Population Estimates of Sickle Cell
Disease in the U.S.
Kathryn L. Hassell, MD
Background: The number of individuals with sickle cell disease (SCD) in the U.S. is unknown.
Determination of burden of disease, healthcare issues, and policies is best served by representative
estimations of the SCD population.
Purpose: To update SCD population estimates by using recent U.S. Census and birth-cohort SCD
prevalence for at-risk populations as available through the centralized reporting of universal newborn
screening for hemoglobinopathies, with an effort to demonstrate the potential effect of early mortality.
Methods: National and state SCD populations were estimated based on the 2008 U.S. Census, using
total, African-American, and Hispanic birth-cohort disease prevalence derived from the National
Newborn Screening Information System. Estimates were corrected for early mortality for sickle cell
anemia using data from the CDC’s Compressed Mortality Report and published patient-cohort
survival information.
Results: National SCD population estimates ranged from 104,000 to 138,900, based on birth-cohort
disease prevalence, but from 72,000 to 98,000 when corrected for early mortality. Several limitations
were noted in the available data, particularly for SCD mortality in adults.
Conclusions: The number of individuals with SCD in the U.S. may approach 100,000, even when
accounting for the effect of early mortality on estimations. A paucity of high-quality data limits
appropriate estimation. State-to-state variability may preclude application of state-specifıc information to other states or to the nation as a whole. Standardized collection and centralized reporting, a
surveillance system, will be necessary to assess the size and composition of the U.S. SCD population.
(Am J Prev Med 2010;38(4S):S512–S521) © 2010 American Journal of Preventive Medicine
Introduction
T
he number of individuals with sickle cell disease
(SCD) in the U.S. is unknown. Thirty years ago, the
U.S. sickle cell anemia population was estimated to
be 32,000 –50,000, based on reported gene frequencies derived from testing of African-American neonates.1 Subsequent population estimates of over 50,000 – 80,000 for
both SCD and sickle cell anemia (a common form of
SCD) are noted in a variety of publications, usually without a specifıc reference. When a source is documented,
information reported by the Agency for Health Care
Policy and Research (AHCPR),2 the CDC,3 or the National Heart, Lung and Blood Institute4 is most often
cited. Specifıc methods used to obtain these fıgures are
From the Colorado Sickle Cell Treatment and Research Center, University
of Colorado Denver Health Sciences Center, Aurora, Colorado
Address correspondence and reprint requests to: Kathryn L. Hassell,
MD, Colorado Sickle Cell Treatment and Research Center, 13121 East 17th
Avenue, Mail Stop C-222, P.O. Box 6511, Building L28 Room 351, Aurora
CO 80045. E-mail: [email protected].
0749-3797/00/$17.00
doi: 10.1016/j.amepre.2009.12.022
S512 Am J Prev Med 2010;38(4S):S512–S521
not provided but are usually discussed in the context of the
frequency of sickle cell anemia in the U.S. African-American
population as determined by newborn screening data. Although some authors note a potential effect of early mortality,1 specifıc adjustment to population estimates is not apparent. Recent work by Strouse et al.5 corrects California
SCD population estimates for early mortality, utilizing data
from state administrative data sets, but this correction has
not been applied to national estimates.
To date, there are no published national population
estimates that use contemporary birth cohort-disease
prevalence data that clearly delineate specifıc types of
SCD or that incorporate the emergent Hispanic community, a growing segment of the U.S. population that is also
affected. Yet the availability of more-representative population estimates is critical when anticipating public
health and healthcare service needs, conducting health
economics analysis, and developing policy. In the absence of current available information, examples of extrapolation based on old population estimates continue
into 2009,6 which may underestimate the total burden
and need.
© 2010 American Journal of Preventive Medicine • Published by Elsevier Inc.
Hassell / Am J Prev Med 2010;38(4S):S512–S521
S513
The number of individuals with SCD was estimated by multiplying population estimates determined by the 2008 U.S.
Census7 by reported or derived birth cohort-SCD prevalence. For this analysis, the birth cohort-disease prevalence
was assumed to represent the prevalence of SCD in the
broader resident population in which the births occurred.
Several sources were used for estimation of SCD prevalence
based on birth cohort information.
Genetic Services. Hispanic population disease prevalence for
the Eastern states was based primarily on data from New
York and Florida, reported to reflect Caribbean rather
than Mexican background.2
For purposes of the current analysis, AHCPR prevalence
estimates were applied to the non-Hispanic black and nonblack
Hispanic populations of each state as reported by 2008 U.S.
Census data.7 The Eastern Hispanic prevalence estimate reported by AHCPR was applied to states in the Northeast region
as defıned by the U.S. Census (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont, New Jersey,
New York, and Pennsylvania) and Florida. The Western Hispanic estimates were applied to all other states.
