Cause-of-Death Disparities in the African Diaspora: Exploring Differences Among Shared-Heritage Populations

Abstract

Objectives. We investigated changes in life expectancy (LE) and cause-specific mortality over time, directly comparing African-descent populations in the United States and the Caribbean.

Methods. We compared LE at birth and cause-specific mortality in 6 disease groups between Caribbean countries with a majority (> 90%) African-descent population and US African Americans.

Results. The LE improvement among African Americans exceeded that of Afro-Caribbeans so that the LE gap, which favored the Caribbean population by 1.5 years in 1990, had been reversed by 2009. This relative improvement among African Americans was mainly the result of the improving mortality experience of African American men. Between 2000 and 2009, Caribbean mortality rates in 5 of the 6 disease groups increased relative to those of African Americans. By 2009, mortality from cerebrovascular diseases, cancers, and diabetes was higher in Afro-Caribbeans relative to African Americans, with a diabetes mortality rate twice that of African Americans and 4 times that of White Americans.

Conclusions. The Caribbean community made important mortality reductions between 2000 and 2009, but this progress fell short of African American health improvements in the same period, especially among men.

In the United States, African American health is regularly reported to be worse than that of White Americans.^1,2 These regular observations from health surveillance, death registration, and other epidemiological studies have motivated a major research effort to understand the reasons for the observed differences. Social determinants have been identified as strong factors in the health differential,^3–6 and this collective evidence has led to US government initiatives aiming to eliminate socially determined and unwarranted health disparities: Healthy People 2010 and Healthy People 2020.^7,8

The Caribbean has a significant African-descent population that, because of its shared heritage, is considered to have ethnic and genetic similarities to African Americans. Studies have leveraged this relationship, with Caribbean-descent study participants used as a comparison group to African Americans in studies investigating, for example, depressive disorders, cancer, and cardiovascular diseases.^9–11 Other international research efforts have examined the health of geographically dispersed African-descent populations, investigating obesity, hypertension, vascular disease, and cancer, among other conditions.^12–16

US-funded Caribbean-based studies have subsequently influenced US health and research guidelines, notably for glaucoma, and continue to contribute to US strategic health planning.^17,18 To date, these US–Caribbean comparative efforts have not been guided by a regional public health strategy. Moreover, comparisons of African-descent populations have mostly been unilateral, with Caribbean public health not yet looking for evidence-based synergies with their African American neighbors.

The National Institute on Minority Health and Health Disparities has funded a 5-year grant (U24MD006959) to explore and compare for the first time health disparities among African-descent populations in the Caribbean and the United States. This project, a capacity-building collaboration between the Sullivan Alliance¹⁹ and the University of the West Indies,²⁰ has created a new entity known as the USA–Caribbean Alliance for Health Disparities Research and is using a wide range of published work, Caribbean and US health databases, and open-access data to build an evidence-based picture of Caribbean health and health disparities, comparing these disparities among Caribbean people with relevant populations in the United States. This article represents early output from this research program and is 1 in a series of analyses reporting disparities in a range of health domains. We anticipate that with the creation of a comprehensive disparities situation analysis for the region, priorities for public health both regionally and on a country level can then be based on contextually relevant evidence. In this article, we focus on changes in life expectancy (LE) and cause-specific mortality over time, directly comparing African-descent populations in the United States and the Caribbean.

METHODS

We report LE and cause-specific mortality. For the Caribbean, we extracted country-level population totals and LE at birth from the United Nations (UN) World Population Prospects, 2012 Revision,²¹ and we extracted the number of age-specific deaths by cause from the World Health Organization Mortality Database.²² This database is a compilation of medically certified mortality data by age, sex, and cause of death, classified according to the International Classification of Diseases (ICD). Deaths are reported annually by member states from civil registration systems, and postsubmission data cleaning by World Health Organization analysts verifies ICD codes but does not adjust for undercoverage. For the United States, data were required by self-reported race. We extracted the number of age-, sex-, and race-specific deaths by cause from the Wide-ranging Online Data for Epidemiological Research database.²³ We obtained LE at birth by race for the United States from the National Center for Health Statistics 2012 health report.²⁴

Defining African-Descent Populations

We focused on the African-descent population in the Caribbean and in the United States. What countries are considered to constitute the Caribbean depends on historical and sociopolitical factors as well as geography,²⁵ and no universally recognized definition exists for which countries should be included. We used the UN definition, which identifies 30 island territories in the Caribbean basin (for details, see the supplementary file available with the online version of this article at http://www.ajph.org). We added 4 territories to the UN definition because of their historical and sociopolitical links to the Caribbean: Belize in Central America and the Guianas in northeastern South America, consisting of Guyana, Suriname, and French Guiana.

