Association Between Sickle Cell Trait With Selected Chronic Medical Conditions in U.S. Service Members

Abstract

Introduction

Sickle cell trait (SCT), the heterozygous carrier state for hemoglobin S, is present in an estimated 1.6% of all newborns and 7.3% in black individuals in the USA. SCT has long been considered a benign condition with anticipated normal life expectancy and no increased risk for chronic diseases. The medical literature is inconclusive on the potential association between SCT and chronic medical conditions (CMC) including chronic kidney disease, venous thromboembolism, and stroke. Studies addressing these questions are lacking particularly in non-Black young adults.

Materials and Methods

We conducted a retrospective cohort study among U.S. active duty, enlisted, service members who entered from 1992 to 2012 using existing Department of Defense (DoD Military Healthcare System databases). SCT positive subjects (1,323) were matched by demographic characteristics to SCT negative subjects (3,136) and followed through 2013 for CMC that included deep vein thrombosis, diabetes mellitus and hematologic, pulmonary, and renal conditions.

Results

The rate of developing any of the included CMC was higher for those with SCT (incidence rate ratio = 1.71 95% CI 1.61–1.81) compared with those who were SCT negative and their healthcare utilization rate for any of CMC studied was higher for SCT positive compared with negative individuals (URR = 2.45 95% CI 2.41–2.50), with the highest rate ratios observed for hematologic and renal conditions. SCT positive compared with negative individuals were more likely to have encounter diagnoses of sickle cell disease and diabetes Type II and were less likely to have encounter diagnoses of other hemoglobinopathies and diabetes type I.

Conclusion

SCT in these racially diverse, young adults increased both the incidence of and healthcare utilization for thromboembolism, diabetes mellitus type II, sickle cell disease, pulmonary, and chronic renal conditions. These findings suggest that clinicians treating young adults with SCT should exercise heightened surveillance for these CMC to ensure both early diagnosis and access to treatments.

INTRODUCTION

Sickle cell trait (SCT) was historically considered a benign carrier condition and not a disease. Individuals heterozygous for hemoglobin S tended to be free of clinical manifestations associated with sickle cell disease and had a normal life expectancy.¹ SCT occurs when one allele of the beta globin gene carries the sickle hemoglobin mutation and the other allele codes for normal hemoglobin (A), resulting in hemoglobin AS (HbAS). Comprehensive newborn screening in the U.S. demonstrated an overall SCT prevalence of 1.6% and a race-specific prevalence of 7.3 and 0.7% among Blacks and Hispanics, respectively.²

Individuals may have received SCT screening as part of preconception counseling. The National Collegiate Athletic Association (NCAA) mandated screening for all Division-I athletes.³

The Department of Defense conducts pre-induction health screening and medical examinations for all recruits, assessing medical suitability and physical capability for service in the military.⁴ As a part of this assessment, the Navy, Marine Corps, and Air Force screens all recruits for SCT while the Army screens those entering selected occupations only.⁵ While individuals homozygous for hemoglobin S are disqualified from entry, SCT is not. SCT prevalence estimates in the U.S. military based on a cohort from 1992 to 2012 were 5.02% of non-Hispanic blacks, 1.08% of Hispanics, and 0.1% of Whites.⁶

SCT has been associated with selected chronic medical conditions (CMC). Coexistence of SCT with other hemoglobinopathies, such as beta thalassemia and hemoglobin C are considered to have a similar clinical course to sickle cell disease with the associated clinical manifestations. Studies reported both positive and negative associations regarding the risk of chronic kidney disease (CKD),^7–9 venous thromboembolism (VTE),^10,11 diabetes,¹² cardiovascular,^13,14 pulmonary,¹⁵ and stroke¹⁶ in individuals with SCT. The variation in findings was due in part to variation in study designs, types of populations included, and length of follow-up for outcomes.