Subsequent data voluntarily reported to a national data
set by state newborn screening programs from 1990 to
1999 were analyzed by Therrell and Hannon.8 They reported incidences for sickle cell anemia (Hb SS, 1:3044);
hemoglobin SC disease (Hb SC, 1:7386); and Hb SS⫹SC
(1:2474) in the overall U.S. population. For this analysis,
these prevalence estimates were applied to the total U.S.
population in 2008.
Agency for Health Care Policy and Research
California Newborn Screening Program
This article seeks to update SCD population estimates
by using recent U.S. Census and birth cohort-SCD prevalence for at-risk populations, as made available through
the expansion and voluntary centralized reporting of universal newborn screening for hemoglobinopathies. An
effort to demonstrate the potential effect of early mortality on such estimates is also undertaken.
Methods
Birth Cohort–Based Population Estimate
Figures for the prevalence of SCD in the African-American
and Hispanic populations by AHCPR were published in
1993.2 That report combined all forms of SCD into a single
estimate of prevalence for African-American and Hispanic
populations, distinguishing Hispanic populations from
Eastern and Western states, as noted in Table 1. The
estimates were based on a review of the literature at
the time and available newborn screening data from at least six
states (Wisconsin, Louisiana, Texas, California, Michigan, and
Virginia) as collected by the Council of Regional Networks for
Several publications used data from the California newborn
screening program to estimate the prevalence of SCD. Using
data from 1990 –1996, disease prevalence estimates for three
major forms of SCD—Hb SS, Hb SC, and Hb S␤thalassemia
(HbS␤thal)—for both African-American and Hispanic populations were calculated by Lorey et al.9 and are listed
in Table 1. No distinction was made between Hb
S␤⫹thalassemia (Hb S␤⫹thal) and Hb S␤othalassemia
(HbS␤othal). Updated incidences of specifıc types of SCD
per 100,000 total infants screened in California from 1998
through 2006 were
recently published, wiTable 1. Birth cohort-SCD prevalence estimates
thout designation of
race/ethnicity.10 The auSource of prevalence estimate
thors reported an inciPopulation, type of SCD AHCPR2 1993
California9 1990–1998
NNSIS 2005–2007
dence of 8.5 per 100,000
(1:11,764) for Hb SS
African-American
and 4.4 per 100,000 (1:
All types of SCD
1:346
1:365
22,727) for Hb SC; 1.4
HbSS
1:700
1:601
per 100,000 (1:71,428)
for Hb S␤⫹thalassemia;
HbSC
1:1,297
1:1,127
and 0.8 per 100,000 (1:
HbS␤thalassemia
1:4,056
1:4,198
250,000) for Hb S␤o
Hispanic
thalassemia. For this
analysis, these estimates
All types of SCD
East: 1:1,114
1:16,305
were applied to the 2008
West: 1:31,847
California population
HbSS
1:45,622
1:18,642
and used for compariHbSC
1:364,976
1:57,700
son to fıgures obtained
HbS␤thalassemia
1:729,953
1:175,233
for California from
pooled national data
AHCPR, Agency for Health Care Policy and Research; HbSS, sickle cell anemia; HbSC, hemoglobin SC disease;
(e.g., AHCPR).
NNSIS, National Newborn Screening Information System; SCD, sickle cell disease
April 2010
S514
Hassell / Am J Prev Med 2010;38(4S):S512–S521
The National Newborn Screening Information
System–Based Estimates
The National Newborn Screening Information System
(NNSIS) was established in 2005 to sustain timely and
meaningful centralized national newborn screening data.8
Each state newborn screening program is asked to voluntarily report and distinguish suspected from confırmed cases of
hemoglobinopathies using their own case defınition and
sources of race/ethnicity information; national standardization of submitted data has not been adopted.
State-specifıc information was accessed from this resource for the years 2005–2007, and it included the number
and type of confırmed cases of hemoglobinopathies, including Hb SS, Hb SC, and Hb S␤thal, and race/ethnicity of some
identifıed cases.11 No distinction was made between Hb
S␤⫹thal and Hb S␤othal in the database. Other types of
SCDs were not included in this analysis.
The NNSIS data set was found to be incomplete, as some
states did not report information and others began universal
newborn screening for hemoglobinopathies after 2005.
State-specifıc birth cohort-prevalence estimates were generated only for states that reported data from 2005–2007. Data
from these reporting states were pooled to derive a birth
cohort-prevalence estimate that was applied to states without NNSIS information to permit state-specifıc birth-cohort
estimates.
