Increased Prevalence of Fractures in Congenital Adrenal Hyperplasia: A Swedish Population-based National Cohort Study.

CONTEXT: Low bone mineral density has been reported in individuals with congenital adrenal hyperplasia (CAH), but the prevalence of fractures is unclear.
OBJECTIVE: To study the prevalence of fractures in CAH. DESIGN, SETTING, AND PARTICIPANTS: Patients with CAH (n = 714, all 21-hydroxylase deficiency) were compared with controls matched for sex and year and place of birth (n = 71 400). Data were derived by linking National Population-Based Registers. MAIN OUTCOME MEASURES: Number and type of fractures.
RESULTS: Mean age was 29.8 ± 18.4 years. Individuals with CAH had more fractures compared to controls [23.5% vs 16.1%, odds ratio (OR) 1.61, 95% CI 1.35-1.91], and this was found in both sexes (females: 19.6% vs 13.3%, OR 1.57, 95% CI 1.23-2.02; males: 28.7% vs 19.6%, OR 1.65, 95% CI 1.29-2.12). Fractures were significantly increased in patients born before the introduction of neonatal screening but not in those born afterwards. Any major fracture associated with osteoporosis (spine, forearm, hip, or shoulder) was increased in all individuals with CAH (9.8% vs 7.5%, OR 1.34, 95% CI 1.05-1.72). The highest prevalence of fractures was seen in SV phenotype and I172N genotype while nonclassic phenotype and I2 splice genotype did not show increased prevalence. A transport accident as a car occupant and fall on the same level were more common in patients with CAH, both sexes, than in controls.
CONCLUSIONS: Patients with CAH had an increased prevalence of both any fracture and fractures associated with osteoporosis (both sexes) but not for patients neonatally screened. We conclude that fracture risk assessment and glucocorticoid optimization should be performed regularly.

Congenital adrenal hyperplasia (CAH) belongs to a group of autosomal recessive disorders affecting the steroid synthesis in the adrenal cortex, resulting in variable cortisol and in most variants also aldosterone deficiency, together with adrenal androgen overproduction (1). 21-hydroxylase deficiency (21OHD), caused by mutations in the CYP21A2 gene, is the dominant variant of CAH affecting 95% to 99% of all cases (2,3). Phenotypically, CAH can be divided into classic, including salt-wasting (SW) and simple virilizing (SV), and nonclassic (NC) groups (1). The SW phenotype is not compatible with life if glucocorticoid replacement is not initiated within the first few weeks of life, while the SV phenotype traditionally was found in the neonatal period in girls due to ambiguous genitalia and in young boys due to rapid postnatal growth and virilization (3). However, nowadays neonatal screening of 21OHD has been introduced in many countries, and almost all classic cases are diagnosed in the neonatal period (3,4). The NC phenotype has 20% to 70% residual enzyme activity resulting in fewer symptoms and signs than the classic forms (5).

Glucocorticoid supplementation, the main treatment in CAH, was introduced in the 1950s and made survival possible in SW CAH and reduced the adrenal androgen production (2,3). However, balancing the treatment to avoid hyperandrogenism as well as hypercortisolism is challenging both for clinicians and patients. The glucocorticoid doses needed for adequate hypothalamic-pituitary-adrenal axis suppression are mostly supraphysiological, causing negative clinical outcomes (6).

