Research Questions and Testable Predictions . | Results . | Conclusions . |
---|---|---|
City-Level Organization and Action | ||
A. Were levels of residential segregation, residential density, and illiteracy associated with non-White/White mortality ratios? | Cities with high prepandemic infectious disease mortality are also hit hard by the pandemic, but city demographics are unlikely to explain the specific mortality signature of the 1918 pandemic. | |
A1. Cities with high prepandemic mortality are hit harder by the 1918 pandemic A2. Illiteracy, density, and segregation associated with mortality disparities in 1918 | A1. Yes A2. No evidence in support | |
B. Did nonpharmaceutical interventions (NPIs) offer protections and disproportionately benefit non-Whites? | NPIs help to explain cumulative total mortality but are unlikely to explain reduced racial disparities. | |
B1. NPI implementation associated with race-independent (total) mortality | B1. Yes | |
B2. NPI implementation associated with race-specific mortalities and mortality disparities | B2. No evidence in support | |
Social Immunology | ||
C. Did racial differences in herald wave exposure generate differences in partial immunity, disproportionately protecting non-White communities? | Partial immunity is unlikely to explain reduced fall disparities, though herald wave exposure may have mattered for nonimmunological reasons (e.g., behavioral changes). (Note: Low statistical power may reduce ability to detect associations; calculations are based on potentially noisy/biased proxy measures.) | |
C1. Non-Whites had greater spring wave exposure | C1. Yes | |
C2. Greater spring wave exposure associated with lower fall wave mortality | C2. Mixed evidence (inconsistent) | |
D. Did racial differences in early childhood flu exposure to the 1890 influenza virus produce greater influenza mortality among Whites compared with non-Whites in 1918? | We cannot rule out 1890 exposure as a major driver of high mortality among White young adults, and thus reduced disparities, in 1918. If 1890 exposure is to fully account for such high mortality, historical flu exposure and immunological imprinting parameters need to line up in relatively narrow ways, albeit ones consistent with some historical influenza literature. (Note: We are unable to directly measure imprinting.) | |
D1. Reduced 1918 disparities are driven by cohorts who would have had 1890 exposure | D1. Yes | |
D2. Proportion of city residents with urban origins is greater for White than for non-White populations | D.2. Yes | |
D3. Proportions of city residents with urban origins positively associated with mortality | D3. Yes (though only suggestive) | |
D4. Aggregate mortality in the 20–29 and 30–39 age bands is consistent with “reasonable” mortality among the hypothetically imprinted | D4. Yes (but only if imprinting is close to ubiquitous in some cohorts) | |
D5. Estimated imprinting rates are consistent with “reasonable” historical influenza attack rates | D5. Yes (but the parameter space is highly constrained for the 20–29 age-group) |
Research Questions and Testable Predictions . | Results . | Conclusions . |
---|---|---|
City-Level Organization and Action | ||
A. Were levels of residential segregation, residential density, and illiteracy associated with non-White/White mortality ratios? | Cities with high prepandemic infectious disease mortality are also hit hard by the pandemic, but city demographics are unlikely to explain the specific mortality signature of the 1918 pandemic. | |
A1. Cities with high prepandemic mortality are hit harder by the 1918 pandemic A2. Illiteracy, density, and segregation associated with mortality disparities in 1918 | A1. Yes A2. No evidence in support | |
B. Did nonpharmaceutical interventions (NPIs) offer protections and disproportionately benefit non-Whites? | NPIs help to explain cumulative total mortality but are unlikely to explain reduced racial disparities. | |
B1. NPI implementation associated with race-independent (total) mortality | B1. Yes | |
B2. NPI implementation associated with race-specific mortalities and mortality disparities | B2. No evidence in support | |
Social Immunology | ||
C. Did racial differences in herald wave exposure generate differences in partial immunity, disproportionately protecting non-White communities? | Partial immunity is unlikely to explain reduced fall disparities, though herald wave exposure may have mattered for nonimmunological reasons (e.g., behavioral changes). (Note: Low statistical power may reduce ability to detect associations; calculations are based on potentially noisy/biased proxy measures.) | |
C1. Non-Whites had greater spring wave exposure | C1. Yes | |
C2. Greater spring wave exposure associated with lower fall wave mortality | C2. Mixed evidence (inconsistent) | |
D. Did racial differences in early childhood flu exposure to the 1890 influenza virus produce greater influenza mortality among Whites compared with non-Whites in 1918? | We cannot rule out 1890 exposure as a major driver of high mortality among White young adults, and thus reduced disparities, in 1918. If 1890 exposure is to fully account for such high mortality, historical flu exposure and immunological imprinting parameters need to line up in relatively narrow ways, albeit ones consistent with some historical influenza literature. (Note: We are unable to directly measure imprinting.) | |
D1. Reduced 1918 disparities are driven by cohorts who would have had 1890 exposure | D1. Yes | |
D2. Proportion of city residents with urban origins is greater for White than for non-White populations | D.2. Yes | |
D3. Proportions of city residents with urban origins positively associated with mortality | D3. Yes (though only suggestive) | |
D4. Aggregate mortality in the 20–29 and 30–39 age bands is consistent with “reasonable” mortality among the hypothetically imprinted | D4. Yes (but only if imprinting is close to ubiquitous in some cohorts) | |
D5. Estimated imprinting rates are consistent with “reasonable” historical influenza attack rates | D5. Yes (but the parameter space is highly constrained for the 20–29 age-group) |