Naik et al estimated an increased risk of CKD associated with SCT based on 1,247 SCT African American individuals from five large population based, prospective cohort studies (OR = 1.57, 95% CI 1.34–1.84).⁹ Bucknor et al reported on the risk of pulmonary embolism (PE) and CKD among 13,964 SCT Kaiser Permanente patients in Northern California. The risk of PE was increased with SCT (RR = 1.37, 95% CI 1.07–1.75). The risk of CKD also increased with SCT status (RR = 1.13, 95% CI 1.03–1.23).¹³

Little et al reported on the risk of VTE and PE associated with SCT in a cohort of patients 18- to 75-yr-old from 609 UK general practices, demonstrating the risk of VTE was increased with SCT (OR = 1.78, 95% CI 1.18–2.69), while the risk of PE was also increased with SCT (OR = 2.27, 95% CI 1.17–4.39).⁹ Bell et al reported a lifetime risk of VTE in two cohort studies that included 19,599 middle age and older individuals of 8.1% (95% CI 7.1–8.7%) with the risk increasing to 18.2% in SCT individuals.¹⁰

Caughey et al reported results from the Atherosclerotic Risk in Communities, a population-based cohort study that included 3,497 black individuals of whom 223 were SCT with follow-up from 1987 until 2011. They observed an increased risk of ischemic stroke associated with SCT (HR = 1.4, 95% CI 1.0–2.0).¹⁶

Liem et al reported results from the Coronary Artery Risk Development in Young Adults (CARDIA) study. This retrospective, longitudinal cohort study of young adults ages 18–30 yr, followed for 25 yr, included 136 SCT individuals, reporting that SCT was not an independent risk factor for hypertension, diabetes, or metabolic syndrome.¹³

Lacy et al reported findings based on two combined retrospective, longitudinal cohort studies, including 4,620 participants with 367 SCT. They observed the prevalence of prediabetes and diabetes was lower in participants with SCT compared with participants without SCT. They also concluded that Hemoglobin A1c may systematically underestimate previous glycemic levels in black patients with SCT.¹²

This study explored the associations between SCT and CMC, including thromboembolic, hematologic, diabetes mellitus, pulmonary, and chronic renal diseases in U.S. enlisted service members.

METHODS

Study Subjects

We conducted a retrospective cohort study among U.S. enlisted military service members on active duty during 1992–2012 with follow-up through 2013, using the existing DoD Military Healthcare System databases. Eligible SCT positive individuals were identified in the following two ways: (1) medical encounter SCT diagnosis (ICD-9-CM 282.5) in ambulatory or hospitalized encounter records provided by Armed Forces Health Surveillance Center (AFHSC) or (2) documentation of a positive (by either qualitative or quantitative assays) laboratory SCT test result in the existing Military Health System (MHS) HL-7 laboratory database from 2006 to 2012 provided by the Navy and Marine Corps Public Health Center (NMCPHC). A stratified random sample of those with no documentation of SCT diagnosis, from active duty service members during the same time period identified the unexposed, SCT negative, comparison group. Four SCT negative individuals were selected for every SCT positive individual matched on gender, racial ethnic group (non-Hispanic White, non-Hispanic black, Hispanic, non-Hispanic Other/Unknown), service branch as well as quarter and year of enlistment. Demographic data provided by AFHSC included: age at enlistment, gender, race/ethnicity, and date of enlistment and discharge. CMC (any CMC, each CMC individually, and each CMC subcategories) were defined in three ways based on one or more inpatient and/or two or more outpatient encounter records with the same relevant ICD-9 CM diagnostic codes.

The SCT population, irrespective of CMC in the study compared with the total DoD force was described previously.⁶ In brief, the SCT population is over represented compared with DoD with non-Hispanic blacks (81.4 versus 20.1%), females (33.2 versus 13.7%), 18–19 yr olds at enlistment (53.5 versus 8.6%) and Navy service members (48.6 versus 26.8%).

Selected CMC examined for potential associations with SCT status included: deep vein thrombosis (ICD-9 CM 451.Z-453.X subcategorized as thromboembolism and thrombophlebitis), diabetes mellitus (ICD-9 CM 249.X, 250.X subcategorized as type I and II), hematologic (ICD-9 CM 282.X subcategorized as sickle cell disease and hemoglobinopathies), pulmonary (conditions ICD-9 CM 490.X-493.X, 496.X subcategorized as asthma and chronic obstructive pulmonary disease) and renal (chronic 585.X) conditions.

