percent of diabetes cases) had two or more prescription claims for acarbose during 2006 and 2007, and only 85 diabetes cases in the other three northern DHB's had claims for acarbose during this period. Blood pressure-lowering agents
The NZGG recommends annual cardiovascular risk assessment, including blood
pressure monitoring for people with type 2 diabetes 23. Hypertension is common in
individuals with diabetes 30. The prevalence of hypertension amongst adults with type
2 diabetes is typically between 40% and 60% internationally, with risk of
hypertension increasing with age 31 32. Appropriate management of blood pressure
can significantly reduce the risk of cardiovascular morbidity and mortality
experienced by a person with diabetes 32 33. Management of hypertension in people
with diabetes includes lifestyle interventions and blood pressure-lowering medication
23. In CMDHB, 69.7% of diabetes cases had two or more prescription claims for blood pressure-lowering agents during the study period, compared with 66.1% of diabetes cases in the other three northern DHB's. Given the limitations of the data, it is not possible to evaluate the appropriateness of blood pressure medication prescribing amongst diabetes cases in the northern region. However, differences in utilisation of blood pressure-lowering medications between regions, by ethnicity and by deprivation can be described amongst diabetes cases and can provide insight into prescribing patterns. Drugs that affect the renin-angiotensin system, including ACEangiotensin II antagonists, have a number of advantages over other blood pressure-lowering agents in diabetes, including favourable toxicity profiles, no adverse effects on lipid metabolism and protective effects on the progression of nephropathy 34 35. These agents are used as first-line therapy in people with diabetes and 4 Angiotensin converting enzyme microalbuminuria or overt nephropathy 23. Within CMDHB, 61.1% of diabetes cases
had pharmaceutical claims for agents that affect the renin-angiotensin system, while
55.4% of diabetes cases in the other three DHB's in the northern region had two or
more claims for such agents in 2006 and 2007. The majority (77.9%) of claims by
CMDHB diabetes cases in this category were for ACE inhibitors, with angiotensin II
antagonists accounting for 10.7% of claims. Claims for these agents were
inconsistent across ethnicity categories in all four northern DHB's, with diabetes
cases of Asian ethnicity notable in particular for lower proportions of claims for these
agents (Figure 20). Further analysis is required to explore whether the lower
proportion of claims for those of Asian ethnicity was due an age effect, as found for
blood pressure medication use in the 2006/07 NZHS 7. The distribution of claims for
renin-angiotensin agents amongst diabetes cases by deprivation (average NZDep2006
decile for CAU) showed a relatively even spread across all ten deciles, with higher
utilisation levels for CMDHB diabetes cases within each stratum, except for decile
seven (Figure 21). Claim patterns were similar between genders, with dispensing
claims sightly more common amongst males in all four northern DHB's.
Figure 24: Proportion of diabetes cases with two or more claims for agents that affect renin-
angiotensin system in two years (2006-2007)
Northern region (ex. CMDHB)
Figure 25: Percentage of diabetes cases with two or more claims for renin-angiotensin agents by
deprivation level for CMDHB and three other northern DHB's
Northern region (ex. CMDHB) Table 12 shows the proportion of identified diabetes cases within CMDHB and the
other remaining northern region DHB's who have two or more claims for blood
pressure-lowering medication in 2006 and 2007. Drugs that affect the renin-
angiotensin system are the most commonly dispensed group of any pharmacological
blood pressure-lowering agent and more than half of diabetes cases in the northern
region have at least two claims for this therapeutic grouping. By way of reference,
13.4% of adults (with diabetes and without) in the 2006/07 NZHS in CMDHB
reported that they were currently taking medication for high blood pressure 7.
Table 12: Proportion of diabetes cases in CMDHB and three other DHB's in northern region
with claims for blood pressure-lowering medication
Proportion of CMDHB Proportion of northern Agents affecting renin-angiotensin Angiotensin II antagonists Beta adrenoceptor blockers Calcium channel blockers Thiazide diuretics Lipid modifying agents
Dyslipidaemia is common in people with diabetes and is an important contributor to
cardiovascular disease in this group 36. Interventions to manage dyslipidaemia
include lifestyle modifications (e.g. weight loss and reduction in intake of cholesterol,
saturated and ‘trans' fats) and medications such as HMG CoA reductase inhibitors
(statins) 37 38. Within the New Zealand pharmaceutical Schedule (August 2006), the
therapeutic group level 2 category ‘lipid modifying agents' contained agents of
several classes, including fibrates, statins, resins and cholesterol absorption inhibitors
12. The entire therapeutic group ‘lipid modifying agents' has been examined, together
with statins, the most commonly prescribed of these agents (Table 13).
Table 13: Proportion of diabetes cases in CMDHB and northern region with two or more claims
for lipid modifying agents in 2006-2007
Lipid modifying agents HMG CoA reductase inhibitors (statins) There is no easy way to recognise which diabetes cases in the reconstructed population ‘should' have been taking lipid modifying agents. Suitability for these medications is dependent on individual clinical profiles. However, recommendations for the use of lipid modifying agents in clinical guidelines suggest that a high proportion of individuals with diabetes are likely to benefit from their use 3 21. The finding that CMDHB had the highest proportion of diabetes cases with claims for these agents therefore seems positive on face value. Further research looking at the appropriateness of statin prescribing in diabetes cases would add further perspective to this finding.
As with ACE inhibitors and angiotensin II antagonists, the statins group (which
contained the drugs atorvastatin and simvastatin) was analysed further to look at
utilisation by ethnicity and deprivation (Figures 22 and 23). A greater proportion of
diabetes cases in CMDHB had claims for two or more statin prescriptions compared
with the other three northern DHB's, across all ethnic groups and at all levels of
deprivation.