Information from NNSIS was used to calculate the
birth cohort-SCD prevalence as follows: Total birth cohortprevalence was calculated for SCD (HbSS⫹HbSC⫹HbS␤thal)
for each state by dividing the number of total confırmed
cases into the total birth cohort using combined data from
2005 to 2007. As noted, this calculation was performed only
for states with 3 years of data. Variation in individual state
population demographics precluded pooling data to derive a
representative total birth-cohort prevalence.
Specifıc African-American and Hispanic birth-cohort
prevalence fıgures were calculated in a similar fashion for
Hb SS, Hb SC and Hb S␤thal, and SCD (HbSS⫹HbSC⫹
HbS␤thal), but in this case for the African-American and
non-white Hispanic birth cohorts in each state using combined data from 2005 to 2007. Only states that reported race
and ethnicity information for at least 75% of the identifıed
cases were included in this part of the analysis. This threshold for inclusion in the analysis was used to permit data from
states with only a few cases of SCD, for example, when three
of four cases (75%) were assigned race and ethnicity. Data
from states with suffıcient information were also pooled to
calculate birth-cohort prevalence estimates for AfricanAmerican and Hispanic populations. These fıgures were
then applied to the specifıc at-risk populations in states that
had not reported data for 2005–2007 or that reported race/
ethnicity for ⬍75% of identifıed cases.
It was recognized that this approach potentially resulted
in an underestimate of the birth cohort-disease prevalence,
as up to 25% of a cohort may not have been designated as
African-American or Hispanic as a result of missing data.
Additionally, those assigned to different race categories
(e.g., Caucasian) were not captured. To assess for this effect,
population estimates based on the total birth cohort-disease
prevalence, as described above, were compared to those
based on estimates using African-American and Hispanic
population birth cohorts.
Mortality-Adjusted Population Estimates
Population estimates using birth cohort-disease prevalence
and census data did not consider the effect of early mortality.
There were no available national mortality data or contemporary life expectancy fıgures for individuals with SCD.
Three different sources of information were used.
Cooperative Study of Sickle Cell Disease
Published fıgures for life expectancy (median of 45 years for
Hb SS, 65 years for Hb SC) from the Cooperative Study of
Sickle Cell Disease (CSSCD) were derived from subjects
followed between 1978 and 1988.12 The probability of survival over the next 10 years according to age, beginning at
birth, was also estimated.
Pediatric and Jamaican Cohorts
Three publications reported the survival rates of pediatric
cohorts with Hb SS born after 1982 in Dallas,13 East London,14 and Jamaica.15,16 However, long-term survival estimates and the proportion of individuals alive after age 18
were not addressed. A report from Jamaica documented an
estimated median survival of 53 years for men and 58 years
for women with Hb SS,16 and a recent publication noted that
44% of Jamaican individuals with Hb SS born prior to l943
were still alive.17 Information describing the proportion of
individuals alive at different ages was available from these
publications, estimated in most cases from a specifıc group
of individuals; this information was used as discussed below
to estimate the effect of early mortality on population
estimates.
Compressed Mortality Report from the CDC
The CDC collects national mortality data from death certificates in the CDC Wonder online database and provides
compressed mortality reports.18 For this analysis, the ages at
death for individuals with SCD (indicated on the death
certifıcate as sickle cell anemia with and without crisis, double heterozygote sickling disorders, or other sickle cell disorders) were tabulated and used to extrapolate the relative
proportion of individuals still alive in each 5-year age
bracket. Separate analysis of individuals with sickle cell
anemia and those with other types of sickle cell disorders,
including double heterozygous sickling disorders, was
planned as life expectancies differed by 20 years in the
CSSCD.12 Unfortunately, the small number of deaths
www.ajpm-online.net
Hassell / Am J Prev Med 2010;38(4S):S512–S521
(⬍5%) reported in conditions other than sickle cell anemia
precluded meaningful analysis, so all types of SCD were
grouped together for this analysis. Similarly, there were only
a small number of reported deaths in individuals identifıed
as Hispanic, precluding separate analysis based on race/
ethnicity. Because ⬎95% deaths were associated with sickle
cell anemia, the correction for early mortality was applied
only to the population estimates for sickle cell anemia. The
CSSCD reported that the life expectancy for HbSC (age 65)
was nearly that of the African-American population at the
time (68 years),12 so a strong early mortality effect for HbS
disorders other than sickle cell anemia was not anticipated.
In contrast to the other sources used, these data reflected a
cross-sectional view of age at death, rather than a longitudinal assessment of survival for a given cohort.