Of the 30 island territories, 9 with populations below 90 000 do not have published UN demographic estimates and are not considered in this article. Mortality data stratified by race are not publicly available for most Caribbean territories. Therefore, to approximate African-descent populations in the Caribbean, we selected territories reporting a population with a high percentage of self-reported African descent (arbitrarily set as ≥ 90%) for inclusion. From published census reports, we determined the percentage of the population from each country self-reporting their race as Black or mixed Black. From the 21 territories with populations higher than 90 000, we determined 8 to be of primarily African descent: Antigua and Barbuda, Bahamas, Barbados, Grenada, Jamaica, Haiti, St. Lucia, and St. Vincent and the Grenadines. Of these, 2 (Jamaica and Haiti) are considered to have poor vital registration systems (Haiti especially so) and have submitted little mortality data to the World Health Organization Mortality Database. Two further territories are likely to have high African-descent populations (Guadeloupe and Martinique) but are prevented by law from collecting race information at each census round. Therefore, we selected 6 territories in this report as Caribbean indicator countries with high African-descent populations.

We explored the effect of country choice on regional LE in a brief sensitivity analysis. In the United States, mortality and LE data stratified by race, age, and sex are available. For 1979 to 2002, all 50 states and the District of Columbia collected race data on death certificates using 4 single-race categories (American Indian or Alaska Native, Asian or Pacific Islander, Black, and White), allowing only a single race to be reported. Starting in 2003, some states began allowing 1 or more race categories to be reported. To account for this differential race reporting, multiple race responses were imputed (or bridged) to a single race; this imputation process is described in detail elsewhere.²⁶ Although the broad grouping Black includes foreign-born African-descent subgroups, we use this grouping as a proxy for African American. For comparative purposes, we also included the White American population in this article.

Statistical Analysis

For the Caribbean, we calculated average LE as the weighted average of country life expectancy using country population as the weighting factor. We then calculated the life expectancy gap (LEG; the LE difference, an absolute disparity measure) for African-descent Caribbean and for White American populations compared with African Americans, separately for women and men. We report age-standardized mortality rates in the Caribbean for 2000 and 2009 to give context to the observed LE differences. For the chosen time frame, all countries reported cause of death using the ICD 10th Edition (ICD-10).²⁷ Mortality rates were age standardized to the standard US 2000 population²⁸ and are presented per 100 000 population. All-cause mortality and 5 broad disease areas are presented: diseases of the heart (heart disease), malignant neoplasms (cancers), chronic lower respiratory diseases, cerebrovascular disease (strokes), and accidents (unintentional injuries). These 5 disease groups were the 5 leading causes of death in the United States in 2009.²⁹ Because of regionwide concerns about the health burden of diabetes mellitus, we included deaths resulting from diabetes as a 6th cause of death (diabetes was the seventh leading US cause of death in 2009). We calculated simple metrics to highlight different aspects of the comparison between Afro-Caribbean and African American populations. The change and percentage change in mortality rate between 2000 and 2009 described the absolute performance of each population group, and the mortality rate ratio between Afro-Caribbeans and each US population described the relative performance of the Caribbean compared with the United States.

