Ethnicity Data Quality

Question 1

What is the name of the ethnicity protocols?

Answer 1

HISO 2017 Ethnicity Data Protocols

Question 2

How many Māori are undercounted in the health sector?

Answer 2

One in five Māori (21%) are not identified as Māori in the National Health Index, compared to the self-identified Census.

Around 30% are missed for Māori males and those Māori aged 20-24 years.

Question 3

Why is good quality ethnicity data important in the health sector?

Answer 3

• Adequate funding allocation to Māori needs (rather than under-funding)
• Delivery of targeted health services to Māori needs (rather than under-servicing)
• Measure Māori health needs
• Monitor the health system’s performance
• Māori workforce representation

Question 4

How is ethnicity defined?

Answer 4

Ethnic group/s that people self-identify with or feel they belong to.

Cultural affiliation with unique interests, customs or language, a common ancestry and geographic origin.

Question 5

Why do we use ethnicity in health?

Answer 5

Ethnicity data allows us to have a more comprehensive and complete understanding of people’s health experiences and outcomes.

It is an important factor in developing responsive policies and procedures and providing appropriate healthcare.

Historical and political contexts shape the way in which ethnic groups are conceptualised, defined and measured.

Different from race (physical characteristics), descent (biological ancestry), nationality (country), citizenship, and culture.

Question 6

What is ethnicity?

Answer 6

Social category influencing people’s experiences and outcomes.

Ethnicity is relevant to the understanding and measurement of health because the biggest influence on inequities are social determinants.

Self-identification is important as it aligns with self-determination and the right for Māori to name ourselves.

Question 7

Why do we need to ask ethnicity in a standardised way?

Answer 7

Changes to the question can impact the results (such as wording, layout, options).

It has implications for monitoring trends and inequities over time.

Multiple ethnicities (identify with multiple ethnic groups) and ethnic mobility (changes over time).

Example: 1996 Census change of wording encouraged more individuals to identify with multiple ethnic groups.

Question 8

Who collects ethnicity data?

Answer 8

1. Primary and secondary care - National Health Index (ongoing)
2. Hospitals - National Health Index (ongoing)
3. Stats NZ - Census (5 yearly)
4. Central health agencies - Workforce surveys and NZ Health Survey
5. Birth and death registrations (at time)

Question 9

What is the action plan for improving ethnicity data quality?

Answer 9

1. Leadership
2. Knowledge, education and training
3. Compliance
4. Accountability and monitoring
5. IT systems improvement

Question 10

In the Primary Care Ethnicity Data Audit Kit, what is Stage 1?

Answer 10

Systems compliance audit checklist

A primary care provider can complete the checklist to find out whether their systems comply with capturing high quality ethnicity data. There is a clear criteria to calculate scores and address issues. To be completed by someone with a good understanding of the process to input ethnicity data.

Question 11

In the Primary Care Ethnicity Data Audit Kit, what is Stage 2?

Answer 11

Staff survey

Staff are distributed a survey to complete within a primary care provider. Each survey is marked and scored. Results are distributed along with any quality improvements and training requirements.

Question 12

In the Primary Care Ethnicity Data Audit Kit, what is Stage 3?

Answer 12

Ethnicity data quality audit

A sample of 100 consecutive patients is collected from a primary care provider. The data is compared using the Patient Management System to identify the level of match (match, partial and total mismatch). Results are calculated to identify any quality improvements.

Question 13

What are the protocols for collection and confirmation?

Answer 13

Paper form, electronic form, telephone, interpreter.

Collected at any time but must be at least every 3 years. Confirmed at any time.

Collector must not guess, correct or complete on their behalf.
Collector must not ask to choose a principal ethnicity.
If it has not been filled in, collector should ask to confirm whether they have left the question intentionally. Electronic form should not allow the question to be left blank.
Links to more information about ethnicity if electronic form.
Can have a “I don’t know” or “I do not want to state”

Assisted response using an aid or proxy response using a nominee or next of kin or guardian.

There is no legal or recommended age, when the child is capable of understanding the concept of ethnicity.

Question 14

What is the ethnicity question?

Answer 14

Rigorously tested by Stats NZ.

“Which ethnic group do you belong to? Mark the space or spaces that apply to you.”

List of ethnicities must all be present and, in order shown in the graphic.
- NZ European
- Māori
- Samoan
- Cook Island Māori
- Tongan
- Niuean
- Chinese
- Indian
- Other. Please state:

No additional categories may be added.

Preferably that the categories are listed vertically.

Any collection method must allow multiple ethnic groups to be selected and must allow multiple ethnic groups to be entered in the ‘other’ section.

Question 15

What are the protocols for classifying, recording and storing?

Answer 15

Recorded at Level 4 (most detailed level).

