Sickle cell trait and renal disease among African American U.S. Army soldiers

Summary

Sickle cell trait and certain renal disorders are disproportionately prevalent among African American individuals, so a clear understanding of their association is important. We conducted a longitudinal study using the Stanford Military Data Repository to examine sickle cell trait in relation to the incidence of acute kidney injury (AKI) and chronic kidney disease (CKD). Our study population consisted of African American U.S. Army soldiers on active duty between January 2011 and December 2014. The cumulative incidence was 0.51% for AKI (236 cases out of 45,901 soldiers) and 0.56% for CKD (255 cases out of 45,882 soldiers). Discrete time logistic regression models adjusting for demographic-, military- and healthcare-related covariates showed that sickle cell trait was associated with significantly higher adjusted odds of both AKI (odds ratio [OR]: 1.74; 95% confidence interval [CI]: 1.17 – 2.59) and CKD (OR: 2.00; 95% CI: 1.39 – 2.88). Elevated odds of AKI and CKD were also observed in association with prior CKD and AKI, respectively, and with obesity and prior hypertension. Individuals with sickle cell trait and their providers should be aware of the possibility of increased risk of AKI and CKD to allow for timely intervention and possible prevention.

Introduction

Individuals with sickle cell trait (SCT) are heterozygous for the sickle cell mutation in the haemoglobin beta gene (HBB), leading to the presence of both wild-type haemoglobin (Hb) and HbS. It is estimated that 7–8% of African Americans are born with SCT, based on Centers for Disease Control and Prevention (CDC) findings (Centers for Disease Control and Prevention, 2016a) and on the most recent results from 17 states that provided information on both SCT and race/ethnicity collected during newborn screenings (Ojodu et al, 2014). While SCT is thought to be a generally benign condition, recent reports have raised concerns that individuals with SCT may be at increased risk of a number of renal disorders, including chronic kidney disease (CKD), end-stage renal disease (ESRD), renal medullary carcinoma, a more rapid yearly decrease in estimated glomerular filtration rate than non-carriers, a defect in urinary concentrating ability (hypostenuria), and the urinary findings of albuminuria and/or haematuria (Alvarez et al, 2015; Becton et al, 2010; Bucknor et al, 2014; Dueker et al, 2017; Key et al, 2015; Naik et al, 2016a). As a set, renal disorders are associated with serious morbidity. Due to the higher prevalence among African Americans of pre-disposing conditions, including hypertension and diabetes (Flessner et al, 2009), an accurate estimate of the risk of renal dysfunction in association with SCT in this population is critical.

Renal disorders are diverse but can be divided into two broad categories – acute kidney injury (AKI) and CKD – depending on the aetiology and subsequent course of the condition. AKI is a sudden episode of a decrease in kidney function and structural kidney damage occurring within a few hours or a few days, caused by decreases in renal perfusion (pre-renal), obstruction of the urinary tract (post-renal) or direct damage to the kidneys (intrinsic).

A consensus-based, widely accepted definition of AKI is provided by the KDIGO (Kidney Disease: Improving Global Outcomes) guideline that uses time-based changes in serum creatinine or urine volume as its criteria (Kellum et al, 2012). CKD is defined by abnormalities of kidney structure or function, present for three or more months, with serious implications for health (KDIGO, 2013). Depending on the stage of CKD, signs and symptoms can vary from very few to substantial impairments. CKD can have widespread physiological effects, including substantial morbidity and early mortality (Hallan et al, 2012).

There is an increased risk of having recurrent AKI or of developing other health disorders, such as kidney failure, stroke or heart disease after an initial episode of AKI (National Kidney Foundation, 2017). Recurrent AKI after an initial hospital AKI diagnosis is relatively common, and risk stratification approaches to these patients based on presenting factors have been proposed (Siew et al, 2016). Data from a retrospective follow-up of patients treated in the Veterans Affairs (VA) system in the United States showed a significant risk for the development of CKD in the year following a single hospital-acquired AKI event, even in a mild form (Heung et al, 2016).