CAH is associated with increased morbidity and mortality (7-14), and much of this increase is probably related to over- or underreplacement of glucocorticoids. There is a clear association between glucocorticoid use and osteoporosis in patients who receive immunosuppressive doses (15). Glucocorticoids have demonstrated direct and indirect effects on bone, causing initial increased resorption and a later decrease in bone formation with the end result of microarchitectural distortion and increased fracture risk (16). Secondary hyperparathyroidism may also occur in patients on glucocorticoids since they inhibit the calcium reabsorption in the renal tubule. Moreover, gonadal and adrenal androgens stimulate osteoblast proliferation and differentiation in both sexes (17). It has been demonstrated that dehydroepiandrosterone sulphate (DHEAS) and other adrenal androgens affect bone metabolism, mainly on cortical bone and especially during adrenarche (18). A lack of physiological rise in DHEAS levels during childhood, effectively accounting for absence of a typical adrenarche, has been shown in children with classic CAH (19). Thus, low DHEAS levels due to a blunted response of the adrenals and the glucocorticoid effect can impair growth and osteoblast function in patients with CAH and may be the cause of decreased bone mineral density (BMD) seen in most; however, not all studies of BMD in patients with CAH (14,20-30). In a recent meta-analysis of adults with CAH, a slight decrease in BMD compared to matched controls was found (31). Also, the clinical importance is not BMD by itself but fragility fractures since these are the main cause of morbidity in osteoporosis. However, the association between BMD and fracture risk, particularly in glucocorticoid-induced osteoporosis, is poor (16).

Few studies have previously investigated fractures in patients with CAH; they were all quite small (12,18,24,25,28-30,32-35), and the reported fracture rate varied between 0% and 53.3% and the osteoporotic fracture rate varied from 0% to 20% (36). However, how fractures were identified differed between the studies, and many did not mention how fractures occurred. To be able to better evaluate the fracture risk, comparisons with controls are required. Five small studies with patients with CAH have included controls (24,25,28,30,33). One of these included only adult females with CAH and showed a clear increased risk for any fractures but just a tendency for osteoporotic fractures (24). Another study including only adult males with CAH did not show an increase, but the number of fractures in matched controls was quite high (25). The 3 other studies hardly found any fractures at all (28,30,33).

Thus, the aim of the present study was to investigate the prevalence of fractures in all individuals with 21OHD in Sweden and compare to matched controls. In addition, we assessed whether the outcomes differed by sex, age group, the different pheno- and genotypes, and before and after the introduction of the nationwide neonatal screening.

Methods

Patients

Patients with 21OHD (with a complete personal identification number) born between 1910 and 2013 were identified using the National CAH Registry (n = 640) (2) as well as the National Patient Register, using the International Classification of Diseases (ICD)-8 (255.01, 255.08), ICD-9 (2552, 255C), and ICD-10 (E25.0) (n = 74). If a patient had been registered 3 or more times with these ICD codes, further scrutinization was done by checking all their ICD codes to determine whether an alternative diagnosis was more likely. All patients with 21OHD found via the National Neonatal Screening Program, late-diagnosed patients reported to the screening laboratory, all Swedish patients with a diagnostic CYP21A2 mutations analysis, and all patients known to our hospital through previous or current clinical contacts or studies (since the 1940s) are included in the National CAH Registry. This procedure has been reported in detail in previous studies (7,9,11,13,37,38). In total, 714 patients, 404 females and 310 males, with 21OHD were included in the study.

Patients were arbitrarily divided into 3 age groups (0-19, 20-49,and ≥50 years) as well as by birth year before or after the introduction of neonatal screening (1986). If data were available, the patients were also divided into the 3 phenotype groups (ie, SW, SV, and NC) as well as into the 5 most frequent genotype groups (ie, null, I2 splice, I172N, P30L, and V281L) according to CYP21A2 mutation analysis, as previously described (2,39). The mildest mutation defines the genotype group in compound heterozygotes. Null is associated with the SW phenotype, I2 splice is usually associated with SW; I172N, with SV; and V281L, with the NC form (39). The severity of P30L is between SV and NC (40), but it was defined in this study as SV. The NC group included patients with clinically and biochemically and/or genetically verified NC disease.

