Cohort Tracking of Sickle Cell Trait–Positive Births Identified by Newborn Screening, California, 1991-2013: Public Health Surveillance for Sickle Cell Trait

Abstract

Objectives:

Sickle cell trait (SCT) is the carrier status for sickle cell disease, and people with SCT have both hemoglobin A (HbA) and sickling hemoglobin (HbS). SCT is generally regarded as a benign condition, but clinical complications can be substantial. No registry or surveillance system exists to track health outcomes for people with SCT; this study aimed to identify methodology for surveillance.

Methods:

This longitudinal analysis included all live births with SCT identified by the California Department of Public Health Newborn Screening (NBS) Program from 1991 through 2013 and 3 matched controls per newborn, linked to death data in California for 1991-2013.

Results:

There were 94 240 live births with SCT and 282 720 matched healthy controls; 693 (0.74%) deaths occurred in the SCT group, and 1910 (0.68%) deaths occurred among the matched controls. Those with SCT had an increased mortality hazard ratio (MHR) compared with matched controls (11% higher; P = .02). When stratified, the MHR was higher among those aged 1 to 4 years (44% higher; P < .001) and 5 to 14 years (48% higher; P = .005) than among the matched controls. Examination of causes of death showed only a slightly higher-than-expected risk of death due to respiratory causes among people with SCT.

Conclusions:

These findings highlight the need for population-level research, including investigation into causes of death, to inform clinical management and counseling for SCT. Other states may replicate this methodology with population-based data sources. Further surveillance of the health of those with SCT is needed.

Sickle cell trait (SCT) is an inherited blood disorder in which an individual makes both hemoglobin A (HbA), or normal hemoglobin, and sickling hemoglobin (HbS). SCT is the carrier status for sickle cell disease (SCD). SCT was once viewed by the medical community as a generally beneficial condition because of its associated reduced risk for malarial infection.^1,2 SCT’s more recent importance has been in preconception or prenatal screening to test for partner carrier status in the prevention of SCD (HbS/HbS or other sickling conditions). Positive carrier status has otherwise been described as a benign condition, and most carriers are unaware that they have SCT.^3-7

However, among those with SCT, red cells can sickle and cause clinical and potentially fatal complications when exposed to unphysiological conditions such as severe hypoxia, acidosis, and dehydration.^2,8 Beginning in the late 1980s, researchers investigated reports of sudden exertional-related death and/or rhabdomyolysis among young athletes with SCT.^9-16

More recently, other health concerns have been reported in the literature. ¹⁷ It has been established that those who carry SCT have a higher risk than the general population of renal medullary carcinoma, hematuria, chronic kidney disease, eye complications, adverse pregnancy outcomes, and venous and pulmonary thromboembolism.^13,18-26 While it is rare, those with SCT may also experience splenic complications at high altitude or in low-oxygen environments and severe traumatic hyphema.^27,28 Other complications potentially linked to SCT include coronary artery disease, type 2 diabetes, and hemorrhagic stroke; however, findings are inconclusive.^17,18,27,29

The lifespan of people with SCT is believed to be normal, but data are lacking. An investigation of mortality risks among those with SCT in a systematic review showed no evidence of increased risk of mortality among small adult populations; however, the findings were based on only 5 studies, all rated as low or very low quality by the review authors. ¹⁸ An additional small study of short-term stroke outcomes among a clinical population found an odds ratio (95% CI) of 1.3 (1.1-3.5) for increased 30-day mortality after a hemorrhagic stroke event among those with SCT compared with those with normal hemoglobin. ³⁰ A review of mortality among US Army soldiers found no increase in all-cause mortality among those with SCT but did find substantial differences in death due to exertional rhabdomyolysis among the 14 soldiers with SCT who died compared with those without SCT who died. ¹³