Figure 26: Proportion of diabetes cases with two or more claims for statins in two years (2006-
2007) in CMDHB and northern region
Norther region (ex. CMDHB)
Figure 27: Percentage of diabetes cases with two or more claims for lipid modifying agents by
deprivation level for CMDHB and three other northern DHB's
Northern region (ex. CMDHB) Almost 90% of pharmaceutical claims for statins amongst CMDHB diabetes cases were in adults aged over 45 years (as with the remainder of the northern region). Consistent with the slightly greater number of males identified amongst diabetes cases, 52.2% of diabetes cases with claims for statins were male. Other medications
Depression is more common in individuals with diabetes than in the general
population 39 and micro- and macrovascular complications of diabetes are frequently
associated with the presence of depression 23. There are many methods for the
treatment of depression; antidepressant medication being just one of those methods. It
is unclear what an appropriate level of antidepressant prescribing in diabetes cases
would be. Notwithstanding, 11.7% of diabetes cases in CMDHB had two or more
pharmaceutical claims for antidepressant medication during the two-year period,
compared with 14.7% of diabetes cases in the remaining northern DHB's.
The TG level 3 category ‘antiplatelet agents' contains the drugs aspirin and
dipyridamole (Persantin) 12. Within CMDHB, 46.2% of diabetes cases had two or
more prescription claims for antiplatelet agents, while 40.7% of diabetes cases in the
other three northern DHB's had claims for these agents.
Cost
The total value of community pharmaceutical reimbursement claimsGST) in the reconstructed population for the whole northern region in 2007 was $310
million, of which diabetes cases accounted for about one quarter of that total at $78
million. Within CMDHB, the total cost of pharmaceutical reimbursement claims in
the reconstructed population in 2007 was $92 million, of which $28 million was
attributed to diabetes cases, i.e. 6.3% of the population accounted for more than 30%
of community pharmaceutical reimbursement costs during 2007.
The unadjusted cost of community pharmaceutical claims per person is presented in
Table 14 for both individuals without diabetes in the reconstructed DHB populations
and diabetes cases for each DHB.
Table 14: Unadjusted cost of community pharmaceutical subsidy claims by DHB, per person, for
reconstructed DHB populations and diabetes cases within those populations in 2007 (excl. GST)

District Health Board
Cost/person ($) – Cost/person ($) – no Counties Manukau DHB It is not clear what pharmaceutical costs for people with diabetes ‘should be'. Comparatively, the mean 2007 cost of community pharmaceutical claims for individual diabetes cases in CMDHB was similar to that found in the other three northern DHB's. While it is difficult to compare pharmaceutical costs between countries, a recent report indicated that the average net ingredient cost (NIC)diabetes agents prescribed in primary care (insulin, monitoring agents and oral diabetes agents) in the 1.9 million people with registered diabetes in the United Kingdom in 2006 was around £300 ($750). The average drug cost (as opposed to reimbursement value) for all community pharmaceuticals prescribed to diabetes cases 5 Reimbursement value is the basic drug cost, less patient co-pay, plus pharmacy fee 6 NIC is the basic drug cost, not including dispensing costs, discounts, fees or prescription charges in CMDHB in 2007 was $817. Of the four northern DHB's, the mean drug cost for
diabetes cases in 2007 ranged from $752 in ADHB to $830 in NDHB.
Consistent with the findings for cost of laboratory claims, the peak age group for
community pharmaceutical claims (in absolute terms) was the 55-64 year old group
(Figure 28).
Figure 28: Distribution of dollar value of pharmaceutical claims by age group for diabetes cases
and those without diabetes in the reconstructed CMDHB population, 2007
00) 12500
0
$ ' 10000
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group
Adherence
It is not possible to accurately estimate adherence/compliance to diabetes medications
given the limitations of the pharmaceutical claims data available. Around 60% of
pharmaceutical claims in the northern region data set contained data for ‘total days
supply'. This variable was aggregated for northern region diabetes cases with two or
more claims for a particular medication in two years and then used to obtain an
estimate of the mean number of days of medication dispensed per person for the 2007
calendar year (Table 15). It was therefore possible to obtain a crude appreciation of
adherence to individual medications based on the mean number of days medication
supply in 2007 (assuming that medications which were dispensed were then used in
accordance with prescribing advice). Because of the way insulin dispensing was
described in the pharmaceutical claims data, information on mean days dispensed is
not available for this category.
The figures given in Table 15 are rough estimates only and must be treated with
caution. The estimates include only those diabetes cases that had data recorded for
the variable ‘total days supply'. They do not account for admissions to hospital (and
hence medications used in hospital) and an assumption is made that dispensing of
medications equates to those medications actually being taken. Furthermore, some
medications may have been initiated during the 2007 calendar year (and others
stopped for legitimate medical reasons). Initiation (and cessation) of medications part
way through 2007 would have reduced the mean number of days dispensed for these
medications, falsely implying that adherence/compliance was lower than it really was.
A thirteen month period from December 2006 to December 2007 was used to include
prescription claims which fell just outside the January 2007 to December 2007 period.
There are no established rules for how adherence/compliance to medication should be
measured, although adherence to 80% of prescribed medication is often used as a
threshold in clinical trials 40. The proportion of diabetes cases who were considered
‘adherent', in that they were dispensed medication for >80% of days in 2007, is
presented in Table 15 for each separate medication.
Table 15: Number of diabetes cases in northern region with data on ‘total days supply' and
regular claims for diabetes (and related) medications in 2007, together with mean days supplied
per medication and proportion with >80% medication adherence
Number of diabetes Proportion of year for which medications diabetes cases with >80% adherence Cilazapril with hydrochlorothiazide Metformin hydrochloride Metoprolol succinate Quinapril with hydrochlorothiazide Verapamil hydrochloride Hospital service utilisation
For the calendar year 2007, 11,800 medical and surgical hospital discharges were
recorded in NMDS for CMDHB amongst diabetes cases (1,100 principal diagnosis
discharges, 5,900 secondary and 4,800 with documentation of diabetes), while 57,700
hospital discharges were recorded for the remainder of the reconstructed population.
Sixteen point nine percent of the total hospital discharges in CMDHB for 2007 were
for people identified as diabetes cases using the decision rules. Across the remaining
three DHB's in the northern region, diabetes cases accounted for 14.5% of hospital
admissions in 2007.
Diabetes discharges were spread across a range of different hospital services, although
the greatest number (42.3%) were in inpatient adult medical services. Within the
northern region (ex. CMDHB), 49.3% of hospital discharges for diabetes cases were
for inpatient adult medical services.
