The Right to Be Counted: Unmasking First Nations and Métis Population Undercounts in Winnipeg, Manitoba

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for this work. For me it raises an interesting and important issue: implication of accurate population stats/estimation of population of First Nations peoples for appropriate/equal healthcare resources allocation to local First Nations peoples. I understand that this paper is based on a health project, but I am not quite sure if the paper in its current shape is about health topic that this journal focuses on. For me it is literally about how to estimate the population, rather than health issues, although some very general discussion the connection between population size and health equality. 

My specific comments include: 1) I am bit confused about the methodology used in this paper for population estimation. It would be helpful that you justify it by giving more details, including citing more references, validating how your interview question "Approximately how many First Nations and ...." leading to justification of population census 2021 etc. 2)  In the "Introduction", you cited a number of references about the implication of under-counting population for health service delivery and funding allocation; in Abstract and Discussion, you highlighted this implication, however, there is no such result at all reported in the paper. 

In your revision, you might want to refocus your paper and include more information about how you estimate the population and how this under-counting affects health service.   

For your consideration.

Regards.

Author Response

Thank you for this work. For me it raises an interesting and important issue: implication of accurate population stats/estimation of population of First Nations peoples for appropriate/equal healthcare resources allocation to local First Nations peoples. I understand that this paper is based on a health project, but I am not quite sure if the paper in its current shape is about health topic that this journal focuses on. For me it is literally about how to estimate the population, rather than health issues, although some very general discussion the connection between population size and health equality. 

My specific comments include: 1) I am bit confused about the methodology used in this paper for population estimation. It would be helpful that you justify it by giving more details, including citing more references, validating how your interview question "Approximately how many First Nations and ...." leading to justification of population census 2021 etc. 2)  In the "Introduction", you cited a number of references about the implication of under-counting population for health service delivery and funding allocation; in Abstract and Discussion, you highlighted this implication, however, there is no such result at all reported in the paper. 

In your revision, you might want to refocus your paper and include more information about how you estimate the population and how this under-counting affects health service.

Thank-you for your review.

We focused on the estimation of the size of the First Nations and Metis populations living in the City of Winnipeg, because of the documented inaccuracy of the census counts. In Canada, Census counts drive per-capita health funding (Canada Health Transfer) and every downstream health survey, including the Survey Series on First Nations People, Métis and Inuit (SSFNPMI) so accurate enumeration is a prerequisite for health-equity measurement, reporting, and equitable health service funding,  thereby improving population health outcomes.

In response to the general comment about how the paper responds to the journal’s focus on health issues and health equity, we have strengthened the wording that links accurate population size estimate to population health and Indigenous health equity in the opening paragraph, the second sentence of which now reads: Accurate demographics  are required for health needs assessment, health and social system performance measurement,  and resource allocations, including the Canada Health Transfer.  We additionally note, that the foundational links between accurate population size estimation, population health and health equity are described in the sixth paragraph of the introduction and the second paragraph of the discussion (cut and pasted below).

“The implications of under-counting extend directly to health service delivery and funding. As the Government of Canada notes, “Indigenous peoples are included in the per capita allocations of funding from the federal fiscal transfer and are entitled to access insured provincial and territorial health services as residents of a province or territory” [13]. When populations are systematically under-counted, per capita allocations fail to meet actual needs, exacerbating systemic inequities in healthcare access and quality. Moreover, these systematic undercounts cascade throughout Indigenous health data systems and surveys. The newly launched Survey Series on First Nations People, Métis and Inuit (SSFNPMI) [14] is based on the 2022 Indigenous Peoples Survey (IPS), which in turn is based on the 2021 Census questionnaire. Failure to adequately enumerate Indigenous peoples during the census leads to compounding errors in the reporting of Indigenous health data.”