Using the 2008 U.S. Census population estimates, as divided into 5-year age brackets, the number of individuals
with sickle cell anemia in each bracket was estimated using
total, African-American, and Hispanic birth cohort-disease
prevalence data. This number was then adjusted for the
proportion of individuals who would be expected to still be
alive in that age bracket as derived from the pediatric and
Jamaican cohorts, the CSCCD, and the CDC Compressed
Mortality Report. This permitted assessment of the impact
of the variability of survival on population estimates. National population fıgures were derived using sickle cell anemia prevalence estimates for the African-American (1:601)
and Hispanic populations (1:18642) and summed for a total
national population estimate of sickle cell anemia. This correction was also applied to individual states for which a
state-specifıc total birth-cohort disease prevalence had been
calculated using NNSIS data. The mortality-adjusted sickle
cell anemia population was added to the birth cohortdisease prevalence– based population estimates for HbSC and
HbS␤thal to obtain total population estimates for each state.
S515
2007 to calculate state-specifıc birth-cohort disease
prevalence estimates for African-American and Hispanic populations, which were pooled to calculate
birth cohort-disease prevalence fıgures listed under
NNSIS in Table 1.
Total birth cohort-SCD prevalence estimation,
without regard to race/ethnicity, was possible from 37
states using data reported to the NNSIS. These data
varied signifıcantly by state; SCD occurred in 1:400 –
600 of all births in some states/areas with large at-risk
populations (e.g., Mississippi, District of Columbia)
but in only 1:20,00 –30,000 of all births in states with
small at-risk populations (e.g., Utah, South Dakota).
Because of this wide variability and incomplete information from 13 states, a national total (as distinct from
African-American or Hispanic) birth cohort-disease
prevalence estimate was not made from the 2005–2007
NNSIS data.
The total U.S. SCD population estimates based on disease prevalence as reported by AHCPR2 and as derived
from the NNSIS are noted in Table 2. In addition, when
the total birth cohort-disease estimates as reported from
national newborn screening data from 1990 to 19988 were
used, the total U.S. sickle cell population was estimated to
be 122,900, of which 99,888 (81%) had HbSS. Of the
104,487 individuals with SCD as estimated using NNSISderived prevalence data from 2005 to 2007, there were
66,070 (63%; 64,131 African-American and 1946 Hispanic) individuals with HbSS.
Individual state SCD population estimates were performed using state-specifıc birth-cohort disease prevalence for total, African-American, and Hispanic birth
cohorts and are provided in Table 3. Analysis using total
birth cohort-prevalence resulted in an average increase of
37% (range⫽25%–70%) in population estimates when
Results
compared to the sum based on African-American and
Hispanic cohorts. For those states without 2005–2007
Table 1 lists the birth-cohort prevalence estimates of
data to calculate the total birth cohort-disease prevalence,
SCD for African-American and Hispanic populations
2
9
the fıgure based on African-American and Hispanic coas reported by AHCPR, as reported for California
horts was carried over to permit national population esand as derived from the NNSIS as described in the
timation. These data are shown in the third and fourth
Methods section. Race and ethnicity data were availcolumns of Table 3. The distribution of SCD types
able from 30 states in the NNSIS for the years 2005–
was fairly consistent
across states, with
Table 2. U.S. SCD population estimates based on total U.S. population (2008 census data)
HbSS representing
60%, HbSC reprePopulation estimate based on
senting 30%, and
SCD population
AHCPR prevalence
Hispanic
NNSIS prevalence
Hispanic
HbS␤thal representAfrican-American
105,261
101,840
ing 10% of the popuHispanic
10,180
2,646
lation. Combining fıgures derived from
Total
115,442
9%
104,487
2.5%
these various sources,
AHCPR, Agency for Health Care Policy and Research; NNSIS, National Newborn Screening Information System;
without correction for
SCD, sickle cell disease
April 2010
Hassell / Am J Prev Med 2010;38(4S):S512–S521
S516
Table 3. U.S. and individual state SCD population estimates from state-specific prevalence, corrected for early
mortality
AHCRP-based
(1993)
NNSIS-based
AAⴙHispanic birth