Statistical Analysis

Distributions among those meeting any CMC definition by demographic variables at enlistment and by SCT positive versus negative status were compared and tested with the Chi Square test. The crude and adjusted incidence rate ratios (IRR), healthcare utilization rate ratios (URR), and 95% CI for any and each CMC were calculated using multivariable Poisson regression. The following potential confounders were considered for inclusion in our multivariable models: racial/ethnic group, gender, age at enlistment, length of military service, and service branch. Stepwise variable selection eliminated those variables with P > 0.1 for the most parsimonious model. The length of service in years from enlistment to incident CMC diagnosis and from incident CMC diagnosis until separation from the military were compared by SCT status by calculating the median and interquartile (25–75%) range (IQR) and using the Wilcoxon-Rank-Sum test. A P-value of <0.05 was considered statistically significant. All analyses were conducted using SAS, (Version 9.3), Cary, NC, USA.

The USUHS Institutional Review Board approved this protocol. The study was funded by an Interagency Agreement with the National Heart, Lung and Blood Institute.

RESULTS

Table I described the demographic characteristics for SCT positive versus negative subjects at enlistment who were matched on gender, race, and branch of service and with at least one inpatient and/or two or more outpatient medical encounter records with any of the five selected CMC diagnostic codes. CMCs were more common in the Army versus the other services for SCT positive compared with negative subjects, while the distribution for gender, age and race were similar between these two groups. Similar distributions were found when subjects were restricted to non-Hispanic blacks (results not shown). SCT positive versus negative subjects without any medical record encounters for a CMC diagnosis were more likely to be male, 25 yr of age and older, and in the Navy.

Table I.

Demographics of Study Population by CMC and SCT Status

Demographics	At Least One CMCa			No CMC
	SCT Positive	SCT Negative	P-valueb	SCT Positive	SCT Negative	P-valueb
	N = 1,549 n (%)	N = 3,624 n (%)		N = 13,531 n (%)	N = 56,696 n (%)
Gender
Male	721 (46.5)	1,768 (48.8)	0.1398	9355 (69.1)	38,536 (68.0)	0.0088
Female	828 (53.5)	1,856 (51.2)		4,176 (30.9)	18,160 (32.0)
Age
18–19	772 (49.8)	1,922 (53.0)	0.0697	7,291 (53.9)	31,268 (55.2)	0.0047
20–24	587 (37.9)	1,331 (36.7)		4,993 (36.9)	20,648 (36.4)
25–29	134 (8.7)	271 (7.5)		943 (7.0)	3,684 (6.5)
30+	56 (3.6)	100 (2.8)		304 (2.2)	1,096 (1.9)
Race
Non-Hispanic white	47 (3.0)	114 (3.2)	0.1542	630 (4.7)	2,597 (4.6)	0.8899
Non-Hispanic black	1,323 (85.4)	3,136 (86.5)		10,957 (81.1)	45,982 (81.1)
Non-Hispanic others	88 (5.7)	153 (4.2)		772 (5.7)	3,291 (5.8)
Hispanic	91 (5.9)	221 (6.1)		1,172 (8.7)	4,826 (8.5)
Service
Army	598 (38.6)	1,210 (33.4)	0.0036	2,679 (19.8)	11,942 (21.1)	0.0140
Navy	547 (35.3)	1,354 (37.4)		6,780 (50.1)	27,926 (49.3)
Air Force	318 (20.5)	835 (23.0)		3,170 (23.4)	13,109 (23.1)
Marine	86 (5.6)	225 (6.2)		902 (6.7)	3,719 (6.6)

^aCMC defined as having at least one inpatient &/or 2+ outpatient encounter records.

^bP-value based on Chi Square test.