While a positive relationship between SCT and CKD has been reported by several studies (Naik et al, 2016b; Bucknor et al, 2014), much less is known about SCT’s relationship to AKI. Only one previous study, using hospital-based data, has been published on the topic: it reported adjusted relative risks of AKI for non-Hispanic African American participants with sickle cell disease (SCD), SCT and normal haemoglobin (Bucknor et al, 2014). In that study, while SCD was highly associated with an increased risk of AKI, SCT was not.

Given the seriousness of renal disorders as a set, additional knowledge about possible renal sequelae among those with SCT is essential. We have therefore leveraged data on a large population of sickle-cell tested, African American U.S. Army soldiers with longitudinal health histories to test for associations between SCT and incident AKI and CKD while controlling for co-morbidities, including prior renal disease of the other type.

Methods

Study Design

We conducted a longitudinal study of SCT in relation to renal disease using data from the Stanford Military Data Repository (SMDR). The SMDR includes largely de-identified demographic, administrative and healthcare encounter data for all soldiers on active duty in the U.S. Army anytime between 2011 to 2014 (See Table I for information about the specific data sources that were combined to create the datasets used for this study). This study was approved by the institutional review board of Stanford University and underwent secondary review by the Defense Health Agency’s Human Research Protection Office.

Table I.

Descriptions of the sources and types of data from the U.S. military used to produce the research datasets.

Defense Manpower Data Center

Active Duty Master File – Extract from official Department of Defense (DoD) records of demographic and military service data. Transactions File – Extract from official DoD records of military duty status changes, such as completion of service.

Military Health System Data Repository

Combined Ambulatory Professional Encounter Record: Extract of records of outpatient care in military facilities. Tricare Encounter Data, Non-Institutional: Extract of records of outpatient care in civilian facilities reimbursed by military health coverage. Standard Inpatient Data Record: Extract of records of inpatient care in military facilities. Tricare Encounter Data, Institutional: Extract of records of inpatient care in civilian facilities reimbursed by military health coverage. Clinical Data Repository Vitals File: Extract of records of height and weight readings at outpatient encounters in military facilities. Pharmacy Detail Transaction Service: Extract of records of medications dispensed by military or civilian pharmacies.

Medical Operational Data System

Periodic Health Assessment: Extract of data from required annual health screenings at military health facilities. eProfile: Electronic archive of formal duty and activity restrictions or “profiles,” with associated clinical reasons.

Digital Training Management System

Army Body Composition Program: Data from required, biannual height and weight assessments.

Study population

The study population comprised all African American U.S. Army soldiers serving on active duty between January 2011 and December 2014 who had laboratory results from tests of the sickle cell phenotype. Self-reported race, per data from the Defense Manpower Data Center, was recorded as White, Black, Asian, Pacific Islander, American Native and Other/Unknown. In this study, we included only African American soldiers because of the substantially higher prevalence of SCT in this group.

All soldiers newly enlisting during the study period were included in the analyses because prior renal disease would disqualify individuals from entering military service (US Army, 2008).

Among soldiers who initiated service prior to January 2011, we initiated observation for outcomes at the first health screening encounter in their longitudinal record that was preceded by a minimum of six months of observation time (i.e., in or after July 2011). Health screenings included physical examinations, which occur every five years or due to administrative reasons such as certain pending training programmes (US Army, 2008), or the annual Periodic Health Assessments required of Army soldiers (US Army, 2018), which may lead to full physical examinations. These screenings are useful for detecting any substantial medical problems among soldiers because clinicians must regularly confirm their medical readiness for strenuous duty, including combat deployment. Using these encounters helped us ensure the capture of incident events rather than follow-up care. Only those soldiers with no diagnoses of AKI prior to or at the first observed health screening were included in the AKI analytic population. Similarly, only those soldiers with no indication of CKD prior to or at that encounter were included in the CKD analytic population. However, participants with prior AKI were included in the analyses targeting incident CKD, and participants with prior CKD were included in the analyses targeting incident AKI.