cohorts
(2005–2007)
NNSIS-based
Corrected for early
total birth
mortality in HbSS
cohort
% Hispanic (2005–2007) Using CDC data Using cohort data
State
Total
All states
121,956
7
110,892
3
138,923
84,743
97,930
3,559
⬍1
3,440
0
3,787
2,500
2,851
86
1
84
3
64
35
45
Alabama
Alaska
Arizona
% Hispanic Total
a
b
635b
838
7
894
16
894
Arkansas
1,305
⬍1
1,267
1
1,931
1,088
1,266
California
7,482
5
5,628
5
5,773
4,240
4,707
Colorado
640
5
436
11
444
311
371
1,375
24
1,350
9
1,858
1,078
1,252
Delaware
530
⬍1
648
3
703
487
561
District of Columbia
932
⬍1
1,665
1
1,944
1,243
1,413
11,632
28
8,556
4
14,434
7,407
8,803
8,427
⬍1
8,165
1
116
2
115
4
46
10
51
22
5,607
1
4,107
3
Connecticut
Florida
Georgia
Hawaii
Idaho
Illinois
b
5,797b
115a
70b
82b
51a
31b
36b
8,165
a
545
5,316
3,243
1
1,636
1
1,636
998
1,162b
Iowa
236
2
234
4
422
215
254
Kansas
506
1
376
5
606
372
417
Kentucky
954
⬍1
799
1
914
660
745
Louisiana
4,081
⬍1
4,388
0
5,470
3,420
3,936
53
26
39
3
125
58
75
Maryland
4,803
⬍1
5,591
0
6,587
4,205
4,860
Massachusetts
1,725
24
1,306
3
3,261
1,598
1,957
Michigan
4,129
⬍1
3,340
1
4,784
2,875
3,322
695
1
445
3
908
456
570
Mississippi
3,160
⬍1
3,279
0
4,347
2,761
3,092
Missouri
1,968
⬍1
2,502
1
2,267
Maine
Minnesota
Montana
20
4
20
9
20
Nebraska
236
2
186
5
200
Nevada
628
3
577
7
707
New Hampshire
New Jersey
New Mexico
New York
North Carolina
b
3,720
1,680
Indiana
a
4,981
1,702
a
1,903
b
12
134
148
466
a
74
38
47
5
47
4,749
24
4,647
5
6,031
a
539
b
29
3,681
b
197
13
229
28
229
140
12,065
19
9,580
2
11,968
7,482
5,776
⬍1
5,603
1
a
5,603
14b
b
3,418
33b
4,256
163b
8,661
3,973b
(continued on next page)
www.ajpm-online.net
Hassell / Am J Prev Med 2010;38(4S):S512–S521
S517
Table 3. (continued)
AHCRP-based
(1993)
State
North Dakota
Total
NNSIS-based
AAⴙHispanic birth
cohorts
(2005–2007)
NNSIS-based
Corrected for early
total birth
mortality in HbSS
cohort
% Hispanic (2005–2007) Using CDC data Using cohort data
% Hispanic Total
20a
12b
14b
20
2
20
4
4,003
⬍1
3,774
1
5,730
3,150
3,725
Oklahoma
846
1
846
2
1,045
672
753
Oregon
232
5
190
18
210
159
180
Pennsylvania
4,329
10
3,668
5
6,058
3,184
3,743
Rhode Island
284
32
185
4
343
150
184
3,692
⬍1
4,309
1
5,176
Ohio
South Carolina
South Dakota
3,694
b
19b
27
2
27
5
Tennessee
3,020
⬍1
2,925
1
2,925a
1,784b
2,077b
Texas
8,644
3
8,389
5
8,958
6,326
7,132
110
9
141
31
118
76
82
Utah
Vermont
27
3,196
a
a
23
31
16
4
16
4,484
⬍1
4,308
1
4,006
726
2
428
10
536
West Virginia
188
⬍1
281
1
281
Wisconsin
996
1
1,086
2
1,851
Virginia
Washington
Wyoming
21
6
22
13
22
a
a
16
b
10
11b
2,558
2,961
321
370
b
171
934
13
b
200b
1,146
16b
a
Numbers over from previous column as state data incomplete for years 2005–2007.
Numbers calculated by using pooled correction estimate rather than state-specific analysis as described in Methods section.
AA, African-American; AHCPR, Agency for Health Care Policy and Research; HbSS, sickle cell anemia; NNSIS, National Newborn Screening
Information System; SCD, sickle cell disease
b
early mortality, the average estimate for number of people
with SCD in the U.S. is 119,100⫾11,915.
Correction of population estimates for early mortality
was performed using data from pediatric cohort and Jamaican studies, the CSCCD, and CDC compressed mortality fıles, as described in the Methods section. The proportion of individuals still alive at age 18 years from
pediatric cohorts with HbSS born after 1982 ranged from
0.85 to 0.99.13–16 The Dallas cohort reported 85% of individuals were still alive as they entered adulthood, which
was comparable with the number of deaths reported in
the pediatric-aged group in 2006 (18%) obtained from the
compressed mortality report. The age at death for individuals with SCD reported to the CDC database is displayed graphically in Figures 1 and 2, and showed a
continual shift over time from 1979 to 2006. In addition
to a shift toward death at older ages, deaths occurring
between ages 1 and 4 years appeared to markedly diminish between 1974 and 2006, which temporally correlated
with the expansion of universal newborn screening for
April 2010
hemoglobinopathies, implementation of penicillin prophylaxis,19 and use of conjugated pneumococcal vaccination. There was less shift noted in the age at death in
adults. The observation of a shift in ages at death, however, precluded the pooling of data over enough years to
obtain a suffıcient number of cases to perform meaningful analysis for specifıc SCD types or for specifıc at-risk
populations.