Study Protocol

Controls matched by birth year, sex, and place of birth were identified from the Total Population Register in a ratio of 100 for each case of 21OHD. Matching for immigration to Sweden was done using the Migration Records (Statistics Sweden), which contain all migrations since 1901. By using the unique Swedish personal identification number, unambiguous linkage between the population-based registers was possible. All data were deidentified prior to delivery by the register holders. The National Patient Register (Swedish Board of Health and Welfare) was used to identify all discharge diagnoses according to the ICD for both in- and outpatient specialist care since 1964 and 2001, respectively. The outcome, a fracture, was registered, but type of trauma was also recorded. The prevalence was calculated at the time point the data extraction was done (2013), at the time of death, or migration out of the country, whichever came first. The different ICD codes (Swedish version) used for the separate analyses are shown in Table 1.

Table 1.

Fractures and manner of occurrence based on ICD diagnoses from the National Patient Registry including both in- and outpatient care

Diagnosis	ICD8	ICD9	ICD10
Any fracture	800-829	800-829	S02, S22, S32, S42, S52, S62, S72, S82, S92, T02, T08, T10, T12, T14.2
Any major osteoporotic fracturea	805, 806, 813, 820, 810-812	805, 806, 813, 820, 810-812	S12, S22.0, S22.1, S32.0, S32.7, T02.1, T08, S42, S52, S72, T10
Fracture of skull or facial bones	800-804	800-804	S02
Fracture of rib(s), sternum and thoracic spine	807.00, 807.10, 807.90, 806.21, 806.31, 806.91	805C, 805D, 806C, 806D, 807, 809	S22
Fracture of rib(s)	807.00, 807.10, 807.90	807A, 807B	S22.3, S22.4
Fracture of lumbar spine and pelvis	806.22, 806.32, 806.92, 808.00, 808.10, 808.90	808, 805E-805F, 806E-806F	S32
Fracture of shoulder and upper arm	810-812	810-812	S42
Fracture of humerus	812	812	S42.2, S42.3, S42.4, S42.7
Fracture of forearm	813	813	S52
Fracture of lower end of radius	813.00, 813.10, 813.90, 813.42, 813.52, 813.92	813E	S52.5, S52.6
Fracture at wrist and hand level	814-817	814-817	S62
Fracture of femur	820-821	820-821	S72
Fracture of hip	820	820	S72.0, S72.1, S72.2
Fracture of lower leg, including ankle	823, 824	823, 824	S82
Fracture of foot, except ankle	825, 826	825, 826	S92
Fracture of spine	805, 806	805, 806	S12, S22.0, S22.1, S32.0, S32.7, T02.1, T08
Transport accidents	E807-E846	E800-E849	V01-V99
Pedestrian injured in transport accident	E807.2	E819H	V01-V09
Motorcycle rider injured in transport accident	E819.2, E819.3	E819C, E819D	V20-V29
Car occupant injured in transport accident	E819.0, E819.1	E819A, E819B	V40-V49
Falls	E880-E887	E880-E888	W00-W19
Fall on same level	E885.9, E886.9	E885, E886	W00, W01, W03, W18

aMajor osteoporotic fracture is defined as any fracture of the spine, forearm, hip, or shoulder.

The study was approved by the Swedish Ethical Review Authority, Sweden.

Statistical Analysis

Mean ± SD is reported for continuous variables if normally distributed, while absolute and relative frequencies are used for categorical outcomes. Categorical parameters were compared using Fisher’s exact test and odds ratio (OR) calculations with 95% CI for the composite outcome “any fracture” and “any major osteoporotic fracture” (ie, fractures of the spine, forearm, hip or shoulder) in the entire cohort as well as in some subgroups. A P-value < 0.05 was considered statistically significant.