Among vital registration data, 2 unknown but estimable features of quality are missing deaths (incompleteness) and deaths with a miscoded cause (misclassification). In a brief sensitivity analysis, we estimated the extent of these data-quality issues among the Caribbean indicator countries. To estimate incompleteness, we drew on UN death estimates, which are derived from a variety of country-level sources²¹ and are used to generate the UN’s LE profiles. The difference between UN-estimated deaths and deaths from national statistics provided an estimate of vital registration undercount for each country, which we distributed pro rata across the defined death categories. To estimate misclassification, we applied deaths assigned to ill-defined causes (ICD-10 codes R00–R99) proportionately by age and sex to other causes, apart from injuries. Last, we also redistributed deaths with unknown age or sex pro rata to other causes. We performed all analyses using Stata, version 13 (StataCorp LP, College Station, TX).

RESULTS

The LE profile and LEG between 1990 and 2009 for 3 populations (Afro-Caribbeans, African Americans, and White Americans) are presented in Figure 1 separately for women and men. Between 1990 and 2009, LE increased for all 3 populations. LE increased from 70.6 to 74.1 years among the 6 indicator countries of the Caribbean, a rise of 3.5 years (from 73.2 to 76.7 among women, a rise of 3.5 years, and from 67.9 to 71.5 among men, a rise of 3.6 years). LE increased from 69.1 to 74.7 years among African Americans, a rise of 5.6 years (from 73.6 to 77.7 among women, a rise of 4.1 years, and from 64.5 to 71.4 among men, a rise of 6.9 years). Equivalent LE figures for White Americans were from 79.4 to 81.2 years among women (a rise of 1.8 years), from 72.7 to 76.4 years among men (a rise of 3.7 years), and from 76.1 to 78.8 years among women and men combined (a rise of 2.7 years).

FIGURE 1—

Life expectancy (LE) among Afro-Caribbean and US populations showing (a) LE at birth for women, (b) LE at birth for men, and (c) the LE gap relative to African Americans: 1990–2009.

Note. The 6 Caribbean indicator countries were Antigua, Bahamas, Barbados, Grenada, St. Lucia, and St. Vincent and the Grenadines.

The LE improvement among African Americans was more pronounced than that among Afro-Caribbeans, so that the average LEG, which was 1.5 years in favor of the Caribbean population in 1990, had been reversed by 2009, with African Americans on average having an extra 0.6 years of expected life. This relative improvement among African Americans was primarily due to the improving mortality experience of African American men, who reduced a 3.4-year LE deficit in 1990 to a 0.1-year deficit in 2009. Over the same 20-year period, the US LEG (White American LE minus African American LE) had narrowed from 7.0 to 4.1 years (from 5.8 to 3.5 years among women and from 8.2 to 5.0 years among men).

Age-Standardized Mortality Rates (2000–2009)

Directly standardized mortality rates and percentage change in these mortality rates for the 3 populations in 2000 and 2009 are summarized in Table 1, with mortality rate ratios and 95% confidence intervals for the Caribbean compared with African Americans and White Americans presented in Figure 2.

TABLE 1—

Life Expectancy, Age-Adjusted Mortality Rates (per 100 000 population), and Percentage Change in Mortality Rates Among 6 Caribbean Territories, African Americans, and White Americans: 2000 and 2009

