Newborn screening for sickle cell disease began in the United States in the early 1970s. These initial screening programs grew out of the recognition that sickle cell anemia was associated with significant morbidity and mortality. In 1970, the estimated median expected survival was 20 years for affected persons living in the United States (Scott, 1970). With advances in the diagnosis, treatment, and prevention of

complications, the life expectancy of persons with sickle cell disease has improved. Presently there is an 85 percent chance that infants with Hb SS will survive to age 20 and a 92 percent change that babies born with Hb SC will survive to a similar age, (Leikin, Gallagher, and Kinney, 1989). The principal causes of death in infants with Hb SS disease are overwhelming infections with Streptococcus pneumoniae organisms, cerebral vascular accidents, and acute splenic sequestration crisis (Emond, Collis, Darvill, et al., 1985; Leikin, Gallagher, and Kinney, 1989). The predisposition to pneumococcal infection is secondary to the functional asplenia that develops within the first 2 years of life (Pearson, Gallagher, Chilcote, et al., 1985). Twice-daily oral penicillin reduces both the morbidity and mortality from pneumococcal infection (Gaston, Verter, Woods, et al., 1986). Teaching parents to recognize the early signs of splenic sequestration crisis should reduce deaths from this complication. Although laboratory procedures to detect sickle hemoglobin in newborns have been available for nearly two decades, neonatal screening for sickle cell disease was not widely implemented by State screening programs until the late 1980s (Garrick, Dembure, and Guthrie, 1973; Gilman, McFarlane, and Huisman, 1976; Schneider, Gustafson, and Haggard, 1970). Arguments against neonatal screening included the contention that little could be done to reduce either morbidity or mortality once sickle cell disease was detected. This position prevailed despite evidence that early identification and entry into comprehensive care favorably affected overall morbidity and mortality (Powars, Overturf, Weiss, et al., 1981). The position was invalidated by a randomized controlled clinical trial which demonstrated that twice-daily oral penicillin reduced both mortality and morbidity from infectious complications of sickle cell anemia (Gaston, Verter, Woods, et al., 1986).

Following the report on the beneficial effects of prophylactic penicillin, the National Institutes of Health convened a Consensus Conference to review evidence about the benefits of newborn screening for sickle cell disease. The conference panel concluded that screening could reduce morbidity and mortality, provided the screening program was linked to the provision of comprehensive health care services to affected infants (Consensus Conference, 1987).

Shortly after these findings were disseminated, sickle cell screening in the United States became widespread. The proliferation of screening programs was stimulated by funds from the Bureau of Maternal and Child Health which were earmarked to support screening programs that could demonstrate the ability of neonatal hemoglobinopathy screening to reduce

morbidity and mortality. Today, newborn hemoglobinopathy screening is performed in more than 40 States, the District of Columbia, Puerto Rico, and the Virgin Islands. The remarkable growth in screening programs within the past few years illustrates how Federal and State partnerships can dramatically improve the quality of health care programs.

Purpose of the Guideline

Although the findings of the Consensus Conference (1987) were published 6 years ago, there remain many unresolved issues surrounding neonatal sickle cell disease screening related to:

1. Definition of the essential screening program components and their respective responsibilities.

2. Definition of the population to be screened.

3. Standards for sample identification, collection, and shipment.

4. Standards for laboratory methods, quality control, quality assurance, and

result reporting.

5. Education and genetic counseling services for the parents of identified heterozygotes and infants with disease.

6. Medical care for infants with sickle cell disease.

7. Cost effectiveness of neonatal sickle cell screening.

This guideline was developed to address each of these problematic areas. The guideline emphasizes the required components for a program, defines the population to be screened, addresses important laboratory and genetic counseling issues, and describes essential health services for infants identified with sickle cell disease. The guideline also discusses the cost effectiveness of neonatal screening for sickle cell disorders.

The panel recognizes that financial constraints may prevent a particular screening program from meeting all recommendations within the guideline. This guideline, however, provides the framework for the implementation and conduct of a screening program that will achieve the ultimate objective of reducing infant morbidity and mortality from sickle cell disease.

1

Guideline:

Population To Be Screened

Introduction

One of the panel's primary objectives was to determine the prevalence of sickle cell disease in different racial and ethnic newborn populations. Other objectives included determining the prevalence of other hemoglobinopathies that might be detected during screening. These data can be used to estimate the total cost and benefit of screening programs. Issues for future research are discussed at the end of this chapter.

All newborns should be screened for sickle cell disease by accurate laboratory techniques. The purpose of such screening is to reduce morbidity and mortality from sickle cell disease. Screening also can identify infants with sickle cell trait, as well as homozygotes and heterozygotes for other hemoglobin variants. Screening of populations with a low prevalence of Hb S is cost-effective when the screening is integrated into a laboratory that is also testing samples from a population with a high prevalence of Hb S.

This recommendation is based on the following analysis of various racial and ethnic populations and the reduction of morbidity and mortality from sickle cell disease by early identification, when coupled with comprehensive medical care and utilization of penicillin prophylaxis.