The mean number of years lived (age at death) was 39
years in 2006 based on data reported to the CDC.18 Of 483
reported deaths in 2006, 9% of them occurred at or before
the age of 20 years, 28% between the ages of 20 and 34
years, 28% between the ages of 35 and 44 years, and 35%
at ages ⬎45 years; the proportion of individuals alive at
age 45 years would be 35%. This was lower than noted by
the CSSCD, in which the proportion of individuals alive
at age 45 years was 50%,12 and in the Jamaican cohort,
where 50% of individuals were still alive at age 55 years.16
Application of correction for early mortality using
CDC compressed mortality data resulted in a 50% de-
Hassell / Am J Prev Med 2010;38(4S):S512–S521
S518
Percentage of deaths
crease in the estimated
30
number of individuals
25
with sickle cell anemia,
with a corresponding de20
crease of 39% in overall
SCD population esti15
mates. The population estimates for the 37 states
10
with an NNSIS-derived
5
state-specifıc total birth
cohort-disease prevalence
0
were corrected and are
<1
1–4 5–9 10–14 15–19 20–24 25–34 35–44 45–54 55–64 65–74 75–84 85+
displayed in Table 3. This
Age group of age at death
correction resulted in a
mean decrease of 39%
1979 (n=301)
1989 (n=389)
1999 (n=503)
2006 (n=483)
(range⫽27%–54%) in statespecifıc SCD population
Figure 1. Age at death for individuals with SCD in years 1979, 1989, 1999, and 2006,
estimates as shown in Tafrom CDC compressed mortality reports
ble 3. The impact of early
SCD, sickle cell disease
mortality was greater for
states with higher perto 88,612. Application of these correction factors to the
centages of individuals with HbSS as compared to HbSC
average population estimate of 119,100, without considand HbS␤thal. Although the distribution of SCD types
eration of state-specifıc data, resulted in corrected national
was not known for the states without NNSIS data, the
SCD population estimates that ranged from 72,700 to
correction (39% reduction in total birth cohort popu84,561. The distribution of the estimated SCD population
lation estimates) was applied to these states to permit an
across the U.S., using the NNSIS-derived data for Africanadjusted national population estimate, which yielded a
American and Hispanic populations, corrected for early
fıgure of 84,674.
mortality in HbSS, is depicted in the map in Figure 3.
Application of the same analysis using the estimates of
Information for the state of California was used to
proportions of individuals alive in each age group from
assess consistency of the methods used. In the state of
the published pediatric and Jamaican cohorts in place of
California, the SCD population was estimated to be
the CDC data resulted in a 33% decrease in the estimated
approximately 7400 using birth cohort-disease prevanumber of individuals
with HbSS and a 29% decrease in the total SCD
population. These data
represented the longest
reported survival for both
children and adults; the
total sickle cell population
estimate was 98,635,
based on the adjusted
state-specifıc population
estimates as provided in
Table 3.
Similar application of
information from the
CSCCD resulted in a 43%
decrease in the sickle cell
anemia population estimates, with a 36% decrease in the total sickle
Figure 2. Age at death for individuals with SCD in 1979 and 2006
SCD, sickle cell disease
cell population estimate
www.ajpm-online.net
Hassell / Am J Prev Med 2010;38(4S):S512–S521
S519
lished estimates have not
made use of the full adoption of universal national
newborn screening, incorporated the Hispanic population, or attempted to assess the impact of early
mortality on national population estimates.
When based solely on
the use of birth cohortdisease prevalence as applied to contemporary
U.S. at-risk and total populations, the number of
individuals with SCD
may number between
≥7000
1000–2499
104,000 and 138,900, with
7,000+
a mean estimate of
5000–6999
500–999
500-999
5,000-6,999
119,100. However, this
2,500-4,999
<500
2500–4999
<500
method overestimates the
number of adults, especially older adults, as a reFigure 3. Estimated number of individuals with SCD, based on state-specific Africansult of the early mortality
American and Hispanic birth-cohort disease prevalence and 2008 U.S. Census population, corrected for early mortality
associated with SCD.
SCD, sickle cell disease
When correction is made
for early mortality associated with sickle cell anelence derived from pooled multistate data, and 5600 –
mia,
population
estimates
are
reduced
to 72,000 –98,000.