Across the whole northern region for public and private hospitals (including
CMDHB), diabetes cases were found to stay in hospital more than 50% longer on
average than hospital patients who did not have diabetes. The average length of
hospital stay (LOS) in the northern region for diabetes cases was 3.6 days, whereas for hospital patients without diabetes, the average hospital LOS was 2.4 days. These average LOS estimates for the northern region were consistent with those found in CMDHB for diabetes and non-diabetes cases (3.6 and 2.4 days). Hospital admissions
Hospital discharge rates
Public and private hospital discharges in NMDS for the calendar year 2007 were
aggregated and then examined by diabetes status and DHB. Hospital discharges of all
types were more than twice as frequent in diabetes cases as in the total CMDHB
reconstructed population. In CMDHB, the crude annual hospital discharge rate for
diabetes cases was 436 discharges per 1,000 people, whereas for the total
reconstructed CMDHB population the 2007 hospital discharge rate was 195 per 1,000
people. In the northern region (ex. CMDHB), the crude 2007 hospital discharge rate
for diabetes cases was 460 per 1,000, whereas for the total reconstructed (ex.
CMDHB) northern population it was 168 per 1,000 (Table 16).
When 2007 hospital discharge rates were standardised by age and sex, the discharge
rate in the group of diabetes cases increased to more than 2.5 times that of the total
reconstructed CMDHB population (Table 16). A similar situation was found in the
northern region (ex. CMDHB), where the hospital discharge rate for diabetes cases
was 2.4 times that of the total reconstructed northern population. Note also in Table
16 that age- and sex-standardised hospital discharge rates were higher (narrowly) for
both the total reconstructed CMDHB population and the CMDHB diabetes population
than for the corresponding northern region populations.
Table 16: Crude and age- and sex-standardised hospital discharge rates for CMDHB and
northern region (ex. CMDHB) for 2007
Crude rate per 1,000 Age-, sex-standardised rate per 1,000 people CMDHB reconstructed population Northern region (ex. CMDHB) reconstructed CMMDHB diabetes cases Northern region (ex. CMDHB) diabetes cases
Crude and age-standardised hospital discharge rates for the population of diabetes
cases in CMDHB, by ethnicity are presented in Table 17. Maaori diabetes cases had
particularly high hospital discharge rates (both crude and age-standardised) in
comparison to the other three ethnic groupings.
Table 17: Crude and age-standardised hospital discharge rates by ethnicity for diabetes cases in
CMDHB, 2007
Crude rate per 1,000 people Age-standardised rate per Proportion of diabetes cases admitted to hospital
Further perspective on hospital admissions can be gained by looking at the
proportions of diabetes cases and of those without diabetes who had medical or
surgical admissions in 2007. Almost one quarter (23.4%) of diabetes cases in the
CMDHB reconstructed population had at least one medical or surgical hospital
admission in 2007, compared with only 10.7% of people without diabetes. When
these figures were standardised by age and sex, 28.8% of diabetes cases in CMDHB
were found to have had at least one admission in 2007, compared with 11.2% of those
without diabetes.
Table 18 describes the proportion of diabetes cases in CMDHB who had hospital
admissions in 2007, by ethnicity. Considerably more Maaori diabetes cases had one
or more admissions to hospital in 2007 than diabetes cases of other ethnicities.
Table 18: Proportion of diabetes cases in CMDHB who had medical/surgical hospital admissions
in 2007, by ethnicity
Crude proportion of cases Age-standardised proportion admitted ≥ 1 times in 2007 (%) admitted ≥ 1 times in 2007 (%) Procedures
As with hospital discharge rates, major surgical procedures for the calendar year 2007
in NMDS were aggregated and then examined by diabetes status and DHB.
Procedural DRG's (diagnosis-related groups) were used to identify whether an
individual had a significant surgical procedure or not. The unadjusted frequency of
major surgical procedures in CMDHB diabetes cases in 2007 was about 2.5 times that
of the total CMDHB reconstructed population (Table 19). The CMDHB crude major
procedure rate for diabetes cases in 2007 was 111 procedures per 1,000 people,
whereas the procedure rate was 44 per 1,000 for the total CMDHB reconstructed
population. This difference diminished substantially once procedure rates were
standardised for age and sex.
Table 19: Crude and age- and sex-standardised major surgical procedure rates for CMDHB and
northern region (ex. CMDHB) for 2007
Crude rate per 1,000 Age-, sex-standardised rate per 1,000 people CMDHB reconstructed population Northern region (ex. CMDHB) reconstructed CMDHB diabetes cases Northern region (ex. CMDHB) diabetes cases When major surgical procedures within the population of diabetes cases of CMDHB were analysed, it was apparent that (as was the case with hospital discharges) Maaori diabetes cases had noticeably higher crude and age-standardised major surgical procedure rates than diabetes cases of other ethnicities (Table 20). Table 20: Crude and age-standardised major surgical procedure rates by ethnicity for diabetes
cases in CMDHB, 2007
Crude rate per 1,000 people Age-standardised rate per Ischaemic heart disease
There were 639 NMDS hospital discharges in CMDHB with principal diagnoses of
IHD (ischaemic heart disease, ICD-10-AM codes I20 to I25) in diabetes cases in the
calendar year 2007. This corresponded to a crude hospital discharge rate of 24
discharges per 1,000 diabetes cases. In comparison, there were 1,878 hospital
discharges for IHD (principal diagnosis) in the total CMDHB population in 2007,
corresponding to a crude hospital discharge rate of only 4 per 1,000 people. In the
whole northern region (including CMDHB), the crude hospital discharge rate for
diabetes cases was 31 per 1,000 and for the total population it was 6 per 1,000. Crude
hospital discharge rates for IHD by ethnicity for CMDHB are presented in Table 21.
The age-standardised CMDHB hospital discharge rate for diabetes cases with
principal diagnoses of IHD in 2007 was 12 per 1,000 people, while for the total
CMDHB population it was 5 per 1,000. In the whole northern region (including
CMDHB), the age-standardised rate in diabetes cases was 15 per 1,000 and for the
entire northern population it was 6 per 1,000 people.
Table 21: Crude hospital discharge rates for ischaemic heart disease in CMDHB in 2007, by
ethnicity
Diabetes cases - rate per Total population - rate Cerebrovascular disease (stroke)
In CMDHB, there were 215 hospital discharges amongst diabetes cases with principal
diagnosis codes for cerebrovascular disease (stroke), ICD-10-AM I60 to I69, in
calendar year 2007, corresponding to a crude hospital discharge rate of 8 per 1,000
people. For the whole CMDHB population, the crude hospital discharge rate for
cerebrovascular disease was one quarter that of diabetes cases, at 2 per 1,000 people.
