S T A T I S T I C S

Print this  Email this

Methods and data sources

• Indicator selection and presentation
• Data (numerator and denominator)
• ICD Codes
• Statistical methods (standardisation)
  • Ethnicity
  • Age groups
  • Ninety-five percent confidence intervals
  • Rate ratios
  • Ever Māori Analysis
• Methods
• Results
• Summary

---

Indicator selection and presentation

Health indicators are summary measures that provide an indication of wider health concerns and serve to focus attention on key issues. The format of this publication focuses on presenting key information through graphs and tables, with short explanations that document the results. It is intended to complement other sources of Māori health information.

Indicators were selected using conventional criteria relating to their ability to signal wider health concerns, to focus on salient health issues, to be reliably and validly monitored, and for their responsiveness to change. In particular, indicators were chosen that relate to Māori health priority areas as identified in He Korowai Oranga and the New Zealand Health Strategy. Additional indicators were selected to align with those already in use in annual monitoring at a national level. Where required, advice was sought from experts in the areas.


Data (numerator and denominator)

Numerator
Data sources for the indicators are listed below.

Table 1: Data sources for numerators

Source (agency or collection)	Data	Period
New Zealand Health Information Service	Mortality collection data set - mortality	2000-2002
	National Minimum Data Set (NMDS) - hospitalisations	2002-2004
New Zealand Cancer Registry	Cancer registrations	1999-2001
Statistics New Zealand	Infant mortality	2000-2002
	Low birthweight	2000-2002
	Life expectancy	1951-2001
	Disability	2001
	Demographics	2001
	Population projections	2006-2021
	Socioeconomic indicators	2001
Institute of Environmental Science and Research Limited (ESR)	Infectious disease notifications	2001-2003
2002/03 New Zealand Health Survey	Risk and protective factors, disease prevalences, health service utilisation and self-rated health	2002/03
2002 Children's Nutrition Survey	Risk and protective factors	2002
Action on Smoking and Health (ASH)	Smoking prevalence for youth	2003
Youth 2000	School violence indicators	2000
Get Checked Programme, Ministry of Health	Diabetes indicators	2004
Plunket	Breastfeeding rates	2003/04
National Audiology	Hearing failure rates	2003/04

Full details of ICD-9 codes used for administrative data are provided below. For administrative data the most recent three years of data were aggregated to provide stable rate estimates. For survey data the most recent survey year was presented.

Denominator
Population information was sourced from 2001 Census data from Statistics New Zealand. The relevant Census count multiplied by 3 (for a three-year period) was used as the denominator.


ICD Codes

Table A1-1: Avoidable mortality codes

Condition	ICD-9
Tuberculosis*	010-018,137
Selected invasive bacterial and protozoal infection*	034-036, 038, 084, 320, 481-482, 485, 681-682
HIV/AIDS	042
Hepatitis (all types)	070
Viral pneumonia and Influenza	480,487
Lip, oral cavity and pharynx cancers	140-149
Oesophageal cancer	150
Stomach cancer	151
Colorectal cancer*	153, 154
Liver cancer	155
Lung cancer	162
Melanoma of skin*	172
Non-melanotic skin cancer*	173
Breast cancer*	174
Uterine cancer*	179, 182
Cervical cancer*	180
Bladder cancer*	188
Thyroid cancer*	193
Hodgkins disease*	201
Leukaemia*	204.00, 204.01, 204.10, 204.11
Benign tumours*	210-229
Thyroid disorders*	240-246
Diabetes*	250
Alcohol-related disease	291, 303, 305.0, 425.5, 535.3 571.0-571.3, 760.8
Illicit drug-use disorders	292, 304, 305.2-305.9
Epilepsy*	345
Rheumatic and other valvular heart disease*	390-398
Hypertensive heart disease*	402
Ischaemic heart disease*	410-414
Cerebrovascular diseases*	430-438
Aortic aneurysm	441
Nephritis and nephrosis*	403,580-589,591
Obstructive uropathy and prostatic hyperplasia*	592, 593.7, 594, 598, 599.6, 600
Deep vein thrombosis with pulmonary embolism	415.1, 451.1
Chronic obstructive pulmonary disease	490-492, 496
Asthma*	493
Peptic ulcer disease*	531-534
Acute abdomen, appendicitis, intestinal obstruction, cholecystitis/lithiasis, pancreatitis, hernia*	540-543, 550-553, 574-577
Cirrhosis, chronic hepatitis and other chronic liver disease	571
Birth defect*	740-759
Complications of perinatal period*	764-779
Road traffic injuries, other transport injuries	E810-E819
Accidental poisonings	E850-E869
Falls	E880-E886, E888
Fires, burns	E890-E899
Drownings (swimming)	E910
Suicide and self-inflicted injuries	E950-E959, E980-E989
Violence	E960-E969

^*These conditions are amendble to health care.