Given these findings and that most people with SCT and their health care providers are unlikely to be aware of their status, conducting large-scale studies or surveillance can help determine the incidence and prevalence of comorbidities that may be associated with SCT. ⁴ The likelihood of death due to SCT may be exaggerated in some cases, as noted in a recent article on deaths of Black people in custody.^31,32 Furthermore, while concern for populations such as student athletes with SCT has grown, a recent study found that in the absence of data on risk, not enough is being done to protect these athletes from the threat of severe illness or death. ³³ The 2020 National Academies of Sciences, Engineering, and Medicine’s Addressing Sickle Cell Disease: A Strategic Plan and Blueprint for Action, recommends increased surveillance and research into the effects of SCT. ³⁴ All 50 US states now have universal newborn screening (NBS) programs for SCD, which by necessity also confirms SCT status. ³⁵ If such programs hold and have the capacity to link data on confirmed carriers of SCT, these state NBS datasets may form a foundation for surveillance to determine the true incidence of illness or mortality due to SCT.

Long a pioneer in NBS to prevent complications and death from heritable conditions, California began assessing all newborns for SCD and SCT in mid-1991. An analysis of California data through 2012 found an incidence of SCT of approximately 6.9% of Black live births in California (7.3% in the United States) and 0.6% of Hispanic live births in California (0.7% in the United States). ³⁶ The objectives of the current analysis of existing California NBS data linked to vital records death data were to (1) show that population-based surveillance of those with SCT is feasible and useful and (2) examine the risk of mortality in the confirmed NBS-identified SCT carrier population compared with matched control newborns with normal hemoglobin.

Methods

The California Department of Public Health’s (CDPH’s) NBS Program uses high-performance liquid chromatography to assess dried blood spots for hemoglobin disorders. CDPH conducts universal testing for alpha and beta mutations with ratio computations of specific high-performance liquid chromatography retention times to allow the presumptive diagnosis of unusual or rare mutations. Newborns with SCT are identified with a blood spot pattern that contains fetal hemoglobin (HbF), HbA, and HbS. Standardized thresholds for hemoglobin mutations have been validated, and atypical results are sent for confirmatory DNA testing.^37,38 Families are referred to certified SCT counselors for follow-up education and additional testing as needed. The California Biobank Program in CDPH’s Genetic Disease Branch performed a linkage of live births with SCT identified from 1991, the first year of screening for SCT, through 2013, the most recent year of completed linkage, and all deaths in California for 1991-2013, using multiple statewide death data files. There are no current plans to link data later than 2013. Probabilistic linkage using machine learning was used for this linkage; data fields were matched on mother’s first and last name and date of birth, baby’s first and last name and date of birth, mother’s street number and street name, and a hospital code.

All newborns with SCT with complete data on birth date, sex, race and ethnicity, and hospital and city of birth were included if 3 matched control people with normal hemoglobin were identified. CDPH collects data on race and ethnicity for each newborn at the time of specimen collection. Nineteen categories were used to collect data on race and ethnicity; for this study, we grouped them into 5 broad categories: Black, Hispanic, White, Asian, and Other. The Asian category includes Asian East Indian, Cambodian, Chinese, Filipino, Japanese, Korean, Laotian, Vietnamese, and Other Southeast Asian. The Other category includes Guamanian, Hawaiian, Middle Eastern, Samoan, and Other.

Matched control newborns with normal hemoglobin and complete data on birth date, sex, race and ethnicity, and hospital and city of birth were randomly selected from the NBS cohort. We matched those with normal hemoglobin 3:1 to people with SCT on year of birth, sex, race and ethnicity, and city of birth. We excluded SCT records with <3 matching control records from subsequent analyses. NBS data include only people born in California; similarly, the death data used include only those who died in California or were otherwise issued a California death certificate. Linked death certificates provided the underlying causes of death for all years that the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) coding system was used: 1999-2013. The data from 1991-1998 used the prior scheme (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM]), and crosswalks between the 2 coding schemes have been found to be flawed. As such, the team focused the analysis on underlying cause of death using the better coding scheme.

