Dear Health New Zealand,

Please provide a machine readable dataset in csv/Excel format that shows the total number of people by age and gender that are unvaccinated with any COVID shot from 1 Jan 2020 through to the present day as at the end of each calendar month.

The columns of the dataset would be expected to include:
1) Date: Date at the end of the month with the earliest date being 31 Jan 2020. That is the first year of data will give a baseline of variation by month that subsequent years that included the vaccination programme can be compared to.

2) Age: Age from birth as at the end of the month. 0 is the age up to 364 days, 1 is from 1 year to 1 year and 364 days etc. The highest age category should be 100+


3) Gender: Male/Female/Unknown or Other

4) Number of People: Count of the population of people under the care/enrolled in the NZ public health system who are unvaccinated with any dose of any of the COVID vaccines. The number in this column will change based on the variation in the enrolled population, the aging of the unvaccinated each month, deaths, and by the number of people transferring into the "COVID vaccinated" group.

Once a query has been written for one month, it should be able to be rerun for each subsequent month on a programmed loop - minimising the effort involved. The code for this OIA has previously been written for a previous OIA by HealthNZ.

Cell obfuscation:
a) Where a cell is <5 people please populate this with "<5" for confidentiality. 
b) True zero cells should be shown - or at least not included in the dataset and not marked as <5. Please state which method has been used.

Please do consult as early as possible if there are any issues or questions.

This is similar in intensity and effort to previous OIAs for data that Health NZ has successfully responded to. A full history is being requested as some cell values may have changed as HealthNZ cleans up its data (eg where a gender is now known but was unknown when a previous OIA was fulfilled).

Yours faithfully,

Chris Johnston

Dear hnzOIA,

Thank you for this response and dataset which I have reviewed.

There appears to be a significant problem with this dataset where the numbers greatly diverge from previous Health NZ OIAs and the sequence of numbers over time do not make sense.

The previous OIA dataset to the end of April 2024 was: Appendix One HNZ00041545.xlsx
The file for this OIA to the end of Sept 2024 is: HNZ00067725 Appendix.xlsx

Could you please review and correct these errors of:
1) Consistency
2) Completeness.

To help you diagnose the problem I draw your attention to the following example for Females that are 12 years old.

April 2024 OIA has the following figures:
31/01/2021 - 31,703 12 year old females with no dose of the vaccine
28/02/2021 - N/S - a completeness issue.
31/03/2021 - N/S - a completeness issue.
30/04/2021 - N/S - a completeness issue.
31/05/2021 - 31,426 12 year old females with no dose of the vaccine

For the April figures I was able to clean up the Health NZ data and interpolate the figures to produce the required analysis, however the OIA for the end of Sept 2024 has used "<6" - preventing the ability to distinguish between low numbers of people in a scenario vs the need to interpolate the figure from the surrounding months (consistency). Ideally the dataset would simply include the correct number of people directly (completeness).

Sept 2024 OIA has the following figures:
31/01/2021 - 32,008 - change seems to be due to classifying people with a previously unknown gender.
28/02/2021 - <6 - where the number we are expecting in 30,000+ !
31/03/2021 - <6 - as above
30/04/2021 - <6 - as above
31/05/2021 - 31,688 - small change due to gender classification.

By my estimate the Health NZ dataset produced to the end of Sept 2024 is missing approx:
- 750,000 people each month in the months February to April 2021
- 600,000 to 500,000 people each month May to October 2021
- 30,000 to 160,000 people each month November 2021 to April 2023
- 20,000 to 30,000 people each month Jun 2021 to April 2024 (except possibly Aug 2023)

Hopefully the above is helpful and you can turn this around promptly.

Yours sincerely,

Chris Johnston

Dear hnzOIA,

Can you please acknowledge the receipt of this request on 5 Nov 2024 for the information in the OIA response to be corrected?

"Dear hnzOIA,

Thank you for this response and dataset which I have reviewed.

There appears to be a significant problem with this dataset where the numbers greatly diverge from previous Health NZ OIAs and the sequence of numbers over time do not make sense.

The previous OIA dataset to the end of April 2024 was: Appendix One HNZ00041545.xlsx
The file for this OIA to the end of Sept 2024 is: HNZ00067725 Appendix.xlsx

Could you please review and correct these errors of:
1) Consistency
2) Completeness.

To help you diagnose the problem I draw your attention to the following example for Females that are 12 years old.

April 2024 OIA has the following figures:
31/01/2021 - 31,703 12 year old females with no dose of the vaccine
28/02/2021 - N/S - a completeness issue.
31/03/2021 - N/S - a completeness issue.
30/04/2021 - N/S - a completeness issue.
31/05/2021 - 31,426 12 year old females with no dose of the vaccine

For the April figures I was able to clean up the Health NZ data and interpolate the figures to produce the required analysis, however the OIA for the end of Sept 2024 has used "<6" - preventing the ability to distinguish between low numbers of people in a scenario vs the need to interpolate the figure from the surrounding months (consistency). Ideally the dataset would simply include the correct number of people directly (completeness).

Sept 2024 OIA has the following figures:
31/01/2021 - 32,008 - change seems to be due to classifying people with a previously unknown gender.
28/02/2021 - <6 - where the number we are expecting in 30,000+ !
31/03/2021 - <6 - as above
30/04/2021 - <6 - as above
31/05/2021 - 31,688 - small change due to gender classification.

By my estimate the Health NZ dataset produced to the end of Sept 2024 is missing approx:
- 750,000 people each month in the months February to April 2021
- 600,000 to 500,000 people each month May to October 2021
- 30,000 to 160,000 people each month November 2021 to April 2023
- 20,000 to 30,000 people each month Jun 2021 to April 2024 (except possibly Aug 2023)

Hopefully the above is helpful and you can turn this around promptly."

Yours sincerely,

Chris Johnston

Dear hnzOIA,

Thanks for your reply.

The explanations provided are understood.

I note this paragraph in HealthNZ's reply:
"In the output provided back in April 2024, we differentiated between small
number suppression and secondary suppression by using “<5” and “N/S”
respectively, however, when the outputs were run again in September 2024,
this differentiation did not occur and “<6” was used for both small number
suppression and secondary suppression."

For clarity so that the public know which cells have been suppressed in which way - can the dataset please be provided again with the differentiation "<6" and "N/S". The differentiation is important and the label provided is misleading - and incorrect on its face when N/S is applicable (because the number is actually more than 5).

Please include the explanation of each suppression in the reference notes/legend provided in the spreadsheet.

Yours sincerely,

Chris Johnston