Reducing Ethnic and Geographic Inequities to Optimise New Zealand Stroke Care (REGIONS Care): Protocol for a Nationwide Observational Study

Background Stroke systems of care differ between larger urban and smaller rural settings and it is unclear to what extent this may impact on patient outcomes. Ethnicity influences stroke risk factors and care delivery as well as patient outcomes in nonstroke settings. Little is known about the impact of ethnicity on poststroke care, especially in Māori and Pacific populations. Objective Our goal is to describe the protocol for the Reducing Ethnic and Geographic Inequities to Optimise New Zealand Stroke Care (REGIONS Care) study. Methods This large, nationwide observational study assesses the impact of rurality and ethnicity on best practice stroke care access and outcomes involving all 28 New Zealand hospitals caring for stroke patients, by capturing every stroke patient admitted to hospital during the 2017-2018 study period. In addition, it explores current access barriers through consumer focus groups and consumer, carer, clinician, manager, and policy-maker surveys. It also assesses the economic impact of care provided at different types of hospitals and to patients of different ethnicities and explores the cost-efficacy of individual interventions and care bundles. Finally, it compares manual data collection to routine health administrative data and explores the feasibility of developing outcome models using only administrative data and the cost-efficacy of using additional manually collected registry data. Regarding sample size estimates, in Part 1, Study A, 2400 participants are needed to identify a 10% difference between up to four geographic subgroups at 90% power with an α value of .05 and 10% to 20% loss to follow-up. In Part 1, Study B, a sample of 7645 participants was expected to include an estimated 850 Māori and 419 Pacific patients and to provide over 90% and over 80% power, respectively. Regarding Part 2, 50% of the patient or carer surveys, 40 provider surveys, and 10 focus groups were needed to achieve saturation of themes. The main outcome is the modified Rankin Scale (mRS) score at 3 months. Secondary outcomes include mRS scores; EQ-5D-3L (5-dimension, 3-level EuroQol questionnaire) scores; stroke recurrence; vascular events; death; readmission at 3, 6, and 12 months; cost of care; and themes around access barriers. Results The study is underway, with national and institutional ethics approvals in place. A total of 2379 patients have been recruited for Part 1, Study A; 6837 patients have been recruited for Part 1, Study B; 10 focus groups have been conducted and 70 surveys have been completed in Part 2. Data collection has essentially been completed, including follow-up assessment; however, primary and secondary analyses, data linkage, data validation, and health economics analysis are still underway. Conclusions The methods of this study may provide the basis for future epidemiological studies that will guide care improvements in other countries and populations. International Registered Report Identifier (IRRID) DERR1-10.2196/25374

The applicants document ethnic inequalities in stroke outcomes in New Zealand, and outline their plan to determine whether they represent inequities in the delivery of health care. They point out the need to correct for possible effects of disparities in geography, co-morbidities and socioeconomic factors.
Their arguments are presented clearly and succinctly. Their ultimate goal, evidence-informed NZ stroke strategy, should be advanced by this proposed study as planned.

Design and Methods
Score: 5 Hypotheses and Aims are stated clearly. Studies have already shown stroke service size and location impacts on access to optimal stroke care and patient stroke outcomes. However, specific geographic peculiarities unique to NZ may be important here. The impact of ethnicity is self-evidently important to study.
The design of the proposed study presents a comprehensive approach to the issue at hand. The prospective portion of Part 1 should provide an accurate snapshot of present stroke service treatment patterns, outcomes and costs. The registry data broadens this to a nationwide perspective. Part 2 is designed to identify potential barriers to optimal care access and quality.
I have a number of concerns about the Methods: Part 1a: Although I agree that both ischemic and hemorrhagic stroke patients probably suffer from the same inequalities, including the latter adds a layer of stratification to the analysis that is probably unnecessary. At least in the US, 87% of all strokes are ischemic. Adequate numbers of ischemic strokes should therefore be available to achieve adequate cases for statistical analysis.
The use of both final mRS scores and score change in defining favourable outcome is justified. However, using a simple dichotomy robs the study of considerable statistical power (see below). There may be the need for family consent in cases where the stroke vistim is unable to give informed consent. If no provisions are made, the study risks a bias towards mild stroke.
Part 1b: I have the same concerns about including hemorrhagic strokes in the mix and in a simple dichotomous primary outcome. I like the choice of 'best practice stroke care' as a secondary outcome.

Part 1 Data Linkage & Analysis:
The applicants admit that the small numbers make it "challenging to recruit a sufficient sample…" This problem will only be exponentially increased for the stratified analyses. More robust approaches, such as sliding dichotomy, ordinal logistic regression, etc. have been shown to add statistical power and reliability to TBI and stroke studies (1,2) using ordinal outcome scales. I strongly suggest the applicants adapt one of these approaches.
I am not an epidemiologist and may not be using the correct terms, but I suspect the applicants rum the risk of ascertainment (sampling) bias when comparing different geographic areas. Surveys such as this tend to find very low or very high frequencies when sampling locales with sparse populations and few observations. The applicants should weight observations of small finite populations or correct estimates of proportions.

Health significance Score: 6
Treatment of stroke has an enormous public health and economic impact. If successful, this study promises to identify impediments to appropriate stroke care in NZ. Evidence discovered from the proposed study can be used to recommend measures to maximize both the quality and the cost-effectiveness of care.
The applicants provide convincing evidence that inequalities exist in NZ in stroke care and outcomes, that the proposed study can address the causes of the inequalities and that correcting those causes is likely to improve outcomes and reduce the cost of care.

Research Outcomes Score: 7
Dr Anna Ranta, PI of the proposed project, is young but has spent her career specializing in delivery of stroke care. Recipient of several grants and author of several articles on stroke, she possesses the knowledge, the incentive and the skill to see this project through to successful completion. Her team expresses diversity in representation, both geographically and in terms of expertise. The members are from all areas of New Zealand and include experts both in acute stroke care and rehabilitation, epidemiology, public health, Maori medical issues, health care economics. Also included are a stroke care field worker, 2 study planners and coordinators and a stroke survivor. The proposed time commitment by team members is adequate to achieve the study goals.