We analyzed NBS data from the Biobank database using SAS version 9.4 (SAS Institute Inc), with PROC SURVEY commands to identify matched controls. We used PROC LIFETEST to compute Kaplan–Meier curves and 95% CIs for those with SCT and controls. We calculated Cox proportional hazard ratios and 95% CIs using PROC PHREG to compare mortality hazard ratios (MHRs) between people with SCT and controls by age strata, with significance at α = .05.

In addition to determining survival, we calculated MHRs in 4 age strata for those with SCT versus matched controls: the first year of life, ages 1 to 4 years, ages 5 to 14 years, and ages 15 to 22 years. These age strata describe periods of follow-up time, so that an individual may appear in more than 1 stratum, and MHR were conditional to those surviving to the starting age of each stratum.

To examine the distribution of causes of death in the 2 groups (those with SCT carrier status and those with normal hemoglobin), an author with extensive knowledge of the pathophysiology of SCT (E.P.V.) categorized the causes of death included based on the death certificate’s information among both groups (SCT and control) into broader categories, including deaths related to accidents, infection, respiratory issues, sudden infant death syndrome (SIDS), cancer, and congenital causes. A second hematologist also categorized the causes of death, and the two met to adjudicate differences of opinion. A crosswalk for the final recategorization was developed (eTable 1 in the Supplement). Because accidents are unlikely to be related to SCT status, we used this category as a reference category to analyze the relative risk of cause of death for the other categories among those with positive SCT status compared with those with normal Hb; using R 4.5.0, we used the Fisher exact test for this purpose, and we looked at all deaths with ICD-10-CM causes of death (1999-2013) and those deaths that happened only during the neonatal phase, including sudden infant death syndrome (SIDS). We considered Fisher double-sided P < .05 to be significant, with corresponding odds ratios and 95% CIs reported for cause-of-death comparisons with significant P values. We excluded from this analysis those with causes of death listed as missing, due to SCD, or with SIDS at age ≥1 year.

The California Committee for the Protection of Human Subjects (2020-107-CDPH), the Public Health Institute Institutional Review Board (I22-012), and the California Biobank Program Data Use Committee (Study 1479) provided approval and oversight of this work. Data-sharing agreements with the state of California specify that results data with cell sizes <11 cannot be published because of the risk to the privacy of those included in the analysis. ³⁹ Therefore, some results are not shown and are noted as being “small cell size.”

Results

The California Biobank Program found 94 240 people with SCT (or 96.2% of all California newborns identified with SCT during the analysis period) who met the inclusion criteria and 282 720 matched controls born from June 1991 through December 2013 (Table 1). Nearly one-third (31.1%; n = 29 360) of the SCT-identified live births were Hispanic, and 61.0% (n = 57 496) were Black.

Table 1.

Births with sickle cell trait (SCT) identified by newborn screening in California, by race, ethnicity, and sex, 1991 through 2013 ^a

Characteristic	Overall, no. (%) (N = 92 240)	By birth year, no. (%)
		1991-1998 (7 years) (n = 33 444)	1999-2006 (7 years) (n = 32 461)	2007-2013 (6 years) (n = 28 335)
Race and ethnicity
Asian b	808 (0.9)	216 (0.6)	293 (0.9)	299 (1.1)
Black	57 496 (61.0)	21 489 (64.3)	19 211 (59.2)	16 796 (59.3)
Hispanic	29 360 (31.2)	9424 (28.2)	10 559 (32.5)	9377 (33.1)
White	5298 (5.6)	2011 (6.0)	1880 (5.8)	1407 (5.0)
Other c	1278 (1.4)	304 (0.9)	518 (1.6)	456 (1.6)
Sex
Female	46 569 (49.4)	16 556 (49.5)	16 064 (49.5)	13 949 (49.2)
Male	47 671 (50.6)	16 888 (50.5)	16 397 (50.5)	14 386 (50.8)

^aData source: California Department of Public Health Newborn Screening Program.

^bThe Asian category includes Asian East Indian, Cambodian, Chinese, Filipino, Japanese, Korean, Laotian, Vietnamese, and Other Southeast Asian.