Results

Characteristics of the Patients and Controls

The mean age of the 714 patients with 21OHD included in the study was 29.8 ± 18.4 years (oldest included was 83 years). More females with 21OHD (n = 404, mean age 30.5 ± 18.0 years) than males (n = 310, mean age 28.7 ± 18.8) were included (Table 2). Patients were also divided into the 3 different age groups: 0 to 19 years old (n = 223, females n = 113), 20-49 years old (n = 397, females n = 238), and ≥50 years old (n = 94, females n = 53). In 566 patients with 21OHD (79.3%), the severity of the disease could be established. Table 3 and 4 show the specific details concerning the number of patients with 21OHD and their mean age in the different phenotype (SW n = 288, SV n = 188 and NC n = 90, respectively) and genotype groups (null n = 115, I2 splice n = 155, I172N n = 146 and P30L n = 29, respectively). National neonatal screening for CAH was introduced in Sweden in 1986 (4); 335 patients with 21OHD (females n = 197) were born before the introduction and 379 afterwards (females n = 207). Controls, matched for sex and year and place of birth, were included from the Total Population Registry (n = 71 400). This cohort was updated up to 2013 (previously 2009) (13,38,41), and hence more individuals have been included than in the cohort reported by our group in previous studies (7,9,11,37).

Table 2.

Fractures in individuals with congenital adrenal hyperplasia due to 21-hydroxylase deficiency, also divided into females and males, compared with age- and sex-matched controls (100 controls per case)

	CAH individuals	Controls	P-value	CAH females	Controls females	P-value	CAH males	Controls males	P-value
n	714	71 400		404	40 400		310	31 000
Any fracture	168 (23.5)	11 476 (16.1)	<0.00001	79 (19.6)	5387 (13.3)	0.0005	89 (28.7)	6089 (19.6)	0.0001
≥2	95 (13.3)	6486 (9.1)	0.0002	48 (11.9)	2930 (7.3)	0.001	47 (15.2)	3556 (11.5)	0.0487