Outcome Measurement

We examined AKI and CKD as two separate outcomes. Much like civilian settings, and as with any patient, soldiers who present for medical care for particular complaints undergo laboratory tests to clarify diagnoses. Clinicians take a medical history and have access to past electronic medical records. Laboratory testing is done if there is an indication discovered during this process, although it can also be performed as a part of physical examinations required for training. As with other typical medical systems, clinicians make diagnoses using the Army’s electronic health record system (AHLTA). The outcomes were primarily identified with International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes from inpatient and outpatient health care encounters taking place both inside and outside of military facilities. Specifically, AKI was defined as acute renal failure (ICD-9 code 584) and renal failure, unspecified (ICD-9 code 586). The CKD definition comprised chronic kidney disease (ICD-9 code 585), chronic glomerulonephritis (ICD-9 code 583), renal sclerosis (ICD-9 code 587) and nephrotic syndrome (ICD-9 code 581), including ESRD (ICD-9 code 585.6) that results from chronic disease progression. Several additional cases of renal disease were also identified from the “eProfile” system (https://www.dvidshub.net/news/208423/upgraded-e-profile-increasing-readiness-says-army-surgeon-general), which records the clinician’s free text entries of medical conditions associated with duty limitations. To identify additional cases of AKI, we used text searches of eProfile utilizing upper- and lower-case forms of terms such as acute kidney injury, acute renal injury, kidney injury, renal injury and renal failure. Similarly, occurrences of CKD from eProfile were ascertained using search terms such as chronic kidney failure, chronic renal insufficiency, end-stage renal disease and nephrotic syndrome.

Independent Variables

Sickle cell trait:

The primary independent variable was SCT status. SCT was identified from laboratory haemoglobin electrophoresis tests of the Hb AS phenotype.

Demographic factors:

Demographic characteristics included gender and age. Age was categorized into five groups: ≤ 22; 23–27; 28–35; 36–41; 42–49 and ≥50 years of age.

Military-specific factors:

We included service time and military pay-grade category [Defense Finance Accounting Service (DFAS), 2018] in the models for the purpose of adjusting for socioeconomic status. Prior deployment experience was also controlled for, given that deployment could potentially increase the risk for renal outcomes due to stress, exertion or potential injury with blood loss.

Health-related factors:

The most recent body-mass index (BMI; kg/m²) of the participants was categorized based on the standard classification (Centers for Disease Control and Prevention, 2016b) with an additional category for those with missing BMI data. Underweight participants (BMI < 18.5 kg/m²) and normal weight participants (BMI from 18.5 to 24.9 kg/m²) were further combined as the reference group, since fewer than 0.3% of the participants were underweight in both the AKI and CKD analyses.

We also included as covariates prior diagnoses of hypertension (ICD-9 codes: 401, 402 or 405), diabetes (ICD-9 code: 250), and rhabdomyolysis (ICD-9 codes: 728.88 or 791.3), each as a binary variable. Prior history of AKI was included as a covariate in the CKD analyses and prior history of CKD was included as a covariate in the AKI analyses. Prescription of non-steroidal anti-inflammatory drugs (NSAIDs) was also included as a covariate. We calculated the mean of each soldier’s total days’ supply of NSAIDs per month during the most recent six months, and then categorized this variable into: no use, 1 to 5 days, 6 to 9 days and ≥ 10 days.

Statistical Analysis

Descriptive statistics (frequency and percentage) and chi-square tests were used to compare the demographic and health-related characteristics of SCT-tested African American soldiers with and without AKI or CKD. Discrete time logistic regression models were used to test the association of SCT status with AKI and with CKD, adjusting for observed time as well as the other independent variables described. All statistical analyses were conducted with Stata 14 software (StataCorp, College Station, TX) and all reported P-values are two-sided.

Results

A total of 45,901 and 45,882 soldiers were included in the analyses for AKI and CKD, respectively. The cumulative incidence of AKI was 0.51% (236 cases) and the cumulative incidence of CKD was 0.56% (255 cases).

