The COVID-19 pandemic and acute rheumatic fever inequities, Aotearoa New Zealand, 2020–2022

During 2020–2022, the time period characterised by significant COVID-19 public health measures in Aotearoa New Zealand, a marked decrease in reported hospitalisations and public health notifications of acute rheumatic fever (ARF) was observed.¹ The reasons for the decline are not well understood and, subsequently, reported ARF notifications for 2023–2025 have increased back to pre-pandemic levels.¹

ARF is a preventable, immune-mediated disease that develops in a small proportion of people following Group A Streptococcal (Strep A) infection; although it was historically associated with pharyngitis, increasing evidence suggests Strep A skin infection is also a trigger.² Rheumatic heart disease (RHD), the most severe sequelae of ARF, results in permanent heart disease, which can lead to significant morbidity and premature mortality.³ Although rare in most high-income countries, ARF and RHD are significant health issues in Aotearoa and Australia.³

Rates of ARF are disproportionately high in Māori and Pacific peoples in Aotearoa, particularly children aged 5–14 years living in socio-economically deprived areas.⁴ Crude ARF hospitalisation rates (<30 years of age) are nearly 20 and 45 times higher for Māori and Pacific peoples respectively compared with non-Māori, non-Pacific populations, which have a crude incidence of less than one per 100,000 population per year.⁴ Household crowding and barriers to accessing primary healthcare are important modifiable risk factors for ARF prevention.⁵

In Aotearoa, ethnic inequities in ARF are not accidental but reflect the effects of colonisation, privilege and racism.⁶ Globally, colonisation has exploited and oppressed Indigenous peoples and other marginalised groups, enabling the accumulation of economic benefits and wealth by colonial settlers. Power hierarchies, created through colonialism and racial capitalism,⁷ result in the inequitable distribution of risk and protective factors for ARF, and in large and persistent ethnic health inequities in Aotearoa. Colonial structures produce institutions, policies and systems that create intergenerational barriers to determinants of health,^6,7 such as accessing quality housing and healthcare, leading to increased incidence of ARF.⁴

Like other countries implementing a COVID-19 elimination approach, international border controls, physical distancing and lockdowns (stay at home orders) were important parts of the elimination strategy implemented in Aotearoa in 2020–2021,⁸ but also raised concerns about increasing barriers to accessing healthcare and exacerbating ethnic health inequities.⁹ Lockdowns occurred at national and regional levels, with prolonged Auckland and Northland regional-level lockdown in 2021 (Delta variant outbreak, 17 August 2021–2 December 2021). Measures successfully eliminated community transmission until the Delta outbreak but also created significant barriers to healthcare access.^10,11

At the end of 2021, the COVID-19 suppression strategy signalled a change to reducing rather than eliminating transmission. Lockdown restrictions as seen in the elimination strategy were discontinued, some border measures lifted and the stringency of public health and social measures reduced.⁸

Measures eased further in early 2022, with increasing COVID-19 cases in the community. The mitigation strategy focussed on protecting the health system, retaining testing and isolation requirements as well as some public health measures.⁸ Border restrictions were removed in stages and all measures were discontinued by September 2022.

A limited number of studies have examined trends in ARF during the COVID-19 pandemic;^12–16 however, we are not aware of analyses of ethnic inequities during this time. This study aimed to explore the impact of the COVID-19 public health measures on ARF incidence and ethnic inequities within the broader context of large and persistent health inequities for Māori and Pacific peoples in Aotearoa, 2000–2022.

Methods

This study is part of Whāia te Māramatanga, a larger Kaupapa Māori epidemiological study of ARF and RHD hospitalisations and mortality in Aotearoa. A Kaupapa Māori approach provides a rights-based framework for a critical structural analysis of ethnic inequities in ARF, supporting quantitative research that recognises the impacts of, and resists, continued colonisation.¹⁷ Research has been Māori-led, with Kaupapa Māori theory informing the research question, analytical approach, interpretation and solutions identified. Ethics approval was obtained from Auckland Health Research Ethics Committee (AHREC ref. AH26094).

Data sources and definitions of key variables

We obtained data on ARF diagnoses covering 1988–2022 from the National Minimum Dataset to include all available public hospital discharge information available at the time. An encrypted National Health Index number was provided for each event, along with demographic information. Initial ARF hospitalisations and other variables were defined as described in Appendix Table 1. We limited inclusion to people <35 years in our analyses after examining the incidence curves by ethnicity and 5-year age groups up to 75 years of age. ARF is infrequent after this age.

A denominator dataset was created from Stats NZ Census datasets by 5-year age group, sex, prioritised ethnicity, district areas and area-level socio-economic deprivation (New Zealand Index of Deprivation [NZDep] 2013 quintile). Denominator populations were interpolated linearly for 2000–2020 and held constant for 2021–2022 to reflect minimal net migration during the period of stringent border restrictions (correlating with Health New Zealand – Te Whatu Ora current population estimates).

