Professional Documents
Culture Documents
ation between birth weight and CH. There is a large collected on this form before testing for CH. All infor-
excess of low birth weight among infants with most mation for this study came from the computerized
types of malformations. Since the 1950s, studies have records of these collection forms maintained by the
observed that infants with CH are more likely to have California Newborn Screening Program.
increased birth weights compared with unaffected in- The collection form had checkboxes for the following
fants (Childs et al., ’54; Andersen, ’61; Maenpaa, ’72; categories of infant’s ethnicity: white, Hispanic, black,
Smith et al., ’75; Grant et al., ’92; Law et al., ’98). Based Chinese, Japanese, Korean, Cambodian, Laotian, Viet-
on these observations, we hypothesized a U-shaped namese, Filipino, Native American, Middle Eastern,
relationship between birth weight and the prevalence Asian Indian, Hawaiian, Guamanian, Samoan, and
of CH. We have also examined the birth weight and CH other. Data from the collection form indicated that 14%
association stratified by infant’s sex and ethnicity, of the infants included in this study belonged to two or
which past studies have not done. more ethnic groups. Almost all of these infants belonged
In addition, we sought to characterize the relation- to only two ethnic groups, one of which was white. We
ship between ethnicity and CH more accurately by assigned these infants to their nonwhite ethnic group.
examining the prevalence of CH for 13 different ethnic Instances of overlap between nonwhite ethnic groups
groups, most of which have not been examined previ- (e.g., Chinese and black or Chinese and Korean) were
ously. We also sought to determine if there were differ- very rare and were handled by giving preference to the
ences in the prevalence of CH by maternal age or smaller ethnic group. All infants indicated to be Native
whether the prevalence of CH was changing between American or Hawaiian were assigned to those ethnic
1990 and 1998. groups, irrespective of the presence of other nonwhite
ethnic groups. Infants indicated to be both black and
MATERIALS AND METHODS Hispanic were assigned to a separate group. The sam-
ple size for Guamanians, Samoans, Cambodians, and
From January 1, 1990 to December 31, 1998, a total Laotians was too small for meaningful analysis. There-
of 5,049,185 infants, or greater than 98.6% of the total fore, infants from these ethnic groups were included
live births in the state, were screened for congenital with infants from other unspecified ethnic groups and
hypothyroidism by the California Newborn Screening infants whose ethnicity was unknown.
Program of the California Department of Health Ser- The mother’s age at delivery was calculated from the
vices. time elapsed between the mother’s date of birth and
The case definition for this study included only con- the infant’s date of birth. Birth weights of ⬍250 g or
firmed cases of primary CH. In primary CH, the defect ⬎7,500 g were considered improbable and were reas-
in the production of thyroxine occurs at the level of the signed as missing.
thyroid gland. Approximately 85–90% of these cases From January 1990 through October 1997, heel-stick
are due to structural malformations of the thyroid samples of dried blood spots from all screened infants
gland (Sobel and Saenger, ’89; Virtanen et al., ’89). The were assayed for thyroxine (T4). Values of ⱕ9.0 mg/dl
remaining cases of primary CH are hereditary and were considered presumptively positive. In addition,
attributed to autosomal recessive mutations affecting the lowest 5% of the values in each tray of samples
the synthesis of thyroxine. Similar to previous studies were also considered presumptively positive. Presump-
of risk factors for CH, this study did not have the tive positives were assayed for thyroid-stimulating hor-
necessary information to determine the etiology of the mone (TSH). Values of ⱖ25 U/ml were considered
primary CH. Screening programs are designed to de- definitely positive. Infants with positive results were
tect primary CH, but not the underlying cause or eti- referred to a pediatric endocrinologist for further diag-
ology. Because testing to determine the underlying eti- nostic testing. In November 1997, The California New-
ology is complex and does not alter management, some born Screening Program implemented new CH screen-
clinicians consider it optional (LaFranchi, ’99). ing procedures. Since that time, all infant blood
Although the California Newborn Screening Pro- samples have been assayed initially for TSH blood lev-
gram is designed to detect primary CH, some cases of els and values of 25–100 U/ml are considered pre-
secondary and tertiary CH are also detected. These sumptively positive. These infants receive a second
cases are rare and have conditions that indirectly affect assay; if the TSH is still ⬎25 U/ml, they are consid-
the production of thyroxine through effects on the pro- ered definitely positive. Initial assays of TSH above 100
duction of hormones in the pituitary (secondary CH) U/ml are considered definitively positive.