Across the entire northern region (including CMDHB) in 2007, the crude hospital
discharge rate for cerebrovascular disease was 9 per 1,000 for diabetes cases and 2 per
1,000 for the total population. Crude hospital discharge rates by ethnicity for
cerebrovascular disease are given in Table 22.
Table 22: Crude hospital discharge rates for cerebrovascular disease in CMDHB in 2007, by
ethnicity
Diabetes cases - rate per Total population - rate The age-standardised discharge rate for cerebrovascular disease in diabetes cases in CMDHB in 2007 was 4 per 1,000, compared with 2 per 1,000 for the total CMDHB population. These rates were consistent with those found in the total northern region (including CMDHB) in 2007, where the discharge rates were 4 and 2 per 1,000 for diabetes cases and the entire population respectively. Diabetes in pregnancy
There were 16,800 births in CMDHB in 2006-2007. Diabetes in pregnancy
complicates analysis of diabetes in the reconstructed population. A woman may have
known pre-existing diabetes and become pregnant, pre-existing type II diabetes that is
first diagnosed in pregnancy, or gestational diabetes (GDM) caused by changes in
endocrine function during pregnancy 41. Women with GDM are at greater risk of
developing type II (and also type I) diabetes postpartum than the general population 42
43. Women with ICD-10-AM codes for diabetes diagnosed during pregnancy were excluded from the diabetes case group in this study, as resolution of normal glucose metabolism postpartum is likely in the majority of this group and their inclusion may therefore over-estimate prevalence. However, uncertainty around diagnostic categories at the time of clinical coding may result in misclassification of some women with diabetes in pregnancy. In particular, women with first diagnosis of type II diabetes in pregnancy may be missed from the diabetes case group using the existing decision rules. This section aims to quantify absolute numbers of women in CMDHB and the northern region with ICD-10-AM codes for diabetes in pregnancy in order to understand the potential magnitude of misclassification. The ICD-10-AM codes O24.0 to O24.3 are used for pre-existing diabetes in pregnancy. There were 403 women in CMDHB who had hospital discharges in NMDS with principal or secondary codes for pre-existing diabetes between 1990 and 2007, who also had a health event recorded in the 2006-2007 study period. This group of women was included in the diabetes case group (1.5% of CMDHB diabetes case group). Of these women, 129 had discharge codes for pre-existing diabetes in 2006-2007. In the whole northern region, 1,075 women had discharges coded for pre-existing diabetes between 1990 and 2007, of which 344 were coded for this diagnosis in 2006-2007 (Table 23). Table 23: Absolute numbers of women with diagnosis codes for pre-existing diabetes in
pregnancy in northern region, by ethnicity in 2006-2007 reconstructed group

The ICD-10-AM codes O24.4 to O24.9 correspond to NMDS discharges for diabetes
arising in pregnancy, which includes GDM and newly diagnosed type II diabetes.
Women with these diagnosis codes were not included in the diabetes case group
unless they also met the laboratory or pharmaceutical criteria for inclusion. Within
the whole northern region, 1,267 women had discharge codes for diabetes arising in
pregnancy in 2006-2007 (Table 24).
Table 24: Absolute numbers of women with diagnosis codes for new diabetes in pregnancy in
northern region, by ethnicity in 2006-2007 reconstructed group
Of the 1,267 women in the northern region with diagnosis codes for new diabetes, 681 (53.7%) were included in the diabetes group anyway by way of the decision rules for laboratory and pharmaceutical claims (251 of these women in CMDHB). Discussion
This retrospective, cross-sectional study used routinely collected administrative data
from community laboratory and pharmaceutical subsidy claims, together with data on
hospital discharges recorded in NMDS, to create a ‘reconstructed' population for
CMDHB and three other DHB's in the northern region. Within this reconstructed
population, individuals with diabetes were identified using a set of decision rules
which were developed using a review of diabetes literature, consultation with experts
and sensitivity analysis. This study found an age- and sex-standardised prevalence of
diabetes in CMDHB of 7.1%, the highest prevalence of any DHB in the northern
region.
Health inequities, often called health inequalities, are "differences in health that are
unnecessary, unavoidable, unfair and unjust" 44. Ethnicity-based inequity in health,
particularly between Maaori and non-Maaori, has been a consistent feature of the
health landscape in New Zealand for many years 45-47. Like previous studies of
diabetes in CMDHB, this analysis identified inequity in the prevalence of diabetes
between those of Maaori, Pacific and Asian ethnicities and those of Other ethnicity 6 8.
Those of Maaori and Pacific ethnicities were found to have the highest prevalence of
diabetes in CMDHB. Those of Asian ethnicity were also found to have considerably
higher diabetes prevalence than those of Other ethnicity. The disparity in diabetes
prevalence was greatest between Pacific females and females of Other ethnicity.
Inequity in diabetes prevalence is consistent with differences in diabetes mortality
rates. Between 2001 and 2005, the mortality rate for type 2 diabetes (as the
underlying cause of death [ICD-10-AM code E11]) for Maaori aged 65 or more years
was 520 per 100,000, for Pacific it was 440 per 100,000, while for Other it was 85 per
100,000 people 48. These disparities reinforce the need for culturally appropriate
programmes such as Let's Beat Diabetes, which aim to reduce inequity by giving
priority to the most vulnerable groups.
Access to health care is defined as "…timely use of personal health services to
achieve the best possible health outcomes" 49. Differential access to health care is
noted to be an important driver of health inequity 50. Access is dependent on both
utilisation of health services and achievement of health outcomes 49. Although
outcome gaps are clearly evident for diabetes in CMDHB, this study found several
positive signs that utilisation of community laboratory monitoring tests and
pharmaceuticals may be similar between groups of different ethnicity. Utilisation of
tests such as HbA1c, microalbumin and lipid studies was similar between Maaori,
Pacific, Asian and Other diabetes cases, and was similar across the spectrum of
deprivation (although some under-utilisation of monitoring tests by those of Asian
ethnicity was noted in unadjusted analysis). Likewise, patterns of prescribing for
medications used for secondary prevention in diabetes, such as statins and ACE
inhibitors, were similar between the four groups (and across the spectrum of
deprivation). Perhaps initiatives such as CCM, which aim to improve the access of
people with chronic conditions to primary care, have had a role in improving
utilisation of primary health care amongst disadvantaged groups in CMDHB.