Time periods

We analysed the data by month, year and COVID-related time periods (Table 1). We defined COVID-related periods a priori to explore the impact of international border restrictions and other measures on ARF hospitalisation. The year 2000 was assigned as the reference year, being neither unusually high nor low and prior to any nationally coordinated ARF prevention activities.

View Table 1–3, Figure 1–3.

Statistical analysis

Statistical analyses were based on monthly counts of initial acute ARF hospital admissions (<35 years), aggregated by year, month, prioritised ethnicity, sex, 5-year age group, health district and NZDep2013 quintile. Months and strata with no ARF admissions were retained in the analysis by explicitly adding records with zero counts, ensuring exposure time was correctly represented and avoiding zero-truncation of the outcome. The primary response variable for modelling was the count of initial ARF admissions per month per stratum.

To evaluate the impact of COVID-19 public health measures on ARF incidence, we contrasted observed trends with counterfactual estimates from a multivariable Poisson regression model under alternative COVID-19 policy scenarios. COVID-19 exposure variables were constructed as indicator vectors. For example, a national “any lockdown” variable coded 1 for months in which a national or regional COVID-19 lockdown was in place in the relevant health district, and 0 otherwise. By manipulating these indicator variables while holding other covariates at their observed values, we obtained counterfactual monthly ARF rates for COVID-19 time period scenarios, which were then compared to estimate the association of these measures with ARF hospitalisation.

We fitted generalised linear models with a Poisson distribution and logarithmic link function to estimate adjusted rate ratios (aRRs) for ARF hospitalisation. Further description of modelling methods is provided in Appendix Table 2. Statistical significance was assessed at a two-sided α level of 0.05, and 95% confidence intervals (CI) were calculated for all aRRs.

The best-fitting model included main effects for elimination and suppression time period combined, any lockdown, health district, month, age group, ethnicity, sex, year and NZDep2013 quintile, together with interaction terms age group × ethnicity, any lockdown × year, ethnicity × NZDep quintile, ethnicity × sex, and ethnicity × year. All other prespecified two-way interactions were fitted and tested but were not retained because they worsened model fit (higher Akaike information criterion [AIC]) and showed no evidence of interaction (P>0.05).

All analyses were conducted in R (version 4.3.0). Multivariable Poisson regression models were fitted with the lme4 and lmerTest packages, and model assumptions (including dispersion, residual structure and influential observations) were evaluated using the DHARMa package.

Results

Overview of ARF hospitalisations, 2000–2022

During 2000–2022, 3,546 hospitalisation events (all ages) were coded with ARF as the principal diagnosis and 3,236 (91.3%) were initial hospitalisations. Of these initial ARF hospitalisations, 94% (3,042) were <35 years of age (Appendix Figure 1) and are included in the analyses of the period 2000–2022. Crude rates of hospitalisation for ARF <35 years in Māori and Pacific peoples were 14.7 and 32.3 per 100 000 respectively. When adjusted for age, sex, health district, socio-economic deprivation and time periods, Māori were 9.5 times and Pacific peoples 25.5 times more likely to be hospitalised for ARF than non-Māori, non-Pacific peoples (NMNP) (Appendix Table 3).

Of all ARF hospitalisations, 73% of people were aged between 5 and 14 years of age. Adjusted hospitalisation rates for children aged 5–9 and 10–14 years were more than 15 and 22 times higher respectively compared with children aged 0–4 years (the reference group; Appendix Table 3). The highest crude rates were in the 5–9- and 10–14-year-old age groups for Māori (26.2 and 41.4 per 100,000) and Pacific peoples (55.2 and 91.0 per 100,000). Hospitalisation rates varied across time by ethnic groups, as illustrated in Figure 1.

In the period 2000–2022, over half (53%) of all ARF hospitalisations in Aotearoa were in the wider Auckland Region (Auckland, Counties Manukau, Waitematā; Appendix Table 3). ARF hospitalisation rates increased progressively with increasing area-level socio-economic deprivation for all ethnic groups (Figure 2). When adjusted for age, sex, health district, time periods and prioritised ethnicity, this socio-economic gradient persisted. Those living in NZDep2013 quintile 5 (most deprived) were four times more likely to be hospitalised for ARF compared with quintile 1 (least deprived; aRR 4.02, 95% CI 3.13–5.17, P<0.001). This “deprivation gradient” was steeper for Māori (aRR 6.13, 95% CI 4.18–9.0) and for NMNP (aRR 4.00, 95% CI 2.52–6.36) than for Pacific peoples (aRR 2.65, 95% CI 1.7–4.13; Appendix Table 4), although small numbers in some quintiles limit interpretation.