^cOther includes Middle Eastern, Hawaiian, Guamanian, Samoan, and Other.

When linked to death data, 693 (0.74%) of those with SCT had died, while 1910 (0.68%) of the matched controls had died. Those with SCT had generally worse survival experience, with a higher MHR than matched controls (11% higher; P = .02) (Figure 1).

Figure 1.

Product-limit survival estimates among births with sickle cell trait identified by newborn screening (case) and matched controls, California, 1991-2013. Shading indicates 95% Hall–Wellner bands. Data source: California Department of Public Health Newborn Screening Program and California Biobank Program linked death data files.

When examined by age strata, those with SCT had significantly higher odds of death than matched controls in 2 age strata (Table 2, Figure 2): ages 1 to 4 years (MHR = 1.44; 95% CI, 1.16-1.78) and ages 5 to 14 years (MHR = 1.48; 95% CI, 1.13-1.93).

Table 2.

Mortality hazard ratio by age strata, sickle cell trait (SCT) cases identified by newborn screening versus controls in California, 1991 through 2013 ^a

Age, y	Group	No. in age strata follow-up time period	No. of deaths	Mortality hazard ratio (95% CI)	P value b
<1	SCT	94 240	425	1.01 (0.91-1.13)	.80
	Control	282 720	1267
1-4	SCT	89 921	123	1.44 (1.16-1.78)	<.001
	Control	269 804	261
5-14	SCT	73 942	78	1.48 (1.13-1.93)	.005
	Control	221 945	164
15-22	SCT	33 155	67	1.02 (0.78-1.34)	.88
	Control	99 545	218

^aData source: California Department of Public Health Newborn Screening Program.

^bComputed using Cox proportional hazard ratios, with significance at α = .05.

Figure 2.

Mortality hazard ratio by age strata among cases with sickle cell trait identified by newborn screening and matched controls, California, 1991-2013. Data source: California Department of Public Health Newborn Screening Program and California Biobank Program linked death data files.

The categorization and comparison of causes of death revealed that there were 463 deaths among the SCT group and 1250 deaths among the matched controls during this period (Table 3). No death certificates named hemolytic anemia in SCT case deaths. Among those aged <1 year, we found no significant differences in the frequency of death from any of the categorized causes of death compared with who died from accidents for those with positive SCT status compared with matched controls. Among the combined group (aged 0-22 y at death), compared with matched controls, we found 1 significant association: those with SCT had 1.76 (95% CI, 1.00-3.10) times the odds of dying from a respiratory condition compared with dying from an accident.

Table 3.

Categorized causes of death among sickle cell trait (SCT) cases identified by newborn screening versus controls in California, 1999 through 2013 ^a

Category	Age at death, y
	<1			1-22
	SCT	Control	P value b	SCT	Control	P value b
Accident	— c	18	(Ref.)	90	281	(Ref.)
Cancers/malignancies	— c	— c	>.99	27	65	.35
Congenital abnormalities	123	375	.36	163	454	.50
Infection	13	35	.60	21	50	.37
Respiratory	18	31	.81	26	46	.04
SIDS (aged <1 y)	28	53	>.99	28	53	—
No cause of death	48	174	—	79	237	—
Other cause of death	— c	14	.75	29	64	.18
Total	246	705	—	463	1250	—

Abbreviations: —, cell size <11; SIDS, sudden infant death syndrome.

^aData source: California Department of Public Health Newborn Screening Program.

^bUsing double-sided P values from the Fisher exact test for comparisons of SCT status (case vs control) by cause of death (variable index cause vs consistent reference of accidents), with P < .05 considered significant.