and the hypothalamus (tertiary CH) (Sobel and Prevalences were compared using prevalence ratios
Saenger, ’89; De Vijlder et al., ’97). A total of 74 sec- rather than prevalence odds ratios. Strata frequencies,
ondary and tertiary cases was ascertained by the Cal- relative risks, confidence intervals, and logistic regres-
ifornia Newborn Screening Program over the study sion were calculated using the SAS data analysis pro-
period and is excluded from our analysis. gram.
A heel-stick blood sample spotted onto a preprinted
collection form was used for screening. Infant’s ethnic- RESULTS
ity and gender, mother’s date of birth, infant’s date of The distribution of the infant’s ethnicity in our study
birth, date of screening, and infant’s birth weight were population was 45.6% Hispanic, 33.0% white, 7.5%
38 WALLER ET AL.
black, and 13.9% other ethnic groups combined. Over with a female to male prevalence ratio of 2.4. Whites
the study period of 1990 –1998, the proportion of all and Filipinos also had a clear preponderance of female
newborns that were Hispanic increased steadily from cases, with prevalence ratios of 2.0 and 2.3, respec-
40.9% in 1990 to 49.9% in 1998. The proportion of tively. Most of the remaining ethnic groups showed a
infants from all other ethnic groups decreased slightly lower preponderance of female cases, although small
over the same period. numbers make these less precise and more likely due to
The prevalence of primary CH for the total study chance. Among black, Vietnamese, and Asian Indian
population was 35.8 per 100,000 live births (1 per infants, the prevalence ratio in females compared with
2,793) and varied according to the infant’s ethnicity males was less than or equal to 1.0. However, because
(Table 1). Compared with whites, most of the ethnic of the small numbers, the 95% CI values suggest that
groups we examined had elevated prevalence rates,
these estimates are not stable.
including Hispanics, Chinese, Filipinos, Vietnamese,
Table 2 gives the prevalence of CH separately by
Middle Easterners, Asian Indians, and Hawaiians. The
year of birth from 1990 through 1998. Although not
increase we observed for Filipinos was not significant
statistically. However, the 95% confidence interval statistically significant (P ⫽ 0.18), there is a slight
(95% CI) indicates that that an elevation in this group increase in prevalence for the total population over the
is very likely. Prevalence rates for white and Hispanic study period. However, when this secular trend was
infants were the most stable, and there was a 1.6-fold examined separately for four major ethnic groups, eval-
increased prevalence of CH among Hispanic infants uation for linear trend by logistic regression showed no
compared with white infants (prevalence ratio 1.6, 95% significant trend for any of these ethnic groups.
CI ⫽ 1.4 –1.8). Asian Indians had particularly high Table 3 shows the prevalence of CH separately for
rates of CH compared with whites (prevalence ratio different categories of birth weight. Since, on average,
2.9, 95% CI ⫽ 1.9 – 4.4). Black infants had the lowest male infants are larger than female infants (Alexander
CH prevalence, at 8.7 per 100,000 births. This was et al., ’96), the data are stratified by infant sex. Both
one-third the CH prevalence of white infants (preva- low- and high-birth-weight infants had an increased
lence ratio 0.3, 95% CI ⫽ 0.2– 0.4). Infants who were prevalence of CH (Table 3, Fig. 1). Infants weighing
Japanese or Korean and those indicated to be both ⱕ2,000 g had a twofold or greater increase in the risk
black and Hispanic had prevalence rates that were of CH compared with the referent infants who weighed
similar to, or slightly less than, those of whites. How- 3,000 –3,499 g. We examined prevalence ratios for
ever, the rates for these groups are based on small three weight categories of infants weighing ⱖ3,500 g
sample sizes and the 95% CI values suggest that they and found a modest trend of increasing risk of CH with
are not stable. increasing birth weight, especially among female in-
Table 1 gives the ratio of prevalence rates for females fants. For infants weighing ⱖ4,500 g, there was a 2.4-
compared with males separately, on the basis of infant fold increase in the risk of CH among females infants
ethnicity. The female-to-male prevalence ratio was 2.1 and a 2.3-fold increase in the risk of CH among males
overall and varied considerably by infant ethnicity. infants compared with the sex-specific referent groups
Hispanics had the highest excess of female cases of CH of 3,000 –3,499 g (Table 3).