Acute Kidney Injury

Soldiers with SCT comprised 7.4% of those individuals without AKI and 12.3% of those with AKI (Table II, p < 0.004). Soldiers with AKI were significantly more likely to be male and in older age categories (Table II). They were also significantly more likely to be in the overweight and obese categories. The proportion of soldiers with AKI that had CKD in the past six months or more was 5.0 times that of soldiers without AKI (p < 0.001) Participants with AKI also had a significantly higher prior prevalence of other comorbidities than those without the disorder – 2.7 times the prevalence of hypertension, 3.5 times the prevalence of diabetes and 4.3 times the prevalence of rhabdomyolysis (Table II).

Table II.

Characteristics of the analytic study populations, stratified by presence of outcome. Participants were African American U.S. Army Soldiers with SCT test results on active-duty in the U.S. Army between 2011 and 2014. Values are column percentages unless otherwise indicated.

	Acute Kidney Injury Analysis (N=45,901)			Chronic Kidney Disease Analysis (N=45,882)
	No AKI (N = 45,665)	AKI (N = 236)	P-value	No CKD (N = 45,627)	CKD (N = 255)	P-value
SCT Status			0.004			< 0.0001
no SCT	92.6	87.7		92.6	86.3
has SCT	7.4	12.3		7.4	13.7
Gender			< 0.0001			< 0.0001
Female	28.3	9.8		28.3	14.9
Male	71.7	90.2		71.7	85.1
Age Category (years)			< 0.0001			< 0.0001
≤22	33.3	19.9		33.4	4.7
23–27	26.4	20.8		26.5	11.4
28–35	22.6	26.7		22.7	20.8
36–41	11.1	20.8		11.0	31.4
42–49	5.9	9.7		5.8	25.9
50+	0.7	2.1		0.6	5.9
BMI			< 0.0001			< 0.0001
Under/Normal Weight	22.0	14.8		22.1	9.0
Overweight	32.1	34.8		32.1	45.5
Obese	13.4	25.8		13.3	32.9
Unknown	32.5	24.6		32.6	12.5
Renal Disease (Other Type)			< 0.0001			< 0.0001
No	99.4	97.0		99.6	96.1
Yes	0.6	3.0		0.4	3.9
Hypertension			< 0.0001			< 0.0001
No	89.7	72.0		89.9	52.5
Yes	10.3	28.0		10.1	47.5
Diabetes			< 0.0001			< 0.0001
No	98.8	95.8		98.8	94.9
Yes	1.2	4.2		1.2	5.1
Rhabdomyolysis			< 0.0001			< 0.0001
No	99.2	96.6		99.1	96.9
Yes	0.8	3.4		0.9	3.1

AKI: acute kidney injury; BMI: body mass index; CKD: chronic kidney disease; SCT: sickle cell trait

In the adjusted analyses, SCT was associated with a significant, 74% increase in the odds of AKI, after controlling for demographic, military and health-related covariates (odds ratio [OR]: 1.74; 95% CI: 1.17 – 2.59) (Table III). The odds of AKI were also significantly higher among male and obese soldiers. Finally, we also observed substantially higher adjusted odds of AKI among participants with a prior diagnosis of CKD (OR: 2.64; 95% CI: 1.18–5.90), hypertension (OR: 2.79; 95% CI: 1.91–4.05) or rhabdomyolysis (OR: 5.20; 95% CI: 2.53–10.70).

Table III.

Adjusted odds ratios, 95% confidence intervals and p-values from discrete time logistic regression models predicting AKI and CKD among SCT-tested African American U.S. Army Soldiers.