Table A1-2: Avoidable hospitalisation codes

Condition	ICD-9
Tuberculosis*	010-018, 137
HIV/AIDS*	042
Skin cancers*	140, 172, 173
Oral cancers*	141, 143-146, 148-149, 161
Colorectal cancer*	153, 154
Lung cancer*	162
Breast cancer*	174
Cervical cancer*	180
Nutrition*	260-269, 280-281
Alcohol-related conditions*	291, 303, 305.0, 425.5, 535.3, 571.0-571.3
Angina	411.1, 411.8, 413, 786.5
Gastroenteritis*	001-009, 558.9, 779.3, 787.0, 787.9
Other infections*	023, 027, 034-035, 084, 770.0, 771.1-771.2, 771.4-771.9
Immunisation preventable*	032-033, 037, 045, 055-056, 072, 320.0, 771.0, 771.3
Hepatitis and liver cancer*	070, 155
Sexually transmitted diseases*	090-099, 614.0-615.5, 614.7-616.9, 633
Thyroid disease*	240-244
Diabetes*	250, 251.0, 251.2
Dehydration*	276.0, 276.5
Epilepsy*	345, 780.3
E N T infections*	381-383, 461-463, 472.1
Rheumatic fever/heart disease*	390-398
Hypertensive disease*	401-405, 437.2, 276.8
Ischaemic disease*	410, 412, 414, 411.0
Congestive heart failure*	428, 518.4
Stroke*	431, 433, 434, 436
Respiratory infections*	460, 465, 466.0, 480-483, 485-487
CORD*	490-492, 494, 496
Asthma*	493
Dental conditions*	521-523, 525, 528
Peptic ulcer*	531-534
Ruptured appendix*	540
Obstructed hernia*	550.0-550.1, 551-552
Kidney/urinary infection*	590, 599.0
Cellulitis*	680-686
Failure to thrive*	783.3-783.4
Gangrene*	785.4
Burns and scalds	E890-E899
Drowning	E910
Falls from playground equipment	E884.0, E884.5
Indeterminately caused injuries	E980-E989
Poisoning	E850-E869
Road traffic injury	E810-E829
Sports injuries	E886.0, E917.0, E927
Swimming pool accidents	E883.0, E910.5, E910.6
Suicide	E950-E959

^*These conditions are ambulatory sensitive.


Table A1-3: ICD-9 codes used in this report

Condition	ICD-9
Asthma	493
Total cardiovascular disease	390-459
Ischaemic heart disease	410-414
Other forms of heart disease	420-429
Total stroke	430-438
Heart failure	428
Chronic obstructive pulmonary disease (COPD)	491, 492, 494, 496
Rheumatic heart disease	393-398
All revascularisation (CABG and angioplasty) heart disease procedures	360, 361
Diabetes	250
Diabetes complications with renal failure	250.4
Lower limb amputation with concurrent diabetes	841, 250 (together)
All cancer	140-208
Stomach cancer	151
Colorectal cancer	153, 154
Liver cancer	155
Lung cancer	162
Prostate cancer	185
(female) Breast cancer	174
Uterine cancer	179, 182
Cervical cancer	180
All injuries	E800-E999
Motor vehicle traffic	E810-E819
All other transport	E820-E829
Poisonings	E850-E869
Fall	E880-E886, E888
Fires/hot object or substance	E890-E899, E924
Drowning	E910
Suicide and self-harm	E950-E959
Assault	E960-E969
Cut/pierce	E920
Firearm	E922
Machinery	E919
Struck by or against	E916-E917
Suffocation	E911-E913

Statistical methods (standardisation)

Age standardisation allows comparison between ethnic groups with different age distributions. Most of the indicators are presented as age-standardised rates and are expressed as a rate per 100, per 1000 or per 100,000. Direct age standardisation was used in this report, with most rates (unless noted otherwise) standardised to the total Mäori population from the 2001 Census. The 2001 Census Māori population can be found in Appendix 2. For counts less than 5, the age-standardised rate was not calculated.

Standardising to the Māori population provides rates that closely approximate the crude Māori rates (ie, the actual rates among the Māori population) while also allowing comparison to the non-Māori population and the monitoring of inequalities. Readers should note that the use of different standard populations in other reports means that results obtained may differ from those presented in this report, affecting comparability. Also note that some data could not be standardised; these are indicated under relevant tables or figures.