IT must be capable of recording up to 6 ethnicities. Process for reducing if more than 6.

New Zealander has a separate code.

Residual codes (don’t know, refused to answer, outside of scope, response unidentifiable.)

Iwi is coded as Māori.

Question 16

What are the protocols for analysis and output?

Answer 16

Comparable data about groups of interest.

Aggregation of data in the output.

Any output requires clear description of the method and categories used, and limitations.

Where there are multiple ethnicities, the below methods should be used:
- Total response (overlapping)
- Prioritised
- Sole/combination

Question 17

Total response

Answer 17

Each respondent is counted in each of the ethnic groups they reported.

The sum will exceed total population.

Level 1 aggregation.

Most common along with prioritised.

Question 18

Prioritisation

Answer 18

Each respondent is allocated to a single ethnic group using the prioritisation table.

The sum will equal the total population.

Denominator data must also be prioritised.

Level 1 and 2 aggregation.

Inconsistent with the concept of self-identification.

Most common along with total response.

Question 19

Sole/combination

Answer 19

Sole ethnic categories for respondents who report only one.

Combination categories for respondents who state more than one.

The standard Stats NZ single/combination minimum output has nine groups: European, Māori, Pacific Peoples, Asian, Other, Māori/ European, Māori/Pacific Peoples, ‘Two groups Not Elsewhere Identified’ or the category titled ‘Three groups’.

Question 20

What are the underlying principles for high quality ethnicity data?

Answer 20

1. Māori have a right to be counted and therefore, valued.
2. Māori have the right to monitor the Crown.
3. Māori have indigenous rights as tangata whenua to be a priority within ethnicity data analyses.
4. Māori rights to mana motuhake (self-determination) require indigenous data sovereignty principles to be upheld.
5. The Crown requires high quality ethnicity data and analyses for Māori to uphold its obligations under Te Tiriti o Waitangi.
6. Values and positioning should be framed to support equity and Māori indigenous rights.

Question 21

What is the first step to analyses?

Answer 21

Clear understanding of the purpose of the analysis, what questions need to be answered

Following this will determine the best approaches or methods for answering these questions, including the most appropriate sources of information.

The nature of datasets, data quality (including ethnicity data) and availability, reach of the dataset

Presentation of any data should include the data source, quality and impact of these on the presented estimates, methods to address missing or poor data quality

Question 22

Māori population

Answer 22

Examination of a range of information:

Number of Māori in the population (total, regional, local) and their demographic characteristics

Stats NZ - population census, ERP and population projections

Measuring Māori health status and need

Question 23

Estimated Resident Population

Answer 23

Numbers from the associated census adjusted for migration, births and deaths, and missing data

Stats NZ

Currently the best estimate of the Māori population - best measure for size and age-sex structure.

Based on the respective census counts, with adjustments for census non-response (people who did not complete census forms), net census undercount (people who were missed in the census), residents temporarily overseas on census night (RTO), and estimated population change between census date and mid-year.”

Question 24

Māori population projections

Answer 24

Best for planning

Stats NZ

Projected Resident Population

Question 25

Wider determinants of health

Answer 25

Measuring Māori health status and need as health is also driven by wider deteminants of health.

Education, income, employment, deprivation, racism.

Common sources come from Stats NZ, MoE, NZ Health Survey, Te Kupenga.

Question 26

Health service access and quality

Answer 26

Health service performance

PHO enrolments, hospitalisations, cancer registry, PHARMs, PRIMHB (mental health), immunisation.

ASH is a subset of the broader category of PAH

Amenable mortality (timely access to high quality care) and preventable mortality (primary prevention)

Health workforce

Māori rates compared to targets

Patient reported unmet need and satisfaction (NZHS and HQSC)

Question 27

Global measures of health

Answer 27

Reducing inequities

Life expectancy (StatsNZ), mortality rates (StatsNZ), HALE, YLL

Across all diseases, health care and determinants

Question 28

Specific diseases

Answer 28

• How common are different diseases for Māori?
• At what level of disease (e.g., stage) are Māori being diagnosed?
• Which diseases cause the most deaths in Māori?
• How do the above vary by age group, gender, region, SES?
• How are these changing over time?

Question 29

Numbers, proportions and crude rates

Answer 29

Most population: numbers that have been age-standardised and compared to other ethnic groups

“which diseases are the most common for Māori?” or “which diseases cause the greatest numbers of deaths for Māori?”

Rankings: most common registrations and greatest number of deaths

Question 30

Stratification of Māori data

Answer 30

Examine patterns of health and healthcare use within the Māori population by factors such as age, gender, location, and deprivation.

Who is being affected by conditions or using services and who is not.