Analysis of ethnicity using high-level categories such as ‘Pacific' and ‘Asian' is not
without limitation and assumes that individual ethnicities aggregated within these
groups have similar health characteristics. This is not necessarily the case, as was
highlighted in a recent report on Asian health needs which found considerable
heterogeneity in health indicators between the different groups (such as Indian and Chinese ethnicities) which made up the broader Asian group 20. Monitoring and medication for diabetes in CMDHB
There are a number of missing factors which make it difficult to judge the
performance of the CMDHB community in monitoring and appropriately treating
diabetes and preventing admissions to hospital. However, overall CMDHB seems to
have led the other three northern DHB's in several areas during 2006-2007. A greater
proportion of diabetes cases in CMDHB had regular claims for important monitoring
tests such as HbA1c and urinary microalbumin. There are several possible
explanations for the greater utilisation of these tests in CMDHB. Examples of
possible explanations include greater awareness of diabetes (and of the guidelines for
diabetes monitoring) amongst GP's in CMDHB, greater disease severity with
corresponding requirements for more frequent monitoring in CMDHB and initiatives
to improve access to primary care services, such as CCM. Almost 92% of 3,500
people who were enrolled in the diabetes module of CCM across the entire January
2006 to December 2007 period had at least two HbA1c tests performed in the two
years, while just over 60% had four or more HbA1c tests (compared with 83% and
52% respectively for the entire CMDHB diabetes population 51. If individuals in the
CCM diabetes module were removed from analysis, 82% of remaining diabetes cases
had two or more HbA1c tests in two years (the same proportion as the other three
northern DHB's) and 51% had four or more tests (compared with 44% for the other
DHB's). Although CCM probably did influence the frequency of claims for HbA1c
amongst diabetes cases, CMDHB appeared to perform well in comparison to the other
three DHB's even when CCM patients were excluded from analysis.
As is the case with laboratory claims, it is difficult to assess the appropriateness of
prescribing patterns for diabetes in CMDHB using the available data in this study.
However, it is heartening to see that CMDHB again leads the way in the prescription
of medications, such as ACE inhibitors and statins, which are important for secondary
prevention of diabetes complications. What then is the role of programmes such as
CCM, in this instance? It is not possible to give a definitive answer. However, we do
know that half way through the study period in January 2007, there were almost 7,000
people enrolled in the diabetes module of CCM, 78.5% of whom were prescribed
statins 52. If these individuals are removed from the group of CMDHB diabetes cases,
almost 57% of the 20,000 or so remaining diabetes cases had regular pharmaceutical
claims for statin medication. In the other three northern DHB's the proportion of total
diabetes cases with regular pharmaceutical claims for statins ranged between 53% and
56%. It therefore seems probable that the CCM programme did influence the
proportion of diabetes cases prescribed statins, however even without the influence of
CCM, CMDHB performed well in this part of secondary prevention. Further work on
medication adherence and prescribing patterns will help tease the performance of
CMDHB in this area out further.
A key question is whether more intensive follow up of diabetes cases offered by
community-focused programmes such as CCM reduced hospitalisation amongst
diabetes cases in 2006-2007? It is not possible to answer this question using only the
data in this study. CMDHB did have similar hospitalisation and major procedure
rates to the other three DHB's in the northern region, yet the part played by factors
such as severity of local disease and the role of initiatives like CCM is not clear.
Further longitudinal research that compares admission rates for those in CCM with other diabetes cases over time is required. Comparison with Known Diabetes
Several attempts have been made to quantify the burden of diabetes in CMDHB. As
mentioned in the introduction to this report, the Known Diabetes database has
identified around 23,000 people with diabetes who have previously accessed hospital
and specific primary care services in CMDHB 9. The age distribution, sex and
ethnicity compositions of CMDHB diabetes cases from the reconstructed population
were very similar to those of the Known Diabetes group (Figures 29 and 30).
Figure 29: Comparison of age structure of diabetes cases in reconstructed population with that
of cases identified in Known Diabetes database
00-04 05-14 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Age group (years)

Figure 30: Comparison of ethnicity of diabetes cases in reconstructed population with that of
cases in Known Diabetes group for CMDHB
The prevalence rates for diabetes within the reconstructed CMDHB population (by ethnicity) also appear generally consistent with those found in the Diabetes Heart and Health Study, the New Zealand Health Survey 2002/03, the LDB Benchmark Survey and with estimates of diabetes prevalence in CMDHB undertaken by the Ministry of Health 6 8 53 54. The estimate of 26,400 adults aged ≥ 15 years with diabetes in CMDHB (a crude adult prevalence of 8.2%) found in 2006/07 NZHS is also close to 27,000 or so adults estimated to have diabetes in CMDHB during the same period in the reconstructed population used in this study 7. Cost of diabetes care
This study found that diabetes cases in CMDHB (and the other DHB's in the northern
region) had substantially higher costs for community laboratory and pharmaceutical
claims than individuals without diabetes in the reconstructed population. The mean
cost of laboratory claims for diabetes cases in CMDHB in 2007 was almost five times
the cost of claims for people without diabetes. CMDHB had the greatest difference in
mean laboratory claims between diabetes cases and those without diabetes of any of
the northern DHB's. Even so, the difference was still around fourfold for all of the
other three DHB's. In the case of community pharmaceutical claims, the cost for
diabetes cases in CMDHB was more than seven times that of people without diabetes.
Similar results were found for pharmaceutical claims in the other three DHB's. When
crude mean costs for pharmaceutical and laboratory claims (ex. GST) were combined
with the mean cost of hospitalisation for diabetes cases and those without diabetes in
CMDHB in 2007 (using a hospitalisation cost estimate of 1 wies2007/08), diabetes was noted to be responsible for an additional $73 million to the
DHB for the year (Table 25).