Any major op fracturea	70 (9.8)	5339 (7.5)	0.0221	36 (8.9)	2693 (6.7)	0.0876	34 (11.0)	2646 (8.5)	0.1262
≥2	42 (5.9)	2852 (4.0)	0.016	23 (5.7)	1443 (3.6)	0.0304	19 (6.1)	1409 (4.6)	0.172
Fracture head	14 (2.0)	1099 (1.5)	0.3568	5 (1.2)	344 (0.9)	0.4034	9 (2.9)	755 (2.4)	0.5755
Fracture rib(s)	23 (3.2)	877 (1.2)	<0.00001	12 (3.0)	469 (1.2)	0.0034	11 (3.6)	408 (1.3)	0.0033
Fracture LS and pelvis	4 (0.6)	370 (0.5)	0.7903	1 (0.3)	213 (0.5)	0.7283	3 (1.0)	157 (0.5)	0.2117
Fracture shoulder and upper arm	27 (3.8)	1677 (2.3)	0.0178	16 (4.0)	780 (1.9)	0.0095	11 (3.6)	897 (2.9)	0.4928
Fracture humerus	20 (2.8)	923 (1.3)	0.0022	15 (3.7)	518 (1.3)	0.0003	5 (1.6)	405 (1.3)	0.609
Fracture forearm	40 (5.6)	3215 (4.5)	0.173	19 (4.7)	1659 (4.1)	0.5278	21 (6.8)	1556 (5.0)	0.1517
Fracture lower end of radius	34 (4.8)	2448 (3.4)	0.0621	15 (3.7)	1264 (3.1)	0.4718	19 (6.1)	1184 (3.8)	0.0509
Fracture hand	38 (5.3)	2429 (3.4)	0.0092	16 (4.0)	898 (2.2)	0.0266	22 (7.1)	1531 (4.9)	0.0866
Fracture femur	9 (1.3)	578 (0.8)	0.2012	4 (1.0)	321 (0.8)	0.5682	5 (1.6)	257 (0.8)	0.1199
Fracture hip	2 (0.3)	322 (0.5)	0.776	1 (0.3)	231 (0.5)	0.733	1 (0.3)	91 (0.3)	0.6002
Fracture lower leg and ankle	36 (5.0)	2105 (3.0)	0.0025	17 (4.2)	1035 (2.6)	0.0552	19 (6.1)	1070 (3.5)	0.0178
Fracture foot	18 (2.5)	1453 (2.0)	0.3497	11 (2.7)	667 (1.7)	0.1117	7 (2.3)	786 (2.5)	1
Fracture spine	6 (0.8)	433 (0.6)	0.4595	0 (0)	194 (0.5)	0.2719	6 (1.9)	239 (0.8)	0.0358
Transport accidents	56 (7.8)	4244 (5.9)	0.0385	34 (8.4)	2204 (5.5)	0.015	22 (7.1)	2040 (6.6)	0.7292
Pedestrian	0 (0)	169 (0.2)	0.4209	0 (0)	90 (0.2)	1	0 (0)	79 (0.3)	1
Motorcycle rider	7 (1.0)	725 (1.0)	1	2 (0.5)	198 (0.5)	0.7266	5 (1.6)	527 (1.7)	1
Car occupant	26 (3.6)	1382 (1.9)	0.0025	14 (3.5)	756 (1.9)	0.0264	12 (3.9)	626 (2.0)	0.0385
Any falls	190 (26.6)	13 722 (19.2)	<0.00001	102 (25.3)	7118 (17.6)	0.0001	88 (28.4)	6604 (21.3)	0.0034
Fall on same level	111 (15.6)	7065 (9.9)	<0.00001	59 (14.6)	3671 (9.1)	0.0003	52 (16.8)	3394 (11.0)	0.0019