	AKI (N=45,901)		CKD (N=45,882)
	Odds Ratio (95% CI)	P Value	Odds Ratio (95% CI)	P Value
SCT Status
no SCT	Reference		Reference
has SCT	1.74 (1.17, 2.59)	0.006	2.00 (1.39, 2.88)	<0.0001
Gender
Female	Reference		Reference
Male	3.23 (2.08, 5.02)	<0.0001	1.82 (1.28, 2.59)	0.001
Age Category (years)
≤22	Reference		Reference
23–27	0.89 (0.53, 1.50)	0.664	2.70 (1.03, 7.05)	0.043
28–35	0.94 (0.53, 1.67)	0.845	4.08 (1.47, 11.31)	0.007
36–41	0.85 (0.44, 1.64)	0.630	6.51 (2.29, 18.53)	<0.0001
42–49	1.11 (0.53, 2.28)	0.777	11.62 (4.05, 33.33)	<0.0001
50+	0.90 (0.29, 2.84)	0.861	19.24 (6.24, 59.28)	<0.0001
BMI
Under/Normal Weight	Reference		Reference
Overweight	1.21 (0.83, 1.75)	0.322	1.70 (1.11, 2.59)	0.014
Obese	1.64 (1.08, 2.50)	0.020	1.99 (1.27, 3.10)	0.003
Unknown	0.54 (0.22, 1.30)	0.168	0.18 (0.02, 1.35)	0.096
Renal Disease (Other Type)
No	Reference		Reference
Yes	2.64 (1.18, 5.90)	0.018	6.07 (2.87, 12.85)	<0.0001
Hypertension
No	Reference		Reference
Yes	2.79 (1.91, 4.05)	<0.0001	4.13 (3.05, 5.58)	<0.0001
Diabetes
No	Reference		Reference
Yes	1.78 (0.88, 3.59)	0.108	1.06 (0.60, 1.86)	0.841
Rhabdomyolysis
No	Reference		Reference
Yes	5.20 (2.53, 10.70)	<0.0001	2.12 (0.91, 4.93)	0.081

All models additionally controlled for rank category, service time, number of deployments and non-steroidal anti-inflammatory drug use.

AKI: acute kidney injury; BMI: body mass index; CI: confidence interval; CKD: chronic kidney disease; SCT: sickle cell trait

Chronic Kidney Disease

Sickle cell trait was present in 7.4% of soldiers without CKD and in 13.7% of soldiers with CKD (Table II, p < 0.0001). A significantly higher proportion of soldiers with CKD were male and in the older age categories as compared to those without CKD (p < 0.0001). The prevalence of overweight and obesity were also significantly higher for those with CKD (p < 0.0001). Finally, the prevalence of prior comorbidities was much higher among soldiers with CKD – with 9.8 times the prevalence of prior AKI, 4.7 times the prevalence of prior hypertension, 4.3 times the prevalence of prior diabetes and 3.4 times the prevalence of prior rhabdomyolysis (Table II).

In the adjusted analyses, soldiers with SCT had double the odds of developing CKD as compared with soldiers without SCT (OR: 2.00; 95% CI: 1.39 – 2.88), after controlling for all of the demographic and medical covariates (Table III). There was a striking association between age and the presence of CKD, with an OR of 11.62 (95% CI: 4.05, 33.33) for those in the age group 42–49 relative to the reference category of <22 years of age, and an OR of 19.24 (95% CI: 6.24, 59.28) for those in the age group 50 years and over. The odds of developing CKD were also significantly higher for overweight and obese participants (OR: 1.70; 95% CI: 1.11–2.59 and OR: 1.99; 95% CI: 1.27–3.10, respectively). Further, the odds of having CKD were also substantially elevated among those with prior AKI or hypertension (Table III).

Discussion

Using a four-year longitudinal database of SCT-tested African American U.S Army soldiers, we found that SCT was associated with a significantly increased risk of both AKI and CKD in this active, largely healthy population, after controlling for demographic, military and clinical characteristics. To our knowledge, the significant relationship we observed between SCT and AKI is the first of its kind to be reported. These observed associations are important because of the prevalence of SCT among African Americans (7–8%) as well as the serious morbidity and mortality associated with renal disease.

Our findings on SCT in relation to CKD (OR: 2.00; 95% CI: 1.39 – 2.88) are consistent with those of two previous studies, which reported that individuals with SCT had an increased risk of CKD (OR: 1.79; 95% CI: 1.45–2.20) and ESRD (HR: 2.03; 95% CI: 1.44, 2.84) compared with noncarriers (Naik et al, 2014; Naik et al, 2016b). We were not able to analyse ESRD separately, as there were very few ESRD cases (only 4 of the 255 CKD cases). However, a sensitivity analysis where we removed the participants with ESRD did not change our results.