RISK FACTORS FOR CONGENITAL HYPOTHYROIDISM 39
TABLE 2. Prevalence of primary congenital hypothyroidism by infant’s race and year of birth, 1990 –1998,
California per 100,000 live births†
Infant’s
race 1990 1991 1992 1993 1994 1995 1996 1997 1998 P-value*
White 26.1 31.4 28.0 25.7 26.3 36.8 30.2 28.8 22.9 0.93
Black 8.9a 11.0a 13.3a 6.7a 11.5a 9.8a 5.1a 10.6a —a 0.18
Hispanic 41.7 42.9 43.0 49.6 46.2 45.3 42.0 43.3 52.8 0.24
Other 35.2 44.6 39.4 35.5 33.6 30.1 33.1 51.2 37.9 0.84
Total: 32.3 36.5 35.1 36.5 35.2 37.5 34.5 37.8 37.9 0.18
†
Observations missing due to missing variables: 13,618 (0.3% of total births), no cases missing.
a
Less than 10 cases.
*Tested for linear trend using logistic regression.
TABLE 3. Primary congenital hypothyroidism: 1990 –1998 prevalence by infant’s sex and birth weight per
100,000 live births*
Females Males
Prevalence Prevalence
Weight (g) Prevalence Cases ratio 95% CI Prevalence Cases ratio 95% CI
⬍2,000 83.8 47 2.0 1.5–2.7 51.0 29 2.4 1.6–3.5
2,000–2,499 50.2 51 1.2 0.9–1.6 23.4 21 1.1 0.7–1.7
2,500–2,999 41.8 178 1.0 0.8–1.2 19.5 66 0.9 0.7–1.2
3,000–3,499a 41.9 404 1.0 a
21.3 189 1.0 a
DISCUSSION
Fig. 1. Primary congenital hypothyroidism prevalence by gender and We observed a U-shaped association between the
birth weight in 1,786 cases. California, 1990 –1998. prevalence of CH and birth weight that has not been
previously described. It is well known that infants with
The distribution of birth weight varies by ethnic congenital malformations have a large excess of low
group with blacks tending to be smaller than whites birth weight due to growth retardation that often ac-
and Hispanics. To determine whether the U-shaped companies the malformation. Thus, it was not surpris-
relationship observed between prevalence of CH and ing to observe an excess of low birth weight among
birth weight was confounded by infant’s ethnicity we infants with CH. However, except for CH and rare
examined the prevalence ratios for these birth weight genetic syndromes in which infants are born with mac-
categories separately for white, Hispanic, and black rosomia (Jones, ’97), previous studies have not noted
infants (results not shown). In each of these ethnic an association between macrosomia and congenital
groups, there was an increased prevalence of CH both malformations (Milli et al., ’91).
among infants weighing ⱕ2,000 g and among infants Although, a female excess has been reported for neu-
weighing ⱖ4,500 g, compared with the referent weight ral tube defects (NTDs) and some types of cardiac mal-
group. We also noted that the ratio of female to male formations (Seller, ’87; Pradat, ’92; Samanek, ’94;
cases did not vary across these birth weight categories. Shaw et al., ’94), these excesses are neither as consis-
40 WALLER ET AL.
tent nor as large as the female excess associated with (Maenpaa, ’72; Smith et al., ’75; Grant et al., ’92).
CH. Thus, the female excess is a particularly interest- Larger and more refined analyses are needed in order
ing feature of the epidemiology of primary CH. It is to determine whether the association between macro-
unclear, however, whether females are more suscepti- somia and CH is independent of gestational age. That
ble to developing CH or whether females with CH have is, do infants with CH tend to be large for their gesta-
increased in utero survival compared with males with tional age?