	Afro-Caribbean			Afro-Caribbean (Quality Adjusted)			African American			White American
Cause of Death (ICD–10 Code)	MR 2000	MR 2009	Δ%	MR 2000	MR 2009	Δ%	MR 2000	MR 2009	Δ%	MR 2000	MR 2009	Δ%
Women
Life expectancy at birth	74.6	76.7	2.8	—	—	—	75.1	78.0	3.9	79.9	81.3	1.8
All-cause mortality	859.4	814.0	−5.3	1000.2	887.6	−11.3	927.6	763.3	−17.7	715.3	631.3	−11.7
Heart disease (I00–I09, I11, I13, I20–I51)	185.4	163.7	−11.7	220.1	183.3	−16.7	277.6	191.0	−31.2	205.6	143.4	−30.3
Malignant neoplasms (C00–C97)	143.6	141.2	−1.7	171.1	159.0	−7.1	193.8	167.0	−13.8	166.9	147.9	−11.4
Respiratory diseases (J40–J47)	6.1	5.0	−18.0	7.6	5.7	−25.0	22.7	22.3	−1.8	39.5	41.2	4.3
Cerebrovascular diseases (I60–I69)	85.5	72.1	−15.7	104.7	80.9	−22.7	76.2	50.0	−34.4	57.2	37.6	−34.3
Accidents (V01–X59, Y85–Y86)	11.9	15.8	32.8	14.0	17.9	27.9	21.3	19.4	−8.9	22.2	26.2	18.0
Diabetes mellitus (E10–E14)	102.2	84.8	−17.0	124.1	95.1	−23.4	49.4	35.8	−27.5	20.4	15.9	−22.1
Men
Life expectancy at birth	69.6	71.5	2.7	—	—	—	68.2	71.8	5.3	74.7	76.5	2.4
All-cause mortality	1233.0	1171.1	−5.0	1437.6	1277.6	−11.1	1403.5	1123.1	−20.0	1029.4	880.5	−14.5
Heart disease (I00–I09, I11, I13, I20–I51)	251.3	231.5	−7.9	299.1	259.5	−13.2	392.5	289.0	−26.4	316.7	226.6	−28.4
Malignant neoplasms (C00–C97)	227.0	245.1	8.0	270.5	275.5	1.8	340.3	266.7	−21.6	243.9	209.2	−14.2
Respiratory diseases (J40–J47)	26.7	20.9	−21.7	33.1	23.6	−28.7	47.5	40.6	−14.5	57.2	51.4	−10.1
Cerebrovascular diseases (I60–I69)	91.6	98.0	7.0	112.5	110.2	−2.0	89.7	58.8	−34.4	59.9	38.0	−36.6
Accidents (V01–X59, Y85–Y86)	50.9	44.5	−12.6	61.1	49.7	−18.7	58.2	45.2	−22.3	49.2	53.7	9.1
Diabetes mellitus (E10–E14)	83.8	81.7	−2.5	100.9	91.8	−9.0	48.7	43.2	−11.3	26.0	23.3	−10.4

Note. ICD-10 = International Classification of Diseases, 10th Revision²³; MR = mortality rate.

FIGURE 2—

Age-adjusted mortality rate ratios among 6 Caribbean indicator countries for (a) Afro-Caribbean women vs African American women, (b) Afro-Caribbean women vs White American women, (c) Afro-Caribbean men vs African American men, and (d) Afro-Caribbean men vs White American men: 2000 and 2009.

Note. The 6 Caribbean indicator countries were Antigua, Bahamas, Barbados, Grenada, St. Lucia, and St. Vincent and the Grenadines.

All-cause mortality.

Compared with African Americans, the all-cause mortality rate was lower among Afro-Caribbeans in 2000 and higher in 2009. Between 2000 and 2009, all-cause mortality had improved for all 3 populations, with larger improvements among African Americans (a 17.7% drop among women and a 20.0% drop among men) and White Americans (drops of 11.7% and 14.5% among women and men, respectively) than among Afro-Caribbeans (drops of 5.3% and 5.0% among women and men, respectively). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 0.93 to 1.07 among women and from 0.88 to 1.04 among men.

Heart disease.

Between 2000 and 2009, mortality resulting from heart disease had improved for all 3 populations, with larger improvements among African Americans (a 31.2% drop among women, a 26.4% drop among men) and White Americans (drops of 30.3% and 28.4% among women and men, respectively) than among Afro-Caribbeans (drops of 11.7% and 7.9% among women and men, respectively). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 0.67 to 0.86 among women and from 0.64 to 0.80 among men, and the rate for Afro-Caribbeans compared with White Americans from 0.90 to 1.14 among women and from 0.79 to 1.02 among men. So although Afro-Caribbeans reported the lowest heart disease mortality rate in 2000, this was no longer the case in 2009.

Malignant neoplasms.

Between 2000 and 2009, mortality resulting from malignant neoplasms improved in the United States and increased marginally in the Caribbean. Mortality improvements were greatest among African Americans (a 13.8% drop among women, a 21.6% drop among men), followed by White Americans (a 11.4% drop among women, a 14.2% drop among men). There was a modest mortality drop among Afro-Caribbean women (a 1.7% drop) and a mortality increase among Afro-Caribbean men (an 8.0% increase). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 0.74 to 0.84 among women and from 0.67 to 0.92 among men. The rate ratio for Afro-Caribbeans compared with White Americans rose from 0.86 to 0.95 among women and from 0.93 to 1.17 among men. Relative to US populations, Afro-Caribbean cancer mortality rates increased, approached parity with African Americans, and, among men, surpassed death rates among White Americans.