Table II describes the distribution of inpatient, outpatient (two or more occurrences) and inpatient and/or outpatient visits by any CMC and the five individual CMC categories. A higher percentage of SCT positive compared with negative individuals had an inpatient visit (2.63% versus 0.98%), outpatient (9.38% versus 5.72%), and inpatient and/or outpatient encounters (10.27% versus 6.01%) for any of these CMC. A similar pattern of greater percentage of individuals having medical encounters for the individual CMC categories among the SCT positive subjects was observed. These same patterns persisted when analyses were restricted to non-Hispanic blacks (results not shown). All subsequent analyses used the inpatient and/or 2+ outpatient encounter records definition to define CMC.

Table II.

Distribution of CMC Encounter Records (Inpatient Encounter and/or 2 or More Outpatient) by SCT Positive (15,080) and SCT Negative (60,320) Status

Distribution of CMCa encounter records
CMC	CMC Inpatients				CMC Outpatients 2+ Records				CMC Inpatient &/or Outpatient 2+
	SCT Positive		SCT Negative		SCT Positive		SCT Negative		SCT Positive		SCT Negative
	n	%	n	%	n	%	n	%	N	%	n	%
Any chronic condition	396	2.63	589	0.98	1,415	9.38	3,450	5.72	1,549	10.27	3624	6.01
Deep vein thrombosis	24	0.16	35	0.06	57	0.38	121	0.20	66	0.44	132	0.22
Diabetes	48	0.32	81	0.13	106	0.71	225	0.37	117	0.78	236	0.39
Hematologicb	146	0.97	39	0.06	308	2.10	84	0.14	395	2.62	107	0.18
Pulmonary	183	1.21	436	0.72	995	7.07	3043	5.36	1041	6.90	3181	5.27
Renal	12	0.08	10	0.02	33	0.22	42	0.07	38	0.25	47	0.08

^aCMC defined as having at least one inpatient and/or 2+ outpatient encounter records.

^bGlutathione deficiency excluded from hematologic.

Table III shows the incidence and IRR for any CMC and each specific CMC adjusted for age, gender, and race for SCT positive versus negative subjects. Any CMC had an IRR of 1.71 (95% CI 1.61–1.81). Hematologic conditions had the highest IRR of 14.67 (95% CI 11.85–18.16). The IRR was significant for each of the other four CMCs and ranged from 1.31 (95% CI 1.22–1.40) for pulmonary to 3.15 (95% CI 2.05–4.83) for chronic renal diagnoses. Similar results were observed when subjects were restricted to non-Hispanic blacks.

Table III.

Incidence and IRR by SCT Positive Versus Negative Status for Selected CMC^a

Incidence of CMCsa
CMC	SCT Positive 81,958 p-y		SCT Negative 262,609 p-y		IRR	IRR 95% CI	Adjusted Poisson Modelc
	n	Incidence Rate/100 p-y	n	Incidence Rate/100 p-y			IRR	IRR 95%CI
Any chronic condition	1,549	1.89	3,624	1.38	1.38	1.30–1.46	1.71	1.61–1.81
Deep vein thrombosis	66	0.08	132	0.05	1.61	1.20–2.16	1.98	1.47–2.66
Diabetes	117	0.14	236	0.09	1.60	1.28–1.99	1.95	1.57–2.44
Hematologicb	395	0.48	107	0.04	11.89	9.60–14.72	14.67	11.85–18.16
Pulmonary	1,041	1.27	3,181	1.21	1.05	0.98–1.13	1.31	1.22–1.40
Chronic renal	38	0.05	47	0.02	2.60	1.70–3.99	3.15	2.05–4.83

^aCMC defined as having at least one inpatient and/or 2+ outpatient encounter records.

^bGlutathione deficiency excluded from

hematologic and any chronic disease.

^cSCT is independent variable, and model is adjusted for age, gender, and race.

Table IV shows the healthcare utilization rate and URR for any CMC and each specific CMC adjusted for age, gender, and race for SCT positive versus negative subjects. Any CMC had an URR of 2.45 (95% CI 2.41–2.50) and hematologic conditions had the highest URR of 30.94 (95% CI 28.15–34.01). The URR was significant for each of the other four CMCs and ranged from 1.59 (95% CI 1.55–1.63) for pulmonary to 20.07 (95% CI 17.80–22.62) for chronic renal diagnoses. Similar results were observed when subjects were restricted to non-Hispanic blacks.