Ethnicity data quality varies by age and gender so should be considered

Question 31

Age standardisation and trends over time

Answer 31

In addition to understanding which diseases are the most common or have the highest mortality at a single point in time, it is useful to look at trends over time.

The Māori population is ageing over time, increasing the numbers and proportions of Māori in older population groups. Diseases that occur in these older age groups will become more common.

To assess whether there is a greater risk of a disease over time, it is necessary to remove the effect of the ageing population from the analysis.

Assume the same age pattern at each point in time. Usually not real rates of disease.

Māori 2001 population standard - as the population gets olders, becomes less representative. However, calls to update must be balanced against the interruption of time series.

Māori have a younger population age structure to nM. Therefore, age standardisation should be applied to comparator groups.

Question 32

Measuring inequities

Answer 32

All of the methods to measuring health need, access and quality can be compared to another population group to assess inequities.

Comparison of Māori compared to Pākehā, has been criticized as implying that Māori should be seeking to achieve a Pākehā standard. Whether Māori rights to health are being met and indicates the responsiveness of the health system and wider government to address Māori health needs as a right, shifts the focus away from blaming Māori for poor health to highlighting the role of systems in generating disadvantage and privilege. Colonialism and racism as the basic causes of Māori inequities.

Question 33

Comparators

Answer 33

Range of methods to grouping or categorising ethnicity data

Aggregating groups are prioritised, total response or sole/combination. Depends of the questions being asked.

Example - healthcare privilege
Numerator = prioritised (Māori)
Denominator = sole combination (NZ European)

Comparator groups should not be overlapping (total European or total NZ)

1. have clear rationale for the comparator group selected, including in relation to Te Tiriti o Waitangi,
2. select a comparator group that does not overlap with Māori (i.e., is mutually exclusive),
3. understand who is represented in the comparator group, and how this may impact on measured inequities, and
4. understand data quality limitations (including misclassification of ethnicity and missing data) and potential impacts on measured inequities.

Question 34

Māori/non-Māori analysis

Answer 34

The use of Māori/non-Māori analyses acknowledged the fundamental nature of our relationship with the Crown affirmed in Te Tiriti o Waitangi

Non-overlapping comparison group

non-Māori includes ethnic groups which may skew the data and underestimate the inequities (Pacific).

Question 35

Māori/sole European analysis

Answer 35

European as the most socially privileged population

Consistent with a Te Tiriti approach in relation to power and equity).

Relatively clear who is represented in the sole European comparator group.

When sole European is not available, could do nM non Pacific

Question 36

Usually resident population

Answer 36

Count of all people who usually live in NZ or in an area of NZ and who are present on a given census night.

Does not include NZ citizens who are overseas or temporary visitors.

Ethnicity is sourced, in order of priority, from: self-identified ethnicity from the 2018 Census (84.4%); admin enumerated from the 2013 Census (8.2%); admin enumerated from other administrative data sources (6.2%) e.g., birth registrations and health.

Question 37

Health service user

Answer 37

Little documentation about it, no evidence of Maori input into the development (breach of Maori data sovereignty). Lack of information of data quality and limitations.

Subset of the NHI

Estimate of the population using the health system in NZ in a given time period (previous 12 months).

1. Use health services
2. Enrolled in a PHO
3. Received a COVID-19 vaccination

Maori disproportionately represented in the group that is within ERP but not using health services.

Not Te Tiriti compliant for analysing Maori health and equity as it does not measure the total population, excluding Maori who have not had access to services.

Miss classification (under counting) of Maori who have accessed health services and excludes peoples who have not accessed services.

Cannot be used for time-series data because of regular updating and changing size. Also has a 6 month lag for availability.

Benefit: HSU is updated annually which means regular access than the 5 yearly census. Avoidance of numerator-denominator bias.

Question 38

Numerator / denominator

Answer 38

Numerator: number of times an event has occurred in a population

Denominator: choose an appropriate denominator usually related to the population or the total events.

Question 39

IDI Estimated Resident Population

Answer 39

The Integrated data infrastructure (IDI) can be used to create an IDI-ERP which is a subset of “currently resident” individuals linked to the IDI spine. It includes individuals known to be alive and based in NZ (using deaths and migration data), with activity in the preceding year within any of health, ACC, taxation, and education datasets; and for those under the age of 5 years, a birth registration or visa approval (excluding visitor and transit visas) and when the record is linked in the IDI spine. The IDI-ERP has been measured to be 2% larger than the official population estimate, with variations by age and ethnicity.

Access to IDI is limited to those with access to StatsNZ accredited data lab and expertise which is a concern to Maori data sovereignty, lack of consent for collected data to be included in the IDI, risk to personal privacy and risk of harm from inappropriate framing.

Little documentation of the methods and limitations of the dataset.

Ethnicity data is source-ranked