Chronic lower respiratory diseases.

Between 2000 and 2009, mortality resulting from chronic lower respiratory disease improved for African-descent populations, with larger improvements among Afro-Caribbeans (an 18.0% drop among women, a 21.7% drop among men) compared with African Americans (an 1.8% drop among women, a 14.5% drop among men). Among White Americans, the rate increased among women (a 4.3% rise) and decreased among men (a 10.1% drop). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans dropped from 0.27 to 0.22 among women and from 0.56 to 0.51 among men, and compared with White Americans, from 0.15 to 0.12 among women and from 0.47 to 0.41 among men. Relative to US populations, Afro-Caribbeans had markedly lower mortality from chronic lower respiratory diseases.

Cerebrovascular disease.

Between 2000 and 2009, mortality resulting from cerebrovascular disease improved in the United States and for Caribbean women but worsened for Caribbean men. Improvements were seen among African Americans (a 34.4% drop among women and men) and White Americans (a 34.3% drop among women, a 36.6% drop among men) compared with Afro-Caribbeans (a 15.7% drop among women, a 7.0% increase among men). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 1.12 to 1.44 among women and from 1.02 to 1.67 among men. Compared with White Americans, the ratio for Afro-Caribbeans rose from 1.49 to 1.92 among women and from 1.53 to 2.58 among men. Afro-Caribbeans reported the highest cerebrovascular disease mortality rate in 2000, and this mortality excess had increased by 2009.

Accidents (unintentional injury).

Between 2000 and 2009, mortality resulting from unintentional injuries improved among African Americans (an 8.9% drop among women, a 22.3% drop among men) and Afro-Caribbean men (a 12.6% drop), and worsened among Afro-Caribbean women (a 32.8% increase) and White Americans (an 18.0% rise among women, a 9.1% rise among men). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 0.56 to 0.81 among women and from 0.88 to 0.96 among men. Compared with White Americans, the ratio for Afro-Caribbeans rose from 0.54 to 0.60 among women and fell from 1.04 to 0.83 among men. Compared to 2000, accidental deaths in 2009 among African Americans were uniformly lower; among White Americans, they were uniformly higher; and the picture for Afro-Caribbeans was mixed.

Diabetes mellitus.

Between 2000 and 2009, mortality resulting from diabetes improved for all 3 populations, with larger improvements among African Americans (a 27.5% drop among women, an 11.3% drop among men) and White Americans (a 22.1% drop among women, a 10.4% drop among men) compared with Afro-Caribbeans (a 17.0% drop among women, a 2.5% drop among men). Over the same period, the mortality rate ratio for Afro-Caribbeans compared with African Americans rose from 2.07 to 2.37 among women and from 1.72 to 1.89 among men. The ratio for Afro-Caribbeans compared with White Americans rose from 5.01 to 5.33 among women and from 3.22 to 3.50 among men. So by 2009, reported diabetes mortality rates for Afro-Caribbean women and men combined was more than twice that of African Americans and more than 4 times that of White Americans.

Sensitivity Analysis

Changing Caribbean-region membership.

Which countries are included in the definition of African-descent Caribbean inevitably affects regional LE and mortality estimates. The effect of different country choices in 2000 and 2009 is presented in Table 2 for 5 alternative country groupings. In 2009, regional Caribbean LE at birth ranged from 64.7 to 76.7 years, depending on grouping choice. Haiti’s LE at birth of 57.5 years in 2000, which rose to 61.4 years in 2009, was the lowest in the Caribbean and markedly reduced the regional LE in country groupings 4 and 5. Conversely, Martinique and Guadeloupe had the highest regional LEs during much of the analysis period. Between 2000 and 2009, the LE for both islands combined increased from 77.6 to 80.1 (a rise of 2.5 years; from 81.0 to 83.4 among women, from 74.1 to 76.6 among men), was 1.3 years higher than that for White Americans in 2009 (2.2 years higher among women, 0.2 years higher among men), and raised the regional LE in country grouping 3.