The previous literature regarding SCT and AKI is very limited. One prior study showed no significant difference in the prevalence of AKI for patients with SCT compared to an Hb AA cohort (relative risk: 0.97; 95% CI: 0.83 – 1.15) (Bucknor et al, 2014). One potential explanation for the different findings could be the different study populations employed. While Bucknor et al (2014) conducted their study on hospital-registered adults, our study population was comprised of largely healthy, young, active soldiers, which may result in a relatively lower rate of AKI. The two studies also utilized different measures – we used incidence rather than prevalence. Other prior studies reporting on SCT and AKI have only done so in the context of SCT increasing risk of exertional rhabdomyolysis, with AKI as a sequela (George, 1979; Makaryus et al 2007). We controlled for rhabdomyolysis in our model so as to identify any independent effects of SCT on AKI.

The question of why SCT-positive individuals might develop renal functional and structural disorders is an important one. One early study comparing vascular architecture among individuals with SCD, SCT and non-carriers concluded that individuals with SCT may develop vascular defects, although to a substantially lesser degree than those with SCD (Van Eps et al, 1970). Hypoxia within the renal medulla predisposes to sickling: the sickled red blood cell-induced blood flow alterations in the renal vasa recta, which supply the renal tubules, result in changes in the counter current concentration mechanism in the renal medulla. A concentrating defect is common in patients with SCT and is present at an early age although it is usually not as severe as observed with SCD (Francis & Worthen, 1968). The severity of the concentrating defect is directly and positively related to the percentage of Hb S (Gupta et al, 1991). This would suggest that individuals with SCT are part of a heterogenous population with a potential severity of renal dysfunction related to the percent alteration in their haemoglobin composition.

The presence of a concentrating defect complicates the renal maintenance of a stable total body haemodynamic response to situations of induced volume depletion. As volume depletion is, in itself, a risk factor for AKI and may contribute to sickling events, awareness of the presence of SCT is important to minimize the risks of volume depletion in those individuals. The role of volume depletion in the development of rhabdomyolysis may partly explain the increased risk seen with SCT carriers in this active population. This has direct relevance to a population who train and provide mission-related tasks in a variety of external environments.

With regard to other possible mechanisms, more recent studies of individuals with SCD have shown haemolysis-endothelial dysfunction and ischaemic localization to the kidney, which can induce capillary congestion and hyperfiltration (Nath & Katusic, 2012). Hyperfiltration is associated with the development of proteinuria, hypertension and the histological diagnosis of glomerulosclerosis (Ataga & Orringer, 2000). Endothelial dysfunction has been described in SCD and probably also plays an important role in the pathogenesis of sickle cell kidney disease (Naik & Derebail, 2017). Because the prevalence of albuminuria (macro and micro) in patients with SCT is clearly increased over non-carriers, it is reasonable to conclude that the underlying renal microvascular pathophysiology observed may occur in patients with either SCD or SCT (Naik et al, 2014; Sesso et al, 1998). Ataga and Orringer (2000) also indicated that, in SCT patients, spontaneous sickling during hypoxia can contribute to the development of haematuria, hyposthenuria and, in more severe cases, renal papillary necrosis through disruption of adequate perfusion in the vasa recta. There is, therefore, plausible evidence for functional and structural renal damage among SCT-positive individuals based on mechanisms that would contribute to the increased risk of AKI and CKD we observed in our study.

Other health-related factors were also significantly associated with the renal outcomes. Consistent with expectations (Atkins, 2005; Bakris & Ritz, 2009; Chawla et al, 2014), soldiers with hypertension, rhabdomyolysis or renal disease of the other type were at much higher odds of both AKI and CKD. The magnitude of the effect of each of these co-morbidities was substantially greater than that of SCT. Another factor also associated with the elevated risk of both outcomes was obesity, with a similar effect size to that of SCT. Not surprisingly, we observed a dramatic association between age and CKD risk: the adjusted odds ratios were 4.08, 6.51, 11.62 and 19.24 for those in the 28–35, 36–41, 42–49 and 50 years and over age groups, respectively, compared to those in the youngest age category (p=0.007, <0.0001, <0.0001 and <0.0001, respectively).