Table 25: Difference in cost of medical/surgical hospital discharges, community laboratory and
pharmaceutical claims for those with and without diabetes in CMDHB in 2007
Cost per person Cost per person w/o Difference per Number with Additional cost with diabetes ($) Laboratory claims Pharmaceutical claims Inpatient discharges
When the laboratory, pharmaceutical and hospital costs were standardised by age and
sex, there was a slight reduction in the total additional cost for diabetes in CMDHB
(Table 26). However, the additional cost remained considerable, at around $66
million.
Table 26: Age- and sex-standardised cost of medical/surgical discharges, community and
pharmaceutical claims for those with and without diabetes in CMDHB in 2007
Cost per person Cost per person w/o Difference per Number with Additional cost with diabetes ($) Laboratory claims Pharmaceutical claims Inpatient discharges The estimated additional cost of diabetes in this analysis is likely to be an underestimate of the true cost of health care for diabetes in CMDHB. Several aspects of diabetes health care have not been included in the estimate, for example primary care visits and retinal screening. 7 Weighted Inlier Equivalent Separations, cost-weighting applied to hospital admissions Strengths and weaknesses
The main strengths of this study lay in the currency of the data, the low cost (to
CMDHB) of its collection, the extensive detail collected in the three contributing
databases and the ability to link numerators with denominators from the same data set
(Table 27).
Laboratory and pharmaceutical data came from claims processed by the HealthPAC
General Transaction Processing System for reimbursement of community laboratories
and retail pharmacies, meaning that the data became available only a few months after
claims for late 2007 were processed. This meant that very timely estimates could be
made of diabetes in the northern region. Additionally, CMDHB did not incur any
direct financial cost in acquisition of the claims data. Costs were only incurred in
time spent analysing and reporting on the data.
There was a substantial amount of data available for analysis in the final complete
data set. The pharmaceutical data alone contained 42 separate variables.
Furthermore, numerators and denominators in the analyses all came from the same
reconstructed population. The advantage of consistency between numerators and
denominators was that social and demographic variables in the data analyses were
directly linked, thereby avoiding numerator-denominator bias in the analysis of
ethnicity 55 (although ethnicity data in hospital records can still differ substantially
from self-identified ethnicity 56).
The result was a set of timely, cheap and comprehensive data for the resident
populations of CMDHB and the northern region. In this situation the data was used to
explore diabetes in CMDHB; however it is also suitable for analysis of a wide range
of different clinical conditions in various contexts.
In terms of limitations, the data used to create the reconstructed population for
CMDHB and the rest of the northern region was of an administrative nature and was
not designed for assessment of prevalence and other epidemiological analyses. This
meant that decision rules were necessary to identify individuals with diabetes (and
there was a degree of uncertainty around the criterion validity of such rules). For
example, although current advice does not support the use of HbA1c as a screening
test for diabetes 3 21 57, there is some discussion of its use as a screening tool in the
literature 58 59 and anecdotal evidence suggests it is being used for such purposes.
Even though the final decision rule for HbA1c required claims for at least three such
tests in two years, it is still likely that a (probably small) subset of patients may have
been misclassified as having diabetes based on frequency of HbA1c testing when they
had no diabetes diagnosis. Similarly, medications such as metformin may be used in
people without diabetes 60 61, although sensitivity analysis and review of the literature
indicated that for diabetes medications in the August 2006 Pharmaceutical Schedule
such use would not have materially influenced the findings in this study.
Furthermore, some individuals with pre-diabetes (impaired fasting glucose and
impaired glucose tolerance) may have been misclassified as having diabetes using the
decision rules. Any misclassification of these individuals into the diabetes group is
not of great concern, as people with pre-diabetes are at increased risk of developing
diabetes and share some of the risks experienced by those with diabetes, such as
greater risk of cardiovascular disease than the general population 62-66.
Table 27: Table of study strengths and weaknesses

Strengths Weaknesses
• Current data (as recent as December
• Data was not collected for 2007) was available, meaning that epidemiological purposes, creating analyses could be conducted and the need for proxy means of detection results generated in a timely fashion of diabetes cases (decision rules) • The large number of people in the • Decision rules were not formally final data set indicated fairly validated in a published study – rely comprehensive coverage of the on literature review of decision rule CMDHB and northern region appropriateness, expert opinions and populations in the reconstructed sensitivity analysis • ‘Churn' created by people moving in • Availability of community laboratory and out of CMDHB meant some and pharmaceutical data meant a residential address data may have community-level approach to analysis been obsolete, causing was possible for the first time misclassification in both numerators • Creation of the data set from routinely and denominators collected administrative data meant • Not all scripts and lab requests had that it was inexpensive and relatively NHI numbers (meaning small number easy to acquire for CMDHB left out of reconstructed population) • Analyses in the study are easily • Hospital lab tests and pharmaceuticals replicable in future using updated data were not included in analysis • The large number of variables • Those who did not access health available for analysis meant that a services in 2006/2007 were excluded reasonable understanding of from denominator – resulting in laboratory monitoring, prescribing possible prevalence over-estimation habits and hospital utilisation in • Not all prescriptions or laboratory CMDHB was possible tests may have generated subsidy • Availability of data from the other claims, e.g. prescriptions for three northern DHB's meant that medications not on Schedule comparison of CMDHB with other • Some diabetes cases would have populations was possible ‘fallen through the cracks' – not • Study findings are consistent with detected using decision rules 2006 census population estimates and • Data was encrypted, so could not be with findings of other studies of linked to Known Diabetes database, diabetes in CMDHB, i.e. estimates of CCM, or other ‘live NHI' data reconstructed populations are broadly • Likely to be different monitoring/ similar to census 2006 and estimates prescribing patterns in CCM vs. other of diabetes prevalence triangulate diabetes cases – has not been teased approximately with other estimates from previous analyses • Unquantified overseas visitor/student • High rate of service utilisation implies prescription of diabetes medications that diabetes cases are highly likely to and use of community laboratory tests feature in data sets The collection of data for the reconstructed population was dependent on the recording of a hospital event or claim for a community pharmaceutical or a laboratory test during the 2006/2007 period, meaning some residents who did not experience such health events during this period were left out of the analysis. This effect was evident in younger age groups in CMDHB (especially for males), when the reconstructed population was compared with census estimates. A further group of patients may have been excluded due to not having NHI numbers recorded on pharmaceutical or laboratory claims. It is difficult to estimate the size of this second group. It is known that last year (2007) around 94% of prescription claims in Pharmhouse and laboratory claims in Labs had NHI numbers recorded 67. It is therefore likely that this group of missing patients is quite small. The under-quantification of the denominator resulting from both of these factors would have resulted in over-estimation of the diabetes prevalence in CMDHB and the other three northern DHB's. However, the effect of this over-estimation was largely lost when prevalence was standardised by age and sex, as the bulk of diabetes cases occurred in those aged 35 years or more and this group had the best coverage in the total reconstructed population when compared with census data. Recommendations for further analysis