CH. The observation of thyroxine in the cord serum of Primary CH is a relatively common condition with
fetuses with CH suggests that levels of fetal thyroxine almost complete ascertainment in developed countries.
are maintained by transfer of maternal thyroxine to Thus, an excellent sampling frame exists to conduct
the fetus and that CH may not be deleterious to the population-based studies of these infants. To date,
fetus before birth (Vulsma et al., ’89). Also, Dussault et
most studies of risk factors for CH have been based on
al. (’80) reported that the ratio of female to male cases
existing databases and did not distinguish between the
among hereditary cases of CH was approximately 1.0.
different subtypes of primary CH. Future studies of
Among the 74 cases of secondary and tertiary CH that
we excluded from this study, the ratio of females to primary CH should attempt to distinguish between
males was 0.8 (33 females to 41 males), compared with hereditary cases and malformations of the thyroid,
the ratio of 2.0 (1,210 females to 596 males) observed which can be further classified into dysgenesis and
among our cases of primary CH, P ⬍0.001. These ob- agenesis of the thyroid. It appears that the female
servations suggest that fetal hypothyroidism alone excess of primary CH probably varies across these sub-
does not cause differential survival of females com- types. An analysis by sub-types might also shed light
pared with males and that the preponderance of female on the epidemiology of CH among black infants.
cases may be confined to the 85–90% CH which are due The prevention of mental retardation due to CH,
to malformations of the thyroid gland. It also follows through the use of thyroxine replacement therapy, is
that there may be a true increase in the number of an important public health success story. However,
female infants in whom malformations of the thyroid little work has been done to identify modifiable risk
gland develop. factors with the potential to prevent CH. Knowledge of
Our study is consistent with previous studies in ob- such risk factors could decrease the number of infants
serving that Hispanics and Asians have an increased who need lifetime thyroid hormone replacement ther-
and blacks a decreased prevalence of CH compared apy.
with whites (Brown et al., ’81; Sobel and Saenger, ’89;
Lorey et al., ’92; Toublanc, ’92; Ray et al., ’97; Roberts
et al., ’97; Seeherunvong and Sunchai, ’98). The epide- LITERATURE CITED
miology of CH among black infants is strikingly differ- Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. 1996. A
ent from that of other ethnic groups. They have a very United States national reference for fetal growth. Obstet Gynecol
87:163–168.
low prevalence of CH and an equal ratio of the females
Andersen HJ. 1961. Studies of hypothyroidism in children. Acta Pae-
to males. Although our observation of a female to male diatr 50[suppl 125]:38 – 47.
ratio of 1.0 for cases of CH in blacks is not statistically Becerra JE, Khoury MJ, Cordero JF, Erickson DJ. 1995. Diabetes
significant, it is consistent with a similar finding re- mellitus during pregnancy and the risks for specific birth defects: a
ported by Lorey and Cunningham (’92). population based case-control study. Pediatrics 85:1–9.
Brown AL, Fernhoff PM, Milner J, McEwen C, Elsas LS. 1981. Racial
The increased prevalence of CH that we observed differences in the incidence of congenital hypothyroidism. J Pediatr
among infants with macrosomia and infants from spe- 99:934 –936.
cific ethnic groups might be explained by an increased Childs B, Gardner LI. 1954. Etiologic factors in sporadic cretinism. An
prevalence of CH among mothers with diabetes. analysis of ninety cases. Ann Hum Genet 19:90 –96.
De Vijlder JJM, Ris-Stalpers C, Vulsma T. 1997. Inborn errors of
Women with insulin-dependent diabetes (IDDM) and
thyroid hormone biosynthesis. Exp Clin Endocrinol Diabetes
gestational diabetes are at greater risk of having mac- 105[suppl 4]:32–37.
rosomic infants and of having infants with a variety of Dussault JH, Mitchell ML, LaFranchi S, Murphey WH. 1980. Re-
congenital anomalies (Ramos-Arroyo et al., ’92; Becerra gional screening for congenital hypothyroidism: results of screening
et al., ’95). Compared with white women, increased one million North American infants with filter paper spot T4-TSH.
In: Burrow GN, Dussault JH, editors. Neonatal thyroid screening.
diabetes has been reported among pregnant women in New York: Raven Press. p 155–165.