Standardised rate ratios (SRRs) presented in this report are calculated for the Māori population compared to the non-Māori population. Ninety-five percent confidence intervals (CIs) are presented for both rates and rate ratios.

Ethnicity

Most indicators are presented comparing Māori with non-Māori. For data from surveys, prioritised ethnicity was used. A person was classified as Māori if any one of their recorded ethnicities was Māori. So for Māori, prioritised and total response outputs are the same. All other people were recorded as non-Māori. This applies to data derived from the 2002/03 New Zealand Health Survey, 2002 National Nutrition Survey, ASH survey, Youth 2000 Survey and Plunket data.

Official data sets have been shown to undercount Māori (Te Rōpū Rangahau Hauora a Eru Pōmare 2000; Ajwani et al 2003; Cormack et al 2005). Therefore, to improve Māori estimates, analyses of data sourced from NZHIS (including cancer registrations, hospitalisations and mortality) utilised the ‘ever Māori’ method of classification of ethnicity. This involved linking individuals across data sets as outlined below.

Mortality (any causes) from 2000–2002, hospitalisation from 2002–2004, cancer registration from 1999–2001 and a recent copy of the National Health Index (NHI) were obtained from NZHIS. Each record in these data sets has three ethnicity fields, and individuals were classified as Māori if Māori was coded as one of the ethnic groups in any ethnicity field over the time periods described. The remaining people were determined to be non-Māori. View further details on the effect of the ‘ever Māori’ analysis.

The method of ethnicity classification is indicated under each table or figure.

Age groups

For most indicators in the report, the data are presented across all ages in the population. For a selection of indicators, age-specific rates are presented (eg, for meningococcal disease). For all indicators, the relevant age group is noted under each table or figure. Sometimes the term ‘adult’ is used to include the entire population aged 15 years or older.

Ninety-five percent confidence intervals

The results presented have a margin of error. The 95 percent confidence interval (CI) gives an indication of this error. It indicates the interval that has a 95 percent probability of enclosing the ‘true’ value.

The confidence interval is influenced by the sample size of the group. When the sample size is small, the confidence interval becomes wider. Thus, although it may look as if there is no statistically significant difference between Māori and non-Māori, wide confidence intervals would mean we could not exclude the possibility.

When the confidence intervals of two groups do not overlap, the difference in rates between the groups is statistically significant. For example, in the example figure below, the proportion of Māori females stating lack of transport as a reason for not seeing the GP is significantly higher than that of non-Māori females. If the confidence intervals do overlap, the difference could be due to chance (ie, is not statistically significant). The word ‘significant’ is used throughout this report to refer to the 5 percent significance level.

Example figure:


Rate ratios

Age-standardised ratios (see explanation of age standardisation) are used to compare the age-standardised rates between Māori and non-Māori. The rate ratio is equal to the age-standardised Māori rate divided by the age-standardised non-Māori rate. Thus the non-Māori population is used as the reference population. For example, an age-standardised rate ratio of 1.5 means that the rate is 50 percent (or 1.5 times) higher in Māori than in non-Māori, after taking into account the different age structures of these two populations. Rate ratios and their 95 percent confidence intervals are given in the text sections of the document. In this report, if the 95 percent confidence interval of the rate ratio does not include 1, the rates are said to be significantly different from each other.

Ever Māori Analysis

The ‘ever Māori’ method of ethnicity classification was used in the analysis of data on hospital discharges, cancer registrations and death records. This appendix describes the use of the ‘ever Māori’ method in more detail and its impact on Māori mortality, cancer registrations and hospital discharges. Similar methods to those of Cormack et al (2005) were used in this section.

Significant inequalities in health risk and outcomes exist between Māori and non-Māori in New Zealand. Reducing health inequalities that affect Māori is a key Government priority (Minister of Health 2000, Minister of Health and Associate Minister of Health 2002a). High-quality ethnicity data is essential in order to monitor health trends by ethnicity, as well as Government performance to improve health outcomes and reduce health inequalities. It is also needed in order to provide Māori with quality information about Māori health and inequalities (Ministry of Health 2004b).

However, problems have been identified with the collection of ethnicity data. Official health data sets have been shown to undercount Māori in cancer registrations, hospital admissions and deaths (Te Rōpū Rangahau Hauora a Eru Pōmare 2000; Ajwani et al 2003; Cormack et al 2005). There is a need to improve ethnicity data in health information systems. The Ministry of Health has recently released protocols on procedures for the standardised collection, recording and output of ethnicity data for the New Zealand health and disability sector in order to facilitate this (Ministry of Health 2004b).