Data are given a n or n (%). Bolded P-values, P < 0.05; italicized P-values, P = 0.05-0.09.

Abbreviations: CAH, congenital adrenal hyperplasia; LS, lumbar spine; op, osteoporotic.

aMajor osteoporotic fracture is defined as any fracture of the spine, forearm, hip, or shoulder.

Table 3.

Fractures in Individuals with congenital adrenal hyperplasia due to 21-hydroxylase deficiency divided into the phenotypes and compared with age- and sex-matched controls (100 controls per case)

	SW			SV			NC
	All	Females	Males	All	Females	Males	All	Females	Males
n	288	157	131	188	101	87	90	67	23
Age, years	24.5 ± 16.1	25.3 ± 15.6	23.6 ± 16.7	32.5 ± 19.3	32.2 ± 17.5	32.8 ± 21.3	29.3 ± 15.9	30.2 ± 15.7	26.7 ± 16.4
Any fracture	62 (21.5) **	30 (19.1) *	32 (24.4)#	54 (28.7) ***	26 (25.7) ***	28 (32.2) *	14 (15.6)	9 (13.4)	5 (21.7)
≥2	34 (11.8)#	18 (11.5) *	16 (12.2)	25 (13.3)#	12 (11.9) *	13 (14.9)	10 (11.1)	7 (10.5)	3 (13)
Any major op fracturea	27 (9.4)	14 (8.9)	13 (9.9)	20 (10.6)	11 (10.9) *	9 (10.3)	5 (5.6)	3 (4.5)	2 (8.7)
≥2	15 (5.2)	8 (5.1)	7 (5.3)	11 (5.9)	6 (5.9)#	5 (5.8)	3 (3.3)	2 (3)	1 (4.4)
Fracture rib(s)	6 (2.1)	4 (2.6)#	2 (1.5)	11 (5.9) ***	4 (4) *	7 (8.1) ***	3 (3.3)#	3 (4.5) *
Fracture LS and pelvis	1 (0.4)		1 (0.8)	1 (0.5)	1 (1)
Fracture shoulder and upper arm	12 (4.2) *	7 (4.5) *	5 (3.8)	6 (3.2)	4 (4)#	2 (2.3)	2 (2.2)	2 (3)
Fracture humerus	8 (2.8) *	6 (3.8) *	2 (1.5)	5 (2.7)#	4 (4)*	1 (1.2)	2 (2.2)	2 (3)
Fracture forearm	15 (5.2)	7 (4.5)	8 (6.1)	14 (7.5)#	7 (6.9)#	7 (8.1)	2 (2.2)	1 (1.5)	1 (4.4)
Fracture lower end of radius	12 (4.2)	5 (3.2)	7 (5.3)	12 (6.4) *	6 (5.9) *	6 (6.9)	2 (2.2)	1 (1.5)	1 (4.4)
Fracture hand	16 (5.6) *	9 (5.7) *	7 (5.3)	12 (6.4) *	4 (4)	8 (9.2)#	3 (3.3)	1 (1.5)	2 (8.7)
Fracture femur	3 (1)	1 (0.6)	2 (1.5)	2 (1.1)	1 (1)	1 (1.2)	2 (2.2)#	1 (1.5)	1 (4.4)
Fracture hip	1 (0.4)		1 (0.8)
Fracture lower leg and ankle	13 (4.5)#	6 (3.8)	7 (5.3)	12 (6.4) *	7 (6.9) *	5 (5.8)	3 (3.3)	2 (3)	1 (4.4)
Fracture foot	7 (2.4)	4 (2.6)	3 (2.2)	3 (1.6)	1 (1)	2 (2.3)	4 (4.4)#	4 (6) *
Fracture spine	2 (0.7)		2 (1.5)	2 (1.1)		2 (2.3)
Transport accidents	22 (7.6)	12 (7.6)	10 (7.6)	16 (8.5)	12 (11.9) *	4 (4.6)	4 (4.4)	3 (4.5)	1 (4.4)
Motorcycle rider	2 (0.7)	1 (0.6)	1 (0.8)	2 (1.1)	1 (1)	1 (1.2)	1 (1.1)		1 (4.4)
Car occupant	13 (4.5) **	7 (4.5) *	6 (4.6) *	6 (3.2)	4 (4)	2 (2.3)
Any falls	73 (25.4) **	43 (27.4) **	30 (22.9)	62 (33) ***	30 (29.7) ***	32 (36.8)**	22 (22.2)	15 (22.4)	5 (21.7)
Fall on same level	44 (15.3) **	25 (15.9) **	19 (14.5)	33 (17.6) **	16 (15.8) *	17 (19.5)*	10 (11.1)	7 (10.5)	3 (13)

Data are given as n, mean ± SD, or n (%). Data are not displayed if no patient had the condition.*, P < 0.05. **, P < 0.01. ***, P < 0.001. #, P = 0.05-0.099.

Abbreviations: LS, lumbar spine; op, osteoporotic.

aMajor osteoporotic fracture is defined as any fracture of the spine, forearm, hip, or shoulder.

Table 4.

Fractures in individuals with congenital adrenal hyperplasia constituting the four most common CYP21A2 genotype groups compared with age- and sex-matched controls (100 controls per case)