6300 when using state-specifıc at-risk and total populaThese SCD population estimates must be viewed with a
tion birth cohort-disease prevalence estimates. When
great deal of caution. There are major limitations to availcorrected for early mortality using data derived from the
able data and methodology used for this report. Birth
CDC compressed mortality fıgures, CSCCD, and pediatric/
cohort-disease prevalence estimates are assumed to repJamaican cohorts, the overall sickle population estimate for
resent the disease prevalence in the larger population, but
California ranged from 4240 to 4707. This estimate is comthis may not reflect immigrant or other populations with
parable to a population estimate of approximately 4300 cal5
a different gene frequency. Birth cohort estimates rely on
culated by Strouse et al. when using state-specifıc mortality
accurate, consistent, and complete reporting of condata.
fırmed cases identifıed by universal newborn screening
using a standardized approach to data collection. The
Conclusion
NNSIS is an excellent national resource but relies on
voluntary reporting; not all states participate, and those
In the 30 years since estimates of 50,000 were fırst cited, it
that do are not required to use consistent case defınitions;
is likely that the U.S. SCD population has increased. This
specifıc testing (e.g., DNA-based techniques); or stanwould be expected given overall growth of at-risk popudardized designation of race/ethnicity. The methods used
lations, including in the African-American and Hispanic
in this report include three major types of sickle cell
populations, and the availability of prophylactic penicildisease (HbSS, HbSC, HbS␤thal—without separating
lin, vaccinations, and disease-remitting therapies, includHbS␤o from S␤⫹thal) but not other, rarer sickle cell
ing hydroxyurea, which appear to affect mortality in
20
disorders. Populations other than those that are Africanadults. An improvement in mortality is also suggested
American and Hispanic are excluded in some analyses, as
by the shift in the age at death of individuals with SCD as
are cases without a designated race/ethnicity. Use of the
reported to the CDC, as noted in Figure 2. However, the
total birth cohort-prevalence estimates for states with
true number of individuals with SCD remains unknown,
NNSIS data suggests that using only cases designated as
and in the absence of a reliable surveillance system, populations will continue to be quantifıed by estimation. PubAfrican-American or Hispanic may result in an underesApril 2010
S520
Hassell / Am J Prev Med 2010;38(4S):S512–S521
timate (perhaps up to 37%) of the total SCD population in
a state.
Despite the limitations to the data and methods, however, the estimate of overall prevalence of SCD (1:365),
and of sickle cell anemia (1:601) in the African-American
birth cohort, is consistent with other reports, supporting
some validity to the approach and data used for this
report. Similarly, the consistency between state sickle cell
population estimates for California using state-specifıc
data and pooled birth-cohort prevalence estimates derived from the 2005–2007 NNSIS suggests that the use of
the pooled data may be acceptable for a given state in the
absence of state-specifıc data.
The clear challenge to SCD population estimation is
the effect of early mortality. Contemporary national data
regarding sickle cell disease mortality are not readily
available. Published cohort data address sickle cell anemia, but not HbSC or other types of SCD. Assumption
that these other forms of SCD do not cause early mortality
may lead to an overestimation of the number of individuals
with these disorders. Accurate information for sickle cell
anemia is available from pediatric cohorts,13–16 but published data regarding adult survival are limited to data collected prior to the use of disease-modifying therapies
(CSSCD)12 and/or reported from Jamaica,16 –17 where the
course of the disease may be different.17 Interestingly, the
Jamaican group suggested that their ability to carefully
observe the majority of their population may have resulted in a more-accurate determination of the course of
SCD as compared with other countries where limited
cohort data are available.17
Within the U.S., annual mortality associated with the
diagnosis of SCD as reported by the CDC Compressed
Mortality Report seems to vary by percentage of at-risk
populations, but may be influenced by other factors.21
Analysis performed in this report used information available from death certifıcates reported to the CDC in an
effort to obtain national data, but it has a number of
important major limitations. The accuracy of diagnostic
coding is unknown, and there are too few individuals
with forms of SCD other than HbSS and those designated
as Hispanic reported to permit specifıc analysis. The
number of individuals who died with or of SCD without
documentation on the death certifıcate, or even the number of individuals that go unreported, is unknown. Finally, the average age of death (39 years) and the proportion of individuals surviving to age 45 years (35%) were
lower in 2006 than the proportion of individuals surviving (50%) as reported by the CSSCD using data from
before 1994, which seems incongruent with improved
health care and disease-remitting therapy. These fındings
may be due to differences in estimation of survival for a
specifıc cohort as compared to a cross-sectional sampling
of age at death. It is also possible that those in the CSSCD
and the pediatric and Jamaican cohorts received care
within “specialized” sickle cell centers, whereas individuals reported to the CDC did not, resulting in a difference
in outcomes. Unfortunately, it is not possible to sort this
out without accurate national data. Application of a
more-sophisticated approach to population and survival
estimation should be undertaken, but it will be similarly
limited by the paucity of high-quality data available at this
time.