“The implications of this under-counting ripple throughout data collection, policy development and resource allocation. When populations are underestimated by this magnitude, per capita funding allocations fail to meet actual needs, limiting effectiveness of health systems planning, service delivery and exacerbating inequities in health care. The data problem extends beyond the census and compromises all subsequent health surveillance efforts, including the newly launched Survey Series on First Nations People, Métis and Inuit (SSFNPMI), which relies on census-based sampling frames.”

We have added a sentence on the rationale for our population membership interview question:

“As suggested by the STROBE-RDS guidelines, this question was used to ensure that we were accurately estimating the number of connections each participant had to other members of the target population.”

The full, step by step, derivation of the population size estimate is presented in the statistical supplement, and supported by references for the methodology. The presentation of the population estimate in the main manuscript is less-detailed for reasons of space and clarity.

Our second objective was to discuss of implication of persistent under-counting, as opposed to the measurement of them; we feel that the population estimate itself is the health-relevant finding due to its necessity in health system funding transfers and measuring health inequities.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This paper addresses an important topic in the undercounting of Indigenous peoples in Canada. The authors leverage RDS samples and estimates to obtain a more accurate estimate of the targeted population. Over the paper is well written and timely. The following comments are suggested and mainly address the methods.

 Line 118: get rid of “(insert ref)”

Material and Methods section:

Given the risk of bias associated with RDS convenience samples, clarification is needed on how the seeds were selected and whether it is considered a convenience sample or not.

Additionally, clarification is needed on whether any assumptions were checked for the estimator (Volz-Heckathorn estimator I believe) used to get RDS-II estimates which has been shown to be biased if assumptions aren’t met (Gile KJ, Handcock MS. Respondent-Driven Sampling: An Assessment of Current Methodology. Sociological Methodology. 2010 Aug;40(1):285-327. doi: 10.1111/j.1467-9531.2010.01223.x. PMID: 22969167; PMCID: PMC3437336.)

Author Response

This paper addresses an important topic in the undercounting of Indigenous peoples in Canada. The authors leverage RDS samples and estimates to obtain a more accurate estimate of the targeted population. Over the paper is well written and timely. The following comments are suggested and mainly address the methods.

Thank you for your positive feedback and support.  Our specific responses to your comments included below.

Specific Comments & Response to Reviewer

Line 118: get rid of “(insert ref)”
Thank-you for catching this, it has been removed.

Material and Methods section:

Given the risk of bias associated with RDS convenience samples, clarification is needed on how the seeds were selected and whether it is considered a convenience sample or not.

We have amended the following sentence in our methods: Sampling began with a convenience sample of 12 purposively sampled seeds, chosen to achieve diversity across social networks (age, gender, geography).

Additionally, clarification is needed on whether any assumptions were checked for the estimator (Volz-Heckathorn estimator I believe) used to get RDS-II estimates which has been shown to be biased if assumptions aren’t met (Gile KJ, Handcock MS. Respondent-Driven Sampling: An Assessment of Current Methodology. Sociological Methodology. 2010 Aug;40(1):285-327. doi: 10.1111/j.1467-9531.2010.01223.x. PMID: 22969167; PMCID: PMC3437336.)

We have attempted, where possible, to ensure that the population and the sampling process are aligned with the requirements of the Volz-Heckathorn estimator. Based on our knowledge of the population, we believe it is reasonable to assume a single, inter-connected population. We have revised our results section as follows:

We verified that the population and sampling process were consistent with the assumptions of the Volz-Heckathorn RDS-II estimator (17). The largest recruitment chain had 26 waves (excluding seeds) to ensure sample independence from the seeds. There was no evidence of exhaustive sampling (the average degree remained > 30 in later waves), providing evidence that the population size is sufficiently greater than the sample size for the assumption of the RDS-II estimator requiring N>>n. The average degree was highest among the early waves, supporting the assumption that participants with the most connections have a higher probability of recruitment. Convergence plots indicated gender and age group estimates had stabilized (supplemental figures S1, S2).

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for responding to my comments!