Table IV.

Outpatient (2+) URR for Selected CMC by SCT (Positive Versus Negative) Status

CMCa Utilization Rate
CMC	SCT Positive (14,054 p-y)		SCT Negative (56,216 p-y)		Unadjusted Model		Adjusted Poisson Modelb
	Visits	Utilization Rate/100 p-y	Visits	Utilization Rate/100 p-y	URR	95% CI	URR	95% CI
Any chronic condition	17,706	125.99	28,731	51.11	2.04	2.00–2.08	2.45	2.41–2.50
Deep vein thrombosis	755	5.37	1,235	2.2	2.02	1.85–2.22	2.43	2.22–2.66
Diabetes	2,910	20.71	5,025	8.94	1.92	1.83–2.01	2.29	2.19–2.40
Hematologic	3,756	26.73	486	0.86	25.59	23.28–28.13	30.94	28.15–34.01
Pulmonary	8,637	61.46	21,665	38.54	1.32	1.29–1.35	1.59	1.55–1.63
Renal	1,648	11.73	320	0.57	17.05	15.13–19.22	20.07	17.80–22.62

^aCMC defined as having at least one inpatient and/or 2+ outpatient encounter records.

^bNumber of visit is dependent variable, SCT is independent variable, and model is adjusted for age, gender, and race.

Table V specifies the major diagnoses within each CMC, except for chronic renal conditions. Diabetes type II compared with type I was more common in SCT positive (80.2%) versus negative (72.4%) subjects. Sickle cell diseases (includes hemoglobin SS including with or without crisis, sickle cell-hemoglobin C disease and unspecified) compared with hemoglobinopathies (including thalassemias and other forms) was more common in SCT positive (50.4%) versus negative (15.4%) subjects. The distribution of thromboembolism versus thrombophlebitis and for asthma versus chronic obstructive pulmonary disease was not statistically significantly different between SCT positive versus negative subjects. Similar results were observed when analyses only included non-Hispanic Blacks.

Table V.

Frequency of Specific CMC Inpatient and/or 2+ Outpatient Encounter Records by SCT Positive Versus Negative Status

Frequency of CMCa Encounters
CMC	SCT Positive (9,325)		SCT Negative (59,314)		p-Value
	N	%	N	%
Deep vein thrombosis	129	34	250	66
Thromboembolism	51	39.5	87	34.8	0.364
Thrombophlebitis	78	60.5	163	65.2
Diabetes	268	32.9	547	67.1
Type I	53	19.8	151	27.6	0.0154
Type II	215	80.2	396	72.4
Hematologic	649	82.7	136	17.3
Hemoglobinopathies	322	49.6	115	84.6	<0.0001
Sickle cell disease	327	50.4	21	15.4
Pulmonary	1847	25.2	5488	74.8
Asthma	1291	69.9	3912	71.3	0.2566
Chronic obstructive pulmonary diseaseb	556	30.1	1576	28.7
Renal	61	45.5	73	54.5

^aCMC defined as having at least one inpatient and/or 2+ outpatient encounter records.

^bCOPD include Bronchitis NOS, Chronic bronchitis, Emphysema and Chronic airway obstruct NEC.

SCT positive compared with negative subjects experienced a longer time from enlistment to incident CMC diagnosis (median of 1.1 versus 0.6 yr, p < 0.001, results not shown). Deep vein thrombosis, diabetes, and pulmonary disease had similar significant findings while no statistically significant differences were observed for hematologic and chronic renal conditions (p > 0.05). Similar results were observed when subjects were stratified by service and when restricted to non-Hispanic blacks.

SCT positive compared with negative subjects experienced a longer time from incident CMC diagnosis to separation from the military (median of 2.6 versus 2.1 yr, p < 0.001, results not shown). Duration of service was longer for the SCT positive individuals compared with the negative among those that were diagnosed with diabetes and pulmonary conditions. No significant difference was observed for those diagnosed with deep vein thrombosis, hematologic and chronic renal conditions (p > 0.05). Similar results were observed when subjects were stratified by service and when restricted to non-Hispanic blacks.