Importance of Population Characteristics

To evaluate the effectiveness of a screening program, several factors must be assessed (Frame, 1986). These include the incidence (or in the case of genetic abnormalities, prevalence) of the condition to be identified by the screening test. Sickle cell disease is more prevalent in some racial groups than in others. If prevalence differs between identifiable subgroups within the population, then it must be determined if the most cost-effective approach is to screen the entire population or only specific subpopulations.

Difficulties in Assessing Characteristics

Although many studies on the prevalence of sickle cell disease have been published, it is difficult to define accurately the probability that a specific newborn will have this condition. The U.S. population is very heterogeneous and is constantly changing, making it difficult to define with certainty the race of a specific individual. Race cannot be assigned solely by phenotypical characteristics or surnames. Similarly, race cannot be

accurately defined by self-report or by the race of the parents, as these may differ. The lack of uniform standards for recordkeeping of racial or ethnic statistics may contribute to significant errors in the published reports of the prevalence of sickle cell disease in specific racial or ethnic groups.

Assumptions

Because of these difficulties, the guideline makes several assumptions. 1. The same method of assessing race and ethnicity that was used in the studies cited below will be used in the future to implement the guideline if such identification is required.

2. Data in the cited studies represent the U.S. population as a whole, and the studies represent independent samples from that population. 3. There often is significant heterogeneity within an ethnic group. For example, Hispanics (or those with Hispanic surnames), include European Spaniards, as well as persons from the Caribbean Islands, Mexico, and Central and South America. This guideline utilizes the Census Bureau's methods for defining specific ethnic groups.

Methods

A panel subcommittee was formed to develop prevalence estimates. The data used for the prevalence estimates cited in this guideline were derived primarily from a literature search described elsewhere in this guideline. The search produced 1,111 references dealing with population characteristics. After multiple reviews of the title and abstract lists, citations were selected for retrieval and data extraction. Articles were reviewed several times, and data were extracted using a standardized format. The panel's methodologist and subcommittee chair reviewed each article and the extracted data for completeness and accuracy. Twenty-two citations were used in this analysis.

Articles with data on the prevalence of hemoglobinopathies in U.S. population groups were categorized as follows.

■ Group 1-Articles with data on the prevalence of hemoglobinopathies among specific ethnic groups.

■ Group 2-Articles with data by geographic area and information about the ethnic/racial population studied but without prevalence data on specific ethnic/racial groups.

■ Group 3-Articles with data by geographic area alone, with no ethnic or racial data.

Articles that had no data, only had data about non-U.S. populations, or were judged by the reviewers to have serious flaws in methodology were not included. Following categorization, Group I contained 11 articles (Broghamer Jr, Lockwood, and Keeling 1981; Carr and Chapatwala, 1988; Diaz-Barrios, 1989; Gardner and Keitt, 1988; Grover, Wethers, Shahidi, et

al., 1978; Huisman, Harris, Stewart, et al., 1991; Mack, 1989; Meany and Riggle, 1992; Pass, Gauvreau, Schedlbauer, et al., 1986; Powars, 1989; Ralston, Kmetz, Keeling, et al., 1981; Therrell, Simmank, and Wilborn, 1989) and data from the Council of Regional Networks for Genetic Services (CORN). Three articles were included in Group II (AnyaneYeboa, 1989; Harris and Eckman, 1989; Vichinsky, Hurst, Earles, et al., 1988). Seven articles (Barnes, Komarmy, and Novack, 1972; Castro, Winter, Lee, et al., 1981; Foster, Forbes, Hayes, et al., 1981; Lobel, Cameron, Johnson, et al., 1989; Pearson, 1989; Schedlbauer and Pass, 1989; Wethers and Grover, 1986) and additional data from CORN (not broken down by ethnic group) were included in Group III. Only Group I data were used to estimate prevalences.

Data from the articles in each group were collected in separate data bases on IBM compatible microcomputers; dBase IV software was used to build, maintain, and query these three data bases. The software was also used to compute prevalence rates at birth in those cases where the published data only contained raw numbers.

Data were displayed in the form of printed tables for review by the subcommittee. Data in two classes were eliminated from further consideration: (1) data that were clearly incomplete (that is, indicated prevalence rates of trait that were near or below disease rates, or indicated disease rates that were at least an order of magnitude higher than otherwise indicated for the population studied) and (2) data that were superseded by newer, more complete data from the same or a related source.

In some cases, the subcommittee contacted authors to determine whether additional data were available to fill in missing data in the published literature. Additional unpublished data were received from Dr. Astrid Mack' to supplement the published data on Florida prevalences.

A key source of data was the CORN 1990 Newborn Screening Report (Meany and Riggle, 1992), which contained data from many State screening programs. The CORN data represent the most recently published data but do not include a separate category for Hispanics. These data were combined separately from the other literature as well as with the rest of the literature.

The data were analyzed with Bayesian meta-analysis, using the FAST*PRO program (Eddy and Hasselblad, 1992) (with Jeffrey's noninformative priors). Meta-analysis was used to help estimate the prevalence of sickling diseases for each ethnic and/or racial group for which data were available.

'Mack, A., University of Miami, personal communication, 1991.