TABLE 2—

Sensitivity Analysis Exploring the Effect on Life Expectancy at Birth of Changing Caribbean Region Country Membership

Country Grouping (No. of Countries)	Average LE	LE Gap (vs Group 1)	LE Gap (vs African American)	LE Gap (vs White American)
Group 1 (6)a
2000	72.1	—	0.32	−5.18
2009	74.1	—	−0.56	−4.66
Group 2 (8)b
2000	70.2	−1.41	−1.08	−6.58
2009	72.7	−1.47	−2.03	−6.13
Group 3 (10)c
2000	74.6	2.42	2.75	−2.75
2009	76.7	2.51	1.95	−2.15
Group 4 (8)d
2000	61.4	−10.68	−10.36	−15.86
2009	64.7	−9.43	−9.99	−14.09
Group 5 (21)e
2000	69.9	−2.19	−1.87	−7.37
2009	72.1	−2.09	−2.64	−6.74

Note. LE = life expectancy at birth; USVI = US Virgin Islands; WHO = World Health Organization.

^aGroup 1 includes countries with census-confirmed African-descent population greater than 90% and with mortality data available from the WHO mortality database: Antigua, Bahamas, Barbados, Grenada, St. Lucia, and St. Vincent and the Grenadines.

^bGroup 2 includes group 1 countries + 2 countries with census-confirmed African population between 70% and 90%. The 2 additional countries are the Dominican Republic and USVI.

^cGroup 3 includes group 1 countries + 4 countries with unknown percentage of African-descent population. The 4 countries are Aruba, Curaçao, Guadeloupe, and Martinique.

^dGroup 4 includes group 1 countries + 2 countries with census-confirmed African population greater than 90% but with no mortality data submitted to the WHO mortality database. The 2 additional countries are Jamaica and Haiti.

^eGroup 5 includes all 21 Caribbean countries with United Nations life expectancy data. Countries are Antigua and Barbuda, Aruba, Bahamas, Barbados, Belize, Cuba, Curaçao, Dominican Republic, French Guiana, Grenada, Guadeloupe, Guyana, Haiti, Jamaica, Martinique, Puerto Rico, St. Lucia, St. Vincent and the Grenadines, Suriname, Trinidad and Tobago, and the USVI.

Data quality.

Three indices of data quality are presented in the online supplemental file for the 6 Caribbean indicator countries. The estimated mortality undercount in 2000 ranged from 4.5% to 42.0%, for a regional average undercount of 16.1%; in 2009, it ranged from 2.4% to 12.1%, for a regional undercount of 8.8%. Regional cause-of-death misclassification was 2.8% in 2000 and 3.4% in 2009, and age or sex information was missing for 0.3% of deaths in 2000 and 0.4% of deaths in 2009. After adjusting for these quality estimates, all-cause mortality among Caribbean women in 2000 increased from 859.4 to 1000.2 per 100 000 population; in 2009, it increased from 814.0 to 887.6 per 100 000 population. Equivalent increases for men were from 1233.0 to 1437.6 per 100 000 population in 2000 and from 1171.1 to 1277.6 per 100 000 population in 2009. Full details of age- and quality-adjusted mortality rates are presented in Table 1 and in the online supplemental file.

DISCUSSION

Mortality in the Caribbean has been comprehensively reported by the Pan-American Health Organization and others for many years. Published rates have highlighted broadly decreasing mortality across the Caribbean, painting a positive picture of improving national health profiles. Using the many readily available data summaries,³⁰ comparisons are possible within the Caribbean and highlight important disparities between Caribbean nations. Such data also offer external comparisons for Caribbean data using Latin American subregions.³¹ However, context is important for external comparisons, and North and South America’s majority European-descent populations share few cultural similarities to the African-descent population of the Caribbean. Because of their shared heritage, Caribbean people are considered to have ethnic and genetic similarities to African Americans. Studies have leveraged this relationship, and US–Caribbean comparisons exist for a growing number of health domains,^9–16 although most of these publications have not stated the description of health disparities as an explicit goal. This article begins the process of systematically describing US–Caribbean health disparities by focusing on mortality among key disease groups. Much work remains to investigate other health outcomes and explore within-country health disparities. We anticipate that with the creation of a comprehensive disparities situation analysis for the region, priorities for public health both regionally and on a country level can then be based on contextually relevant evidence.