^cCell size <11, too small to report per California Department of Public Health guidelines. ³⁹

Discussion

This analysis of a cohort of 94 240 live births identified with SCT carrier status suggests that passive surveillance of populations with positive SCT status is possible using NBS data. CDPH was able to link data from those with SCT to state mortality data using extant identifiers; there is little reason to believe that other administrative datasets such as hospital discharge and health insurance claims data could not also be linked. These linkages are already taking place at the state level to track SCD via the Sickle Cell Data Collection program at the Centers for Disease Control and Prevention. ⁴⁰ Such surveillance is not as complete as a registry; infants identified as carrying SCT may move out of state or otherwise become difficult to track. However, in the absence of any long-term tracking system and consistent patient and health care provider knowledge of SCT status, a passive surveillance system is a practical and useful way to examine the health effects of SCT at the population level using existing data.

In addition to this attempt to determine the feasibility of population-level surveillance for SCT, this analysis is, to our knowledge, the first of its kind and adds information to the assessment of health risk for the condition. This study reports increased mortality among children and adolescents with SCT compared with those with normal hemoglobin. We found significant differences between the SCT and normal hemoglobin cohorts in 2 age groups: those aged 1 to 4 years at death and those aged 5 to 14 years at death. We found a slightly higher likelihood of those with SCT who died to have a listed cause of death in the ICD-10-CM category related to respiratory injury or condition compared with the normal hemoglobin group, which could be interpreted as significant at P = .04 but should be treated with caution because the sample size was small and the 95% CI included 1.00. A future study using a larger cohort would help to confirm this finding.

The paucity of prior epidemiologic studies of SCT is noteworthy in a condition so common and that is potentially associated with severe conditions such as those described previously. Goldsmith et al laid out a research agenda for SCT that called for an investigation of mortality risks to SCT and their causes. ⁵ Our study begins to fill these information gaps in the effects of SCT at the population level. Furthermore, our study suggests that other states may be able to conduct similar analyses with their SCT NBS data and that as more time passes from the onset of SCT testing, we may be able to gain a fuller understanding of the population-level effect of SCT on mortality and health if NBS SCT data are linked to data sources such as health insurance claims, hospital discharge data, or electronic health records.

The current lack of information on the population-level effect of SCT has important consequences. Clinicians may be unsure of whether to advise patients who do not know their carrier status to test for SCT or how to advise those with SCT of their risks of illness or whether to engage in intense exercise. States queried as a part of the National Academies’ Addressing Sickle Cell Disease: A Strategic Plan and Blueprint for Action did not uniformly report SCT findings from NBS; some states offer counseling for families of newborns identified with SCT, but most do not. ³⁴ These differences likely contribute to the finding that >80% of surveyed people were unaware of their SCT status. ⁴ People who know they carry SCT may be unsure of the long-term health risks of this chronic disease. In a 2010 policy decision that reflects the inadequacy of information about these risks, the National Collegiate Athletic Association implemented a mandate requiring Division I and subsequently Division II and III student athletes to identify their SCT status prior to training or play, a decision that generated controversy and prompted calls for a greater understanding of the effects of SCT on the health and risk of death of athletes and the larger population of carriers.^41-45 The need for information on the health effects of SCT is clear.

Limitations

This study had several limitations that prevent generalizability of the results. Further research into the causes of death among those with SCT compared with those with normal hemoglobin is needed. First, expanding this type of study to other states and using a longer follow-up time would provide both more information and more power for these comparisons. Such an expansion could shine a light on the pathophysiology associated with SCT that may be driving identified differences. Second, cause-of-death information drawn from death certificates is known to be flawed; with data sources such as electronic health records or hospital discharge data linked to SCT status, a more reliable review of differences in illness history and cause of death could be completed. ⁴⁶ Third, this study did not include deaths that occurred out of state or whether race and ethnicity were unknown; however, it is unlikely that there were biases in these variables between the SCT and control groups.

Conclusions

This analysis of an entire cohort of NBS-identified people with SCT is a step toward better understanding the health risks for those with SCT. Future work may include expanding this methodology to more years of data and other states or regions. Methods may also be expanded to include linking other data sources that may have more information about causes of death and outcomes other than mortality; hospital discharge records and electronic health records may be linkable to NBS and mortality data in many jurisdictions. Expanding this work would improve our knowledge of the health effects of SCT and support research into clinical implications and increasing the awareness of SCT among carriers.