DISCUSSION

We found that the incidence of selected CMCs in SCT positive subjects were more common in the Army (the only service that does not universally screen enlistees for SCT) versus the other services, but did not vary by gender, age, and race of subjects. The higher percentage of SCT positive subjects with a CMC was observed for both inpatient and outpatient encounters. Diagnoses of sickle cell disease (e.g., SC disease) compared with hemoglobinopathies and diabetes type II compared with I were more common in SCT positive subjects. The incidence and healthcare utilization rates for any and each CMC individually were higher in SCT positive compared with negative subjects with the highest rates observed for hematologic and chronic renal conditions. Race was not found to be an effect modifier for these associations. Paradoxically SCT positive compared with negative subjects were diagnosed with a CMC later and stayed longer in military service. We speculate that service members known to have SCT may have a self-perceived incentive to stay in the military due to access to universal healthcare and concerns about healthcare and employability outside of the military. All service members, irrespective of SCT status, are subject to the same retention and fitness-for-duty standards, possibly limiting the choice to remain in service for those with a CMC. Further research, including a cost-effectiveness study on retention of individuals with CMCs is warranted.

The findings of this study of a positive association with SCT are consistent with the literature for incidence of the following CMC despite different populations and length of follow-up: deep vein thrombosis,^10,11,14 chronic kidney disease^9,14 Given the relatively young age of the population studied, we did not test for an association of SCT with stroke¹⁶ or cardiovascular disease risk factors.¹³ We observed an association with diabetes, specifically type II, which has not been observed by others.¹² This may be explained by the fact that we relied on medical encounter diagnostic codes without verification of diagnostic criteria, whereas Lacey measured serum glucose and hemoglobin A1c in study subjects.

We are not aware of other cohort studies that have either studied or found an association between SCT and pulmonary conditions as we observed, but studies of other design such as case-control and case series have suggested a relationship.¹⁵

Strengths of this study design include a large sample size of young adult racially diverse individuals, an open cohort with longitudinal follow-up for military and medical outcomes and minimal loss to follow-up during military service. This study is unique in that the outcomes of interest for the selected CMC included both incidence and healthcare utilization; the latter is a proxy for severity of the condition. Our findings may be generalized to other nation’s military forces that are similar in demographic characteristics, accession standards and military service requirements. Limitations to the study include the potential for misclassification of SCT status as false negatives which would likely bias our results towards the null (no association). Finally, the uniquely military enlistment and service requirements limit the generalizability of our CMC incidence and healthcare utilization findings to other physically demanding occupations such as law enforcement and fire fighters where periodic health assessments are required.

Determination of the incidence of CMC assumed that subjects were disease free at enlistment which is not likely a source of misclassification because of the low incidence of CMC in adults under the age of 30 and the medical standards and screening process prior to enlistment. Information for potential confounders such as body mass index, smoking, medications (including oral contraceptives), family history of CMCs (e.g., diabetes, deep vein thrombosis and chronic renal disease) and physical activity prior to enlistment were not available. Finally, the diagnostic accuracy of medical encounter records was not validated, but we have no reason to believe there was a differential bias based on SCT status.

Future research should include expanding this cohort and following subjects beyond military service as retirees (TRICARE beneficiaries) or veterans (Veterans Health Administration beneficiaries) to study the association with other conditions such as cardiovascular disease and stroke that typically have onset in individuals over the age of 50 yr. In addition, this longitudinal study design and including the spectrum of CMCs potentially associated with SCT should be replicated in other young adult populations.

SCT in young adult service members was associated with both the incidence and healthcare utilization for thromboembolism, diabetes mellitus type II, sickle cell disease, pulmonary, and chronic renal conditions. If these findings are reproduced in other young adult populations, clinicians should perform heightened surveillance for these CMC to ensure both early diagnosis and treatment.

FUNDING

The study was funded by an Interagency Agreement with the National Heart, Lung, and Blood Institute/NIH.