1. Longitudinal comparison of CCM diabetes cases with non-CCM diabetes cases.
The CCM programme gives patients with chronic conditions the opportunity for close support and follow up from their primary care provider 68. This programme began in 2001 and was fully operational by late 2003. There were 6,900 patients in the diabetes module in December 2006 and 7,470 in June 2007. Diabetes patients in CCM would have had closer disease monitoring in the community and this may have distorted the analyses of laboratory test monitoring and prescribing patterns. Further work needs to compare patterns of care in each group and hospital admission rates, through encryption of CCM NHI numbers and subsequent linkage to the reconstructed population for CMDHB 2. Domicile codes can be used to map diabetes cases in CMDHB geographically according to CAU, using a geographic information system (GIS). The findings of such analyses can then be used to inform service provision and to describe patterns of diabetes laboratory monitoring and prescribing patterns by locality 3. Further analysis of diabetes in those of Asian ethnicity is required in order to understand the prevalence of diabetes in groups that make up this ethnicity category and to describe patterns of laboratory testing, pharmaceutical prescribing and hospital utilisation in these groups 4. Further work on adherence/compliance to pharmaceuticals within CMDHB. This could be expanded to all medications on the pharmaceutical schedule which are prescribed long-term 5. Further research is needed to understand the expected frequency of laboratory test monitoring in relation to current guidelines and to establish the expected proportion of diabetes cases likely to benefit from particular medications such as ACE inhibitors and statins. This may enable the formulation of targets and indicators for the care of diabetes in the community References
1. Powers AC. Chapter 338. Diabetes mellitus. In: Fauci AS, Braunwald E, Kasper
DL, et al, editors. Harrison's Principles of Internal Medicine. 17th ed. New York: The McGraw-Hill Companies, 2008. 2. Williams G, Pickup JC. Handbook of diabetes. London: Blackwell Science Ltd, 3. New Zealand Guidelines Group. Evidence-based best practice guideline: The assessment and management of cardiovascular risk. Wellington: Ministry of Health and New Zealand Guidelines Group, 2003. 4. Ministry of Health. Modelling diabetes: A summary. Wellington: Ministry of 5. Ministry of Health. The New Zealand health strategy. Wellington: Ministry of 6. Ministry of Health. A portrait of health: Key results of the 2002/03 New Zealand Health Survey. Wellington: Ministry of Health, 2004. 7. Ministry of Health. A portrait of health: Key results of the 2006/07 New Zealand Health Survey. Wellington: Ministry of Health, 2008. 8. Wyllie A, MacKinlay C. Let's Beat Diabetes benchmark survey. Auckland: Phoenix Research for Counties Manukau DHB, 2007. 9. Pearce S. Known Diabetes database: Summary as at 22 November 2007. Manukau City: Counties Manukau District Health Board, 2007. 10. Pearce S. Known diabetes: Personal communication, 2008. 11. Ministry of Health. Ethnicity data protocols for the health and disability sector. Wellington: Ministry of Health, 2004. 12. Pharmac. New Zealand pharmaceutical schedule: August 2006. Wellington: Pharmaceutical Management Agency, 2006. 13. Pharmac. New Zealand pharmaceutical schedule: December 2007. Volume 14, number 3. Wellington: Pharmaceutical Management Agency, 2008. 14. New Zealand Health Information Service. Pharmaceutical claims data mart (pharms) data dictionaryversion 4.0. Wellington: Ministry of Health, 2006. 15. New Zealand Health Information Service. Laboratory claims data warehouse data dictionary, version 2.0. Wellington: Ministry of Health, 2005. 16. New Zealand Health Information Service. National minimum dataset (hospital events). Wellington: Ministry of Health, 2007. 17. New Zealand Health Information Service. Mortality collection data dictionary, version 1.2. Wellington: Ministry of Health, 2004. 18. New Zealand Health Information Service. Current data access policy. Wellington: Ministry of Health, 2007. 19. Klomp H, Lawson JA, Cockcroft DW, et al. Examining asthma quality of care using a population-based approach. Canadian Medical Association Journal 2008;178(8):1013-1021. 20. Gala G. Health needs assessment for Asian people in Counties Manukau: Counties Manukau District Health Board, 2008. 21. American Diabetes Association. Executive summary: Standards of medical care in diabetes - 2008. Diabetes Care 2008;31(S1):S5-S11. 22. Counties Manukau District Health Board. Chronic care management programme: General practice team manual for diabetes version 1.0. Manukau City: Counties Manukau District Health Board, 2005. 23. New Zealand Guidelines Group. Best practice evidence-based guideline: Management of type 2 diabetes. Wellington: Ministry of Health and New Zealand Guidelines Group, 2003. 24. Tracey J, Kenealy T, Sinclair G. Management of type 2 diabetes. NZFP 2002;29(3):194-199. 25. Blakely T, Fawcett J, Tobias M, et al. Decades of disparity II: Socioeconomic mortality trends in New Zealand, 1981-1999. Public Health Intelligence occasional bulletin number 25. Wellington: Ministry or Health, 2005. 26. Carr-Hill R, Chalmers-Dixon P. The Public Health Observatory handbook of health inequalities measurement. Oxford: South East Public Health Observatory, 2005. 27. Gill M, Ockelford P, Morris A, et al. Diagnostic handbook: The interpretation of laboratory tests. 3rd ed. Auckland: Diagnostic Medlab Limited, 2000. 28. Barker AN, editor. LabPlus laboratory handbook. Auckland: Auckland District Health Board, 2008. 29. Nathan DM, Buse JB, Davidson MB, et al. Management of hyperglycemia in type 2 diabetes: A consensus algorithm for the initiation and adjustment of therapy: A consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2006;29(8):1963-1972. 