In order to accurately assess the number of individuals
with SCD nationwide, a coordinated system of data collection and reporting will need to be established. Given
state-to-state variability observed in this report, application of a single state’s data to other states or national
population estimates is not likely to yield representative
information. Individual states can incorporate available
information from newborn screening and administrative
data sets for mortality to estimate the size and composition of their SCD population. Such state-specifıc information is likely to provide the best estimates for who
would address healthcare needs and health policy development within a catchment area. In turn, centralized and
standardized reporting of these data would possibly provide the best national assessment of the SCD population
and burden of disease.
No fınancial disclosures were reported by the author of
this paper.
References
1. Lukens JN. Sickle cell disease. Dis Mon 1981;27:1–56.
2. Sickle Cell Disease Guideline Panel. Sickle cell disease: screening, diagnosis, management, and counseling in newborns and
infants. Clinical Practice Guideline No. 6. AHCPR Pub. No. 93
0562. Rockville MD: Agency for Health Care Policy and Research, Public Health Service, USDHHS, 1993.
3. CDC. Health care professionals: data & statistics. www.cdc.
gov/ncbddd/sicklecell/hcp_data.htm.
4. National Heart Lung and Blood Institute. Who is at risk for sickle
cell anemia? www.nhlbi.nih.gov/health/dci/Diseases/Sca/SCA_
WhoIsAtRisk.html.
5. Strouse JJ, Jordan LC, Lanzkron S, Casella JF. The excess
burden of stroke in hospitalized adults with sickle cell disease.
Am J Hematol 2009;84(9):548 –52.
6. Kauf TL, Coates TD, Huazhi L, Mody-Patel N, Hartzema AG.
The cost of health care for children and adults with sickle cell
disease. Am J Hematol 2009:84:323–7.
7. U.S. Census Bureau. www.census.gov.
8. Therrell BL and Hannon WH. National evaluation of U.S.
newborn screening system components. MRDD Research Reviews 2006:12:236 – 45.
www.ajpm-online.net
Hassell / Am J Prev Med 2010;38(4S):S512–S521
9. Lorey FW, Arnopp J, Cunningham GC. Distribution of hemoglobinopathy variants by ethnicity in a multiethnic state. Genetic Epidemiol 1996;13:501–12.
10. Michlitsch J, Azimi M, Hoppe C, et al. Newborn screening for
hemoglobinopathies in California. Pediatr Blood Cancer
2009;52:486 –90.
11. National Newborn Screening Information System. www2.
uthscsa.edu/nnsis/.
12. Platt OS, Brambilla DJ, Rosse WF, et al. Mortality in sickle cell
disease. Life expectancy and risk factors for early death. NEJM
1994;330:1639 – 44.
13. Quinn CT, Rogers ZR, Buchanan GR. Survival of children with
sickle cell disease. Blood 2004;103:4023–27.
14. Telfer P, Coen P, Chakravorty S, et al. Clinical outcomes in
children with sickle cell disease living in England: a neonatal
cohort in East London. Haematologica 2007;92:905–12.
15. King L, Fraser R, Forbes M, Grindley M, Ali S, Reid M. Newborn sickle cell disease screening: the Jamaican experience
(1995–2006). J Med Screen 2007;14:117–22.
16. Wierenga KH, Hambleton IR, Lewis NA. Survival estimates
for patients with homozygous sickle-cell disease in Jamaica: a
clinic-based population study. Lancet 2001;357(9257):680 –3.
17. Serjeant GR, Serjeant BE, Mason KP, Hambleton IR, Fisher C,
Higgs DR. The changing face of homozygous sickle cell disease: 102 patients over 60 years. Int J Lab Hem 2009;31:585–96.
18. CDC, National Center for Health Statistics. Compressed mortality fıle 1999 –2006. CDC WONDER On-line Database, compiled from compressed mortality fıle 1999 –2006 series 20 No.
2L, 2009. wonder.cdc.gov/cmf-icd10.html.
19. Gaston MH, Verter JI, Woods G, et al. Prophylaxis with oral
penicillin in children with sickle cell anemia. A randomized
trial. NEJM 1986;314:1593–9.
20. Steinberg MH, Barton F, Castro O, et al. Effect of hydroxyurea on
mortality and morbidity in adult sickle cell anemia: risks and
benefıts up to 9 years of treatment. JAMA 2003;289:1645–51.
21. Ashley-Koch A, Yang Q, Olney RS. Sickle hemoglobin (HbS)
allele and sickle cell disease: a HuGE review. Am J Epidemiol
2000;151:839 – 45.
Did you know?
You can sign up for saved search and table of contents email alerts on the AJPM website.
Visit www.ajpm-online.net today!
April 2010
S521