	Null			I2 splice			I172N				P30L
	All	Females	Males	All	Females	Males	All	Females	Males	All	Females	Males
n	115	63	52	155	85	70	146	79	67	29	15	14
Age	23.9 ± 14.8	24.7 ± 13.2	22.9 ± 16.7	23.8 ± 16.7	24.6 ± 17.0	22.7 ± 16.4	32.8 ± 20.4	32.8 ± 18.6	32.7 ± 22.4	25.6 ± 10.3	26.2 ± 10.4	24.9 ± 10.5
Any fracture	30 (26.1) **	16 (25.4) *	14 (26.9)*	28 (18.1)	13 (15.3)	15 (21.4)	44 (30.1) ***	20 (25.3) **	24 (35.8) **	8 (27.6)#	5 (33.3) **	3 (21.4)
≥2	17 (14.8) *	10 (15.9) *	7 (13.5)	16 (10.3)	7 (8.2)	9 (12.9)	20 (13.7)#	9 (11.4)	11 (16.4)	3 (10.3)	2 (13.3)	1 (7.1)
Any major op fracturea	13 (11.3)#	8 (12.7)	5 (9.6)	14 (9)	6 (7.1)	8 (11.4)	16 (11)	9 (11.4)#	7 (10.5)	3 (10.3)	1 (6.7)	2 (14.3)
≥2	7 (6.1)	3 (4.8)	4 (7.7)	8 (5.2)	5 (5.9)	3 (4.3)	8 (5.5)	4 (5.1)	4 (6)	2 (6.9)	1 (6.7)	1 (7.1)
Fracture rib(s)	1 (0.9)	1 (1.6)		5 (3.2) *	3 (3.5) *	2 (2.9)	9 (6.2) ***	3 (3.8)#	6 (9) ***	2 (6.9) *	1 (6.7) *	1 (7.1)
Fracture LS and pelvis				1 (0.7)	1 (1.2)		1 (0.7)	1 (1.3)
Fracture shoulder and upper arm	7 (6.1) *	4 (6.4)#	3 (5.8)	5 (3.2)	3 (3.5)	2 (2.9)	4 (2.7)	3 (3.8)	1 (1.5)	2 (6.9)	1 (6.7)	1 (7.1)
Fracture humerus	5 (4.4) *	3 (4.8) *	2 (3.9)	3 (1.9)	3 (3.5)#		4 (2.7)	3 (3.8)#	1 (1.5)	1 (3.5)	1 (6.7)
Fracture forearm	7 (6.1)	4 (6.4)	3 (5.8)	8 (5.2)	3 (3.5)	5 (7.1)	11 (7.5)#	6 (7.6)	5 (7.5)	2 (6.9)		2 (14.3)
Fracture lower end of radius	6 (5.2)	3 (4.8)	3 (5.8)	6 (3.9)	2 (2.4)	4 (5.7)	10 (6.9)*	5 (6.3)#	5 (7.5)	1 (1.5)		1 (7.1)
Fracture hand	10 (8.7) **	6 (9.5) **	4 (7.7)	5 (3.2)	2 (2.4)	3 (4.3)	9 (6.2)#	1 (1.3)	8 (12) *	3 (10.3)	3 (20) **
Fracture femur				3 (1.9)#	1 (1.2)	2 (2.9)#	2 (1.4)	1 (1.3)	1 (1.5)
Fracture hip				1 (0.7)		1 (1.4)
Fracture lower leg and ankle	8 (7) **	4 (6.4) *	4 (7.7)#	4 (2.6)	2 (2.4)	2 (2.9)	11 (7.5) *	6 (7.6) *	5 (7.5)	1 (3.5)	1 (6.7)
Fracture foot	5 (4.4)#	4 (6.4) *	1 (1.9)	2 (1.3)		2 (2.9)	1 (0.7)		1 (1.5)	1 (3.5)	1 (6.7)
Fracture spine	1 (0.9)		1 (1.9)	1 (0.7)		1 (1.4)	2 (1.4)		2 (3)
Transport accidents	11 (9.6)	8 (12.7)#	3 (5.8)	9 (5.8)	2 (2.4)	7 (10)	15 (10.3) *	11 (13.9) **	4 (6)	1 (3.5)	1 (6.7)
Motorcycle rider				1 (0.7)		1 (1.4)	2 (1.4)	1 (1.3)	1 (1.5)
Car occupant	5 (4.4) *	3 (4.8)	2 (3.9)	6 (3.9)#	2 (2.4)	4 (5.7) *	6 (4.1)#	4 (5.1)#	2 (3)
Any falls	34 (29.6) *	22 (34.9) *	12 (23.1)	33 (21.3)	19 (22.4)	14 (20)	49 (33.6) ***	24 (30.4) **	25 (37.3) **	10 (34.5) *	4 (26.7)#	6 (42.9)
Fall on same level	18 (15.7) *	12 (19.1) *	6 (11.5)	24 (15.5) **	12 (14.1) *	12 (17.1)#	25 (17.1)**	12 (15.2)#	13 (19.4)#	7 (24.1)**	3 (20) *	4 (28.6)#

Severity of the genotype ranges from left to right.Data are given as n, mean ± SD, or n (%). Data are not displayed if no patient had the condition. *, P < 0.05. **, P < 0.01. ***, P < 0.001. #, P = 0.05-0.099.