30. Sowers JR, Epstein M, Frohlich ED. Diabetes, hypertension and cardiovascular disease: An update. Hypertension 2001;37(4):1053-9. 31. Hypertension in Diabetes Study Group. Hypertension in Diabetes Study (HDS): I. Prevalence of hypertension in newly presenting type 2 diabetic patients and the association with risk factors for cardiovascular and diabetic complications. Journal of Hypertension 1993;11(3):309-317. 32. UK Prospective Diabetes Study Group. Tight blood pressure control and risk of macrovascular and microvascular complications in type II diabetes: UKPDS 38. British Medical Journal 1998;317(7160):703-713. 33. Adler AL, Stratton IM, Neil HA, et al. Association of systolic blood pressure with macrovascular and microvascular complications of type II diabetes (UKPDS 36): Prospective observational study. British Medical Journal 2000;321(7258):412-9. 34. Bakris GL. Treatment of hypertension in diabetes mellitus: UpToDate, 2008. 35. Bakris GL. Treatment and prevention of diabetic nephropathy: UpToDate, 2008. 36. McCulloch DK. Overview of medical care in adults with diabetes mellitus: 37. Rosenson RS. Overview of treatment of hypercholesterolemia: UpToDate, 2008. 38. Adult Treatment Panel III. Third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Bethesda, Maryland: National Institutes of Health, 2002. 39. Talbot F, Nouwen A. A review of the relationship between deression and diabetes in adults: Is there a link? Diabetes Care 2000;23(10):1556-1562. 40. Osterberg L, Blaschke T. Adherence to medication. New England Journal of Medicine 2005;353(5):487-497. 41. Jovanovic L. Screening and diagnosis of gestational diabetes mellitus: UpToDate, 42. Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: A systematic review. Diabetes Care 2002;25(10):1862-1868. 43. Ferber KM, Keller E, Albert ED, Ziegler AG. Predictive value of human leukocyte antigen class II typing for the development of islet autoantibodies and insulin-dependent diabetes postpartum in women with gestational diabetes. Journal of Clinical Endocrinology and Metabolism 1999;84(7):2342-2348. 44. Whitehead M. The concepts and principles of equity and health. International Journal of Health Services 1992;22(3):429-445. 45. Ajwani S, Blakely T, Robson B, et al. Ethnic mortality trends in New Zealand, 1980-1999. Wellington: Ministry of Health and University of Otago, 2003. 46. Blakely T, Tobias M, Atkinson J, et al. Tracking disparity: Trends in ethnic and socioeconomic inequalities in mortality, 1981-2004. Wellington: Ministry of Health, 2007. 47. Robson B, Harris R, editors. Hauora: Maori standards of health IV. A study of the years 2000-2005. Wellington: Te Ropu Rangahau Hauora a Eru Pomare, 2007. 48. Papa D. Analysis of NMDS data, 2008. 49. Millman M, editor. Access to health care in America. Washington, DC: National Academy Press, 1993. 50. Reid P, Robson B. The state of Maori health. In: Mulholland M, editor. State of the Maori nation: Twenty-first-century issues in Aotearoa. Auckland: Reed, 2006. 51. Shi F. Personal communication. 2008. 52. Shi F. CCM diabetes summary report for Counties Manukau: 01 Jan 2007 to 31 Jan 2007. Manukau City: Counties Manukau District Health Board, 2008. 53. Sundborn G, Metcalf P, Scragg R, et al. Ethnic differences in the prevalence of new and known diabetes mellitus, impaired glucose tolerance, and impaired fasting glucose. Diabetes Heart and Health Survey (DHAH) 2002-2003, Auckland New Zealand. New Zealand Medical Journal 2007;120(1257). 54. Wright C. Health system 'diagnosis' codes - Chemical, ICD, purchase unit (personal communication). Wellington: Ministry of Health, 2008. 55. Ajwani S, Blakely T, Robson B, et al. Decades of disparity: Ethnic mortality trends in New Zealand 1980-1999. Wellington: Ministry of Health and University of Otago, 2003. 56. Swan J, Lillis S, Simmons D. Investigating the accuracy of ethnicity data in New Zealand hospital records: Still room for improvement. New Zealand Medical Journal 2006;119(1239). 57. McCulloch DK. Screening for diabetes mellitus: UpToDate, 2007. 58. Kenealy T, Elley CR, Arroll B. Screening for diabetes and prediabetes. Lancet 2007;370:1888-1889. 59. Waugh N, Scotland G, McNamee P, et al. Screening for type 2 diabetes: literature review and economic modelling. Health Technology Assessment 2007;11(17). 60. Kashyap S, Wells GA, Rosenwaks Z. Insulin-sensitizing agents as primary therapy for patients with polycystic ovarian syndrome. Human Reproduction 2004;19(11):2474-83. 61. Sinawat S, Buppasiri P, Lumbiganon P, Pattanittum P. Long versus short course treatment with Metformin and Clomiphene Citrate for ovulation induction in women with PCOS. Cochrane Database of Systematic Reviews 2008;Jan 23(1):CD006226. 62. Coutinho M, Gerstein HC, Wang Y, et al. The relationship between glucose and incident cardiovascular events: A metaregression analysis of published data from 20 studies of 95 783 individuals followed for 12·4 years. Diabetes Care 1999;22(2):233-240. 63. DECODE Study Group. Glucose tolerance and mortality: comparison of WHO and American Diabetes Association diagnostic criteria. Lancet 1999;354(9179):617-621. 64. McCulloch DK, Robertson RP. Prediction and prevention of type 2 diabetes mellitus: UpToDate, 2007. 65. Nathan DM, Davidson MB, DeFronzo RA, et al. Impaired fasting glucose and impaired glucose tolerance: Implications for care. Diabetes Care 2007;30(3):753-759. 66. Padwal R, Varney J, Majumdar SR. A systematic review of drug therapy to delay or prevent type 2 diabetes. Diabetes Care 2005;28(3):736-744. 67. Hanham G. Personal communication with Regional Decision Support Team, 68. Counties Manukau District Health Board. Services we fund: Chronic care management. Manukau City: Counties Manukau District Health Board, 2008.

Source: http://www.countiesmanukau.health.nz/assets/About-CMH/Performance-and-planning/health-status/2008-diabetes-in-CMDHB.pdf