Initial ARF hospitalisation and COVID-19 time period analysis

There were 434 initial hospitalisations for ARF, aged <35 years, during 2019–2022 (Table 2). From 2020 to 2021, crude rates were relatively stable for Māori but trended downwards for Pacific peoples. In 2022, the relative risk of hospitalisation increased for Māori (aRR 45) and even more so for Pacific peoples (aRR 77) compared with NMNP, although numbers are small and CIs wide.

There was an estimated 50% reduction in risk of hospitalisation during the elimination–suppression periods combined compared to expected rates for the period (aRR 0.47, 95% CI 0.33–0.69, P<0.001; Table 3). There was insufficient evidence to support an independent association with “any lockdown” in 2020–2021 combined (aRR 0.56, 95% CI 0.30–1.06, P=0.076). However, the association differed when examined by individual year. There was a positive association between “any lockdown” and risk of hospitalisation in 2020 (aRR 4.1, 95% CI 1.9–8.9, P<0.001) and no association in 2021. When “any lockdown” period was combined with the elimination–suppression period in 2020 there was no difference in hospitalisation rates compared to expected (aRR 1.10, 95% CI 0.12–9.74, P=1.000), whereas in 2021 there was a 70% reduction in risk (aRR 0.27, 95% CI 0.13–0.56, P<0.001). There was no evidence to support an interaction between ethnicity and elimination–suppression time period.

Figure 3 shows adjusted estimate rates by month and year during 2019–2022 for three COVID-19 scenarios: i) elimination–suppression strategy (red bars), ii) no elimination–suppression strategy or any lockdown (green bars), and iii) elimination–suppression strategy and any lockdown (blue bars). The difference between the height of the bars indicates the estimated difference in rates between each scenario. In 2020 the difference between blue and red bars is positive; in 2021 the difference is negative. Adjusted estimates of ARF rates in 2022 are low compared with 2019, the year prior to the three COVID-19 scenarios.

Discussion

This Kaupapa Māori epidemiological study provides novel insights into the reported decrease in ARF hospitalisations in Aotearoa during the time of COVID-19 public health measures in 2020–2022. While we found strong evidence for a reduced risk of initial ARF hospitalisation during the period of more stringent COVID-19 public health measures (elimination–suppression approaches), the relationship with lockdowns is more nuanced. In 2020, the reduced hospitalisation rate associated with the elimination–suppression period was counteracted by that of the first national lockdown. In 2021, the reduction in hospitalisation rate associated with elimination–suppression periods was increased further with “any lockdown”, resulting in a greater reduction in estimated rate of ARF hospitalisation. Hospitalisation rates were reduced in 2022 overall, particularly for Pacific peoples, the year when significant COVID-19 public health and social measures were discontinued.

Interpretation of these complex associations occurs within a colonial settler society where racism and coloniality create and maintain ARF inequities for Māori and Pacific peoples in Aotearoa. Ideas of racial hierarchy have been inscribed into institutions, policies and practices, resulting in differential access to resources and opportunities by ethnicity.⁶ Land dispossession, socio-economic deprivation and exclusion from political institutions drive quality housing inequities and differential healthcare pathways and experiences.¹⁸ Māori and Pacific peoples experience systemic failures within the health system, which increase the risk of acute and recurrent rheumatic fever.^18,19 Although COVID-19 public health measures were implemented at regional and national levels, implications for ARF causal pathways vary by ethnic group as a consequence of structural inequities.

Socio-economic gradient trends reported in our study are a reminder that the Māori and Pacific experience in Aotearoa is not homogenous. The flatter gradient for Pacific peoples suggests elevated risk of ARF across all socio-economic deprivation quintiles. Other system-level factors linked with migration/new migrant status may increase risk of ARF independent of socio-economic deprivation, which merits further research.

The persistence of high ARF rates for Māori and Pacific peoples in 2020 likely reflects pre-COVID Strep A community transmission and the characteristic lag between Strep A infection and onset of ARF. The lag time of around 3–12 weeks²⁰ extends the risk of ARF into mid-2020, the time after border closures were introduced and reductions in Strep A strains reported.²¹ In 2021, the risk of ARF hospitalisation reduced during the elimination–suppression period, with an additive effect during lockdowns. At this time, Strep A transmission was likely suppressed to low levels as a consequence of border closures and other public health measures that created physical distancing. A recent study by Bennett et al. found a temporal relationship between reduced Strep A skin and throat swab positivity and ARF notifications from 2020 to 2022.¹⁶

Despite the reduction in circulating Strep A, large ethnic inequities in ARF remain. In 2021, nearly all cases occurred in Māori and Pacific peoples (94%). Risk of hospitalisation was more than 10 and 16 times that of NMNP respectively, suggesting that structural drivers of inequities persisted. Household crowding and barriers to accessing primary healthcare were exacerbated during the time of stringent public health measures,^10,22 which may have increased ARF risk for Māori and Pacific whānau and communities where Strep A strains were still circulating.