We have described a broadly positive picture of improving LE in African-descent populations in the Caribbean and the United States. The improvement among African Americans exceeded that among Afro-Caribbeans, so that the LEG, which was 1.5 years in favor of the Caribbean population in 1990, had been reversed by 2009, with African Americans having an additional 0.6 years of expected life. This relative improvement among African Americans was primarily due to the improving mortality experience of African American men, who reduced a 3.4-year LE deficit in 1990 to a 0.1-year deficit in 2009.

Between 2000 and 2009, mortality rates in 5 of the 6 disease groups in the Caribbean increased relative to those for African Americans; only relative rates for respiratory diseases decreased. A similar pattern of relative increase was seen in women and in men, irrespective of whether the Caribbean data were adjusted for death undercount and cause-of-death misclassification. Using age-adjusted mortality rates, by 2009 rates of diabetes and cerebrovascular disease in the Caribbean were, respectively, roughly 2 times (women, 2.4 times; men, 1.9 times) and 1.5 times (women, 1.4 times; men, 1.6 times) those of African Americans, and data-quality adjustment increased these rates further (for diabetes, women, 2.7 times and men, 2.1 times; for cerebrovascular disease, women, 1.6 times, and men, 1.9 times). The health system response to these chronic noncommunicable disease disparities should be a Caribbean imperative. Noncommunicable disease surveillance is key, and the Barbados government has developed a model for disease monitoring in small island states. Recognizing that the chronic disease burden is not driven by a single disease, the Barbados surveillance system monitors multiple diseases concurrently with centralized staffing and data management functions that offer important economies of scale. Currently, Barbados monitors the incidence of stroke, acute myocardial infarction, and cancers.³² The challenge of managing and preventing noncommunicable diseases then lies with primary health care. Equipping staff with the appropriate competencies to take on the noncommunicable disease challenge is vital to reducing these observed disparities.³³

A definition of the Caribbean region is not absolute, with alternative country inclusions regularly being based on language, geography, history and culture, and organizational affiliation. We began with a widely used organizational grouping (from the UN), and our algorithm for defining an African-descent country grouping was then a pragmatic attempt in the face of unavailable data. Our brief sensitivity analysis suggested that many other country groupings lead to lower estimates of regional LE, and so our primary country grouping (6 countries) offers a somewhat conservative picture of Caribbean–US mortality differences. We obtained higher Caribbean LE estimates only when including islands with extraregional affiliations (e.g., Guadeloupe, Martinique), and it could be useful to understand the reasons for the recent mortality rate improvements in these territories.

Technical challenges arise when aggregating mortality data from vital registration systems. Although all countries in the Americas have operational death registration systems, not all have submitted annual data to regional data repositories^22,34; the 10-year submission rate among Caribbean countries was 70%. Moreover, data submitted by most countries will suffer from some level of incompleteness and cause-of-death misclassification. Variation in these imperfections between countries and across time creates reporting biases that complicate data interpretation. In the face of data-quality challenges, analysts must make pragmatic attempts to understand these data imperfections, reporting results that highlight the possible extent of such limitations.

In this article, simple—and possibly simplistic—quality adjustments have been applied that suggest higher underreporting and misclassification in the Caribbean in 2000 compared with 2009. Adjusted figures worsen the absolute Caribbean mortality experience in each year, suggesting a greater Caribbean mortality improvement between 2000 and 2009, and we have presented this second possible version of the true Caribbean mortality experience throughout the results as quality-adjusted mortality rates. A better understanding of health metric data quality in the Caribbean should be a key analytical priority. When using vital registration data, it will be important to clearly report the methods used and assumptions made to accommodate data imperfections, and sensible sensitivity analyses will be key to understanding the practical importance of these adjustment methods and assumptions.