Until improvements in ethnicity data collection are fully implemented, researchers in Māori health and inequalities have advocated the use of the ‘ever Māori’ method of ethnicity classification (Cormack et al 2005). This has been developed to improve undercounts of Māori in health data sets, and has been shown to produce reasonable estimates of cancer incidence and mortality for the Māori ethnic group for data from 1996 to 2001.


Methods

Cancer registrations from 1999 to 2003 (all cancers 1999–2001; priority cancers 2002–2003), deaths from 1999 to 2002 (any cause), hospital discharges from 1999 to 2004, and a recent copy of the National Health Index (NHI) were obtained from the New Zealand Health Information Service. Each record in these data sets has three ethnicity fields associated with it. Ethnicity fields were pooled by individuals using the encrypted Health Care User Identifier common to all. Where any one of these ethnicity fields had the value ‘21’, the ever-Māori ethnicity indicator was designated as Māori and the remaining people were designated non-Māori.

For the assignment of ‘ever Māori’, the most recently available data were used for each data set back to 1999. More recent data were available for some data sets than others (eg, all cancer data were available up to 2001, while hospital discharges were available up to 2004). Because of this, the years obtained for each data set do not overlap exactly.

In the analyses, the most recent three years of data for each available data source were presented. Table A3-1 shows the effect of the ‘ever Māori’ method assigned using multiple data sets over the 1999–2004 period on the number of Māori deaths, cancer registrations and hospitalisations for the years presented in the profile.

Table A3-1: Māori ethnic group – ever Māori validation using 1999–2004 data

Source				No death registration		Death registration
	Original source	Ever Māori	Increase	Original source	Ever Māori	Original source	Ever Māori
Mortality 2000-2002	7114 (8.55%)	7540 (9.06%)	6.00%
	83,207	83,207
Cancer registrations	3435 (6.38%)	3819 (7.09%)	11.18%	1723	1878	1712	1941
	53,847	53,847		9.00%		9.00%
Public hospital morbidity 2002-2004	291,617 (15.47%)	306,315 (16.25%)	5.04%	288,054	302,502	3563	3813
	1884,650	1884,650		5.02%		7.02%

Results

Mortality (2000–2002)

The ‘ever Māori’ method increased the number of deaths classified as Māori during 2000–2002 from 7114 to 7540, an increase of 6%. This is the same as that found by Cormack et al 2005 for the 1996–2001 period and is close to the 7% undercount identified for the 1996–1999 period by the New Zealand Census – Mortality Study (Ajwani et al 2003). Following the method of Cormack et al (2005) the mortality undercount was used to estimate this method as our standard for estimating the undercount in cancer registrations and hospital admissions.

Cancer registrations (1999–2001)

The ‘ever Māori’ method increased cancer registrations classified as Māori during 1999–2001 from 3435 to 3919, an increase of 11 percent. For public hospital admissions, the ‘ever Māori’ method increased hospitalisations classified as Māori during 2002–2004 from 291,617 to 306,315, an increase of 5 percent.

The ‘ever Māori’ method was used to compare Māori cancer registrations among those people who had a death registered during the time period and those who did not. This was repeated for hospitalisations.

The ‘ever Māori’ method increased Māori cancer registrations between 1999 and 2001 by 9 percent among those people without a death registration, and by 13 percent among those people with a death registration. This indicates that there may still be a residual undercount of Māori cancer registrations among those people without a death registration. Cormack et al (2005) found that cancer registrations increased by 17 percent using the ‘ever Māori’ method among those people with a cancer registration for the years 1996–2001, higher than the 13 percent found here. This may be due to improving ethnicity data collection more recently in cancer registrations. However, it may also be due to the incomplete years of data used in this analysis compared with that of Cormack et al (2005).

Public hospital morbidity (2002–2004)

Similarly for hospitalisations, the ‘ever Māori’ method increased Māori hospitalisations by 5% among those people without a death registration and 7 percent among those with a death registration. Therefore there may still be an undercount of Māori hospitalisations among those people without a death registration.


Summary

The ‘ever Māori’ method of ethnicity data collection is likely to provide reasonable estimates for the number of deaths for Māori. It improves the counts for Māori cancer registrations and hospitalisations, but there is probably still some underestimation of these, as the increase where there is no death record is less than where there is one.