We were concerned that the effect of SCT on CKD might be modified by the prior diagnosis of AKI. However, an SCT-by-AKI interaction term included in a sensitivity analysis was not significant. One previous study observed an increased risk for kidney disease for Dominican Hispanic SCT positive individuals (Dueker et al, 2017). Although Hispanic ethnicity was relatively rare in our study population (2.4%), we performed an additional sensitivity analysis including ethnicity (Hispanic vs. non-Hispanic) in the model. The effect sizes of SCT for both AKI and CKD remained the same, and Hispanic ethnicity itself was not significantly associated with either AKI or CKD. Considering the important influence of age with regard to CKD risk, we also tested different approaches to modelling age and calculated the predicted risk of having CKD in these diverse age categories. These different approaches to modelling age did not impact our effect size for SCT.

Several limitations in our study are worth noting. First, screening for SCT was not conducted for all African American soldiers in the Army. One might expect some differences between the tested and non-tested individuals, given that many SCT tests are likely to have been done on full units in preparation for deployment or among women in association with pregnancy. Individuals in a deployable unit may differ in some key ways; specifically, we may expect those soldiers to be somewhat older (i.e., with more follow-up time, to have the opportunity to be deployed) and, on average, somewhat healthier. We do indeed observe that tested soldiers are more likely to be in the 23–35 years old age range (vs. in the youngest or older age categories), are less likely to be obese, less likely to have either hypertension or diabetes, and more likely to be female. Being in a deployable unit, however, should not be associated with SCT itself, as regulations around deployability do not direct differential statuses on the basis of SCT, but rather only on the basis of medical conditions, for example musculoskeletal conditions and behavioural health conditions (US Army, 2008). We determined that AKI rates are very similar in those tested versus those not tested, and that CKD rates are slightly higher among tested individuals; it is very possible that some of these SCT tests occurred subsequent to the renal diagnoses, rather than prior. In summary, although some differences are observed between SCT-tested and non-SCT-tested populations, it is not clear that our results describing the association of SCT with the kidney disease outcomes would be biased by these differences.

Another concern would be the general limitations of ICD-9 codes to define the outcomes, including potential clinical inaccuracy and the restrictive coding structure. Such limitations could result in misclassification between AKI and CKD, although the similar associations observed with SCT would reduce concerns on this point. An additional diagnosis-related problem could be case under-detection (Fleet et al, 2013; Molnar et al, 2016). However, this is a lesser concern in our study population because the importance of soldiers’ medical readiness means that soldiers with renal-related complaints receive substantial attention (US Army, 2008). Therefore, there is a low threshold for referring soldiers with possible renal conditions to specialists who would then define the severity and chronicity of these conditions.

It also may be the case that average ancestry proportions are somewhat different comparing African Americans with SCT to those without. Because we do not have genetic data in our database, we are unable to control for population substructure. However, the magnitude of the SCT effect we observed in this study is approximately as large as the entire difference between African American soldiers and white soldiers in our dataset (Nelson et al., forthcoming), so a small shift in ancestry proportions would not be enough to drive the entire observed effect.

Finally, although our dataset has a number of strengths, including a large number of individuals with documented SCT status linked with longitudinal health histories, because this is an observational design, we cannot infer causality.

Conclusion

In conclusion, our study identified significant increases in the risk of both AKI and CKD in association with SCT among African American U.S. Army soldiers. Clinicians should be aware that it may be important to assess risks and monitor for early signs of renal dysfunction in this population. Such screening could occur as a part of health screenings or based on history and physical examination findings during routine outpatient care. Although further investigation would be important, we expect that our results may generalize to other comparably active populations. In support of this contention, our results confirm, in our selected healthy population, the previously reported elevated risks of both AKI and CKD in association with obesity, hypertension and prior AKI events. Further research may be warranted to determine the particular genetic or biological properties that put SCT-positive individuals at increased risk of renal outcomes.