Barriers to timely Strep A diagnosis and treatment increased during 2020–2022 and delayed presentations of ARF were reported.²³ There were large reductions in school-based throat swabbing services in Auckland,²⁴ which are targeted to reach Māori and Pacific children. Community antibiotic use, greatest for Pacific peoples, reduced during and after the national lockdown in 2020 and is reported to reflect both reduced access to primary healthcare and decreased circulating pathogens.²⁵ A larger decrease was reported in Auckland, the region with the longest cumulative time in lockdown,²⁶ and where more than half of ARF hospitalisations occurred. COVID-19 policies and processes magnified barriers to accessing healthcare for Māori, Pacific peoples and those with low income during the period of public health measures,^10,11 affecting timely diagnosis and treatment of Strep A infection for Māori and Pacific peoples.

Delayed or missed diagnoses may result in future recurrences being reported as “initial” ARF hospitalisations at an older age and, therefore, missing the critical opportunity to prevent ARF recurrence and cardiac complications. Notification data from Counties Manukau district in 2023–2024 are consistent with this hypothesis, demonstrating an increase in the proportion of persons hospitalised with ARF aged >20 years compared with 2020–2021 (personal communication in: email from the National Public Health Service, 16 April 2024).

The risk of ARF hospitalisation remained lower in 2022 despite the staged opening of borders to international travel and the lifting of other public health measures. Bennett et al. propose that Strep A transmission was suppressed until early 2023;¹⁶ however, there is also evidence of ongoing barriers to healthcare access, and diagnosis and treatment of Strep A and ARF through 2022. Reduced hospitalisation rates were seen for both emergency department presentations and acute hospital admissions (including non-transmissible conditions).²⁷ These did not return to pre-COVID levels following the transition to the staged mitigation strategy from early 2022, with the effect particularly marked for Pacific peoples. Ongoing ethnic inequities suggest that reduced Strep A transmission and healthcare access barriers, which extended beyond the time period of most stringent public health measures, continued to be differentially distributed in 2022.

The rebound in Strep A transmission and ARF cases reported in 2023–2025,^1,16 while outside of our study period, provides further context for interpretation. ARF hospitalisation rates and community-detected Strep A skin and throat infections have returned to the levels seen prior to the COVID-19 pandemic.¹⁶ Māori and Pacific peoples continue to experience high rates of ARF—a stark reminder that colonial policies, practices and institutions persist, resulting in the reproduction of unfair and avoidable ethnic inequities in Aotearoa.

Our study provides important insights into the complex interplay between determinants of ARF and inequities. Of note, our approach provides valuable methodological insights, contextualising analyses of COVID-19 public health measures within a broader landscape of large and unacceptable ARF inequities for Māori and Pacific peoples and producing a model to examine rates for specific time periods. Our results for the broader time period (2000–2022) are consistent with those described previously,⁴ noting the extended time period of this study. Study findings are limited by the data available and sources used. The decision to treat 2021 and 2022 denominators as stable assumes minimal migration and balanced birth and death rates. Birth and death data indicate a small natural increase that is not expected to significantly impact findings.²⁸ We note the exclusion of diagnoses outside of a hospital setting as a minor limitation given primary care diagnoses are unusual. Missed diagnoses are a recognised but unavoidable limitation of studies utilising secondary data.

Ethnicity data and coding of hospitalisation data have well-recognised quality limitations. Ethnicity data are generally complete and timely, but continue to undercount Māori.²⁹ Diagnostic criteria have changed over time and ARF coding misclassification has been identified. While there are limitations to using hospitalisation data, previous audits indicate it is more sensitive for ARF than for notification data.³⁰ These remain critical areas for improvement to more accurately understand trends and inequities in ARF hospitalisation.

Conclusions

Interpreting trends in ARF hospitalisation requires understanding of both the distribution and determinants of ethnic health inequities, which are shaped by racism and coloniality. The dramatic and sustained reduction in ARF incidence for New Zealand Europeans over a period of decades provides clear evidence that ethnic health inequities are not inevitable or fixed in Aotearoa. Temporary reductions in ARF hospitalisations during the period of COVID-19 public health measures highlight the need to address structural inequities that shape the differential distribution of ARF risk and protective factors by ethnicity, including barriers to healthcare access that prevent timely diagnosis and treatment. In Aotearoa, sustained cross-sector primordial prevention strategies to address determinants of Strep A infection and ARF are needed to eliminate health inequities experienced by Māori and Pacific peoples, recognising the need to transform institutions and deliver quality Māori- and Pacific-led health and prevention services.

View Appendix.