Hispanics, Native Americans, Chinese Americans, Fisher DA, Dussault JH, Foley TP, Klein AH, LaFranchi S, Larsen
Asian Indians, and Middle Easterners (Kieffer, ’98). PR, Mitchell ML, Murphey WH, Walfish PG. 1979. Screening for
We also observed an increased prevalence of CH among congenital hypothyroidism: results of screening one million North
American infants. J Pediatr 94:700 –705.
infants belonging to these same ethnic groups. How-
Grant DB, Smith I, Fuggle PW, et al. 1992. Congenital hypothyroid-
ever, we could not directly test the hypothesis of an ism detected by neonatal screening: relationship between biochem-
increased risk of CH associated with maternal diabetes ical severity and early clinical features. Arch Dis Child 67:87–90.
because our study data do not include information on Jones KL. 1997. Smith’s recognizable patterns of human malforma-
diabetes in the mothers. tion. Philadelphia: WB Saunders. p 154, 158, 164, 168.
Kieffer EC. 1998. Diabetes during pregnancy, United States 1993–
Some investigators have suggested that infants with 1995. MMWR 47:408 – 414.
CH have a tendency to prolonged gestation, although LaFranchi S. 1999. Congenital hypothyroidism: etiologies, diagnosis,
only a few have presented data on gestational age and management. Thyroid 9:735–740.
RISK FACTORS FOR CONGENITAL HYPOTHYROIDISM 41
Law WY, Bradley DM, Lazarus JH, John R, Gregory JW. 1998. Seeherunvong T, Sunchai C. 1998. Etiologic study of primary congen-
Congenital hypothyroidism in Wales (1982–1993): demographic fea- ital hypothyroidism. Med Assoc Thai 81:653– 657.
tures, clinical presentation and effects on early neurodevelopment. Seller MJ. 1987. Neural tube defects and sex ratios. Am J Med Genet
Clin Endocrinol 48:201–207. 26:699 –707.
Lorey FW, Cunningham GC. 1992. Birth prevalence of primary con- Shaw GM, Jensvold JG, Wasserman CR, Lammer EJ. 1994. Epide-
genital hypothyroidism by sex and ethnicity. Hum Biol 64:531–538. miologic characteristics of phenotypically distinct neural tube de-
Maenpaa J. 1972. Congenital hypothyroidism. Aetiological and clini- fects among 0.7 million California births, 1983– 87. Teratology 49:
cal aspects. Arch Dis Child 47:914 –923. 143–149.
Milli F. Edmonds LD, Khoury MJ, McClean AB. 1991. Prevalence of
Smith DW, Blizzard RM, Wilkins L. 1957. The mental prognosis in
birth defects among low birth weight infants. Am J Dis Child
hypothyroidism of infancy in childhood. Pediatrics 19:1011–1022.
145:1313–1318.
Smith DW, Klein AM, Hederson JR, Myrianthopoulos NC. 1975.
Pradat P. 1992. Epidemiology of major congenital heart defects in
Sweden, 1981–1986. J Epidemiol Community Health 46:211–215. Congenital hypothyroidism—signs and symptoms in the newborn
Ramos-Arroyo MA, Rodriquez-Pinilla E, Cordero JF. 1992. Maternal period. J Pediatr 87:958 –962.
diabetes: the risk for specific birth defects. Int J Epidemiol 8:503– Sobel EH, Saenger P. 1989. Hypothyroidism in the newborn. Pediatr
508. Rev 11:15–20.
Ray M, Muir TM, Murray GD, Kennedy R, Girdwood RWA, Donaldson Toublanc JE. 1992. Comparison of epidemiological data on congenital
MDC. 1997. Audit of screening programme for congenital hypothy- hypothyroidism in Europe with those of other parts in the World.
roidism in Scotland 1979 –93. Arch Dis Child 76:411– 415. Horm Res 38:230 –235.
Roberts HE, Moore CA, Fernfhoff PM, Brown AL, Khoury MJ. 1997. Virtanen M, Maenpaa J, Pikkarainen J, Pitkanen L, Perheentupa J.
Population study of congenital hypothyroidism and associated birth 1989. Aetiology of congenital hypothyroidism in Finland. Acta Pae-
defects, Atlanta, 1979 –1992. Am J Med Genet 71:29 –32. diatr Scand 78:67–73.
Samanek M. 1994. Boy:girl ratio in children born with different forms Vulsma T, Gons MH, de Vijlder JJM. 1989. Maternal-fetal transfer of
of cardiac malformation: a population-based study. Pediatr Cardiol thyroxine in congenital hypothyroidism due to a total organification
15:53–57. defect or thyroid agenesis. N Engl J Med 321:13–16.