Abbreviations: LS, lumbar spine; op, osteoporotic.

aMajor osteoporotic fracture is defined as any fracture of the spine, forearm, hip, or shoulder.

Any Fracture

More patients with CAH than controls had had a fracture (23.5% vs 16.1%, OR 1.61, 95% CI 1.35-1.91), and this was seen in both sexes (females: 19.6% vs 13.3%, OR 1.57, 95% CI 1.23-2.02; males: 28.7% vs 19.6%, OR 1.65, 95% CI 1.29-2.12) (Table 2). Also, 2 or more different fractures were more common in patients with CAH than in controls, irrespective of sex. When assessing the different age groups, more patients with CAH than controls had any fracture in all 3 age groups (0-19 years: 14.7% vs 10.3%, OR 1.52, 95% CI 1.05-2.20; 20-49 years: 24.7% vs 17.9%, OR 1.50, 95% CI 1.20-1.89; ≥50 years: 39.4% vs 22.6%, OR 2.22, 95% CI 1.47-3.37). The fracture prevalence was also significantly increased in both sexes in the 2 older age groups but not in children (data not shown). In the phenotype groups SW and SV, the prevalence of any fracture was higher than in controls, but this did not quite reach statistically significance in males with the SW phenotype (Table 3). Two or more different fractures were only significantly more common in the females with the SW and SV phenotype. Individuals with NC phenotype had no increased prevalence of fractures. In the genotype groups null, I172N and P30L (only women) were more affected by any fracture, while the patients in the I2 splice group were similarly affected as controls (Table 4). Multiple different fractures were more common in the null group, especially in females.

Any Major Osteoporotic Fracture

The prevalence of any major osteoporotic fracture, predominantly acquired during falls on the same level, was increased in all individuals with CAH (9.8% vs 7.5%, OR 1.34, 95% CI 1.05-1.72), but this did not reach significant levels in either sex, although it came close in females (8.9% vs 6.7%, OR 1.36 95% CI 0.97-1.93) (Table 2). Multiple different major osteoporotic fractures were more common in all individuals with CAH, especially in females. Any major osteoporotic fracture was not significantly increased in any of the age groups (data not shown). In the phenotype groups, only females with SV had an increased frequency of any major osteoporotic fracture (Table 3), while in the genotype groups all patients with null genotype or females with the I172N genotype the increased prevalence almost became significant (Table 4).

Before and After the Introduction of Neonatal Screening

More patients with CAH born before the introduction of neonatal screening had had a fracture compared to controls (29.9% vs 17.8%, OR 1.97, 95% CI 1.56-2.49), and this was seen in both sexes (females: 25.3% vs 14.5%, OR 2.01, 95% CI 1.45-2.77; males: 36.2% vs 22.4%, OR 1.96, 95% CI 1.39-2.79). Similar results were seen in any major osteoporotic fracture (Table 5).

Table 5.

Fractures in individuals with congenital adrenal hyperplasia due to 21-hydroxylase deficiency divided into those born before the introduction of neonatal screening (1986) and those afterwards compared with age- and sex-matched controls (100 controls per case)

	Born before introduction of neonatal screening					Born after introduction of neonatal screening
	CAH individuals	Controls	P-value	CAH females	CAH males	CAH individuals	Controls	P-value	CAH females	CAH males
n	335	33 500		197	138	379	37 900		207	172
Any fracture	100 (29.9)	5950 (17.8)	<0.00001	50 (25.4) ***	50 (36.2) ***	68 (17.9)	5526 (14.6)	0.0678	29 (14)	39 (22.7) #,
Any major op fracturea	35 (10.4)	2389 (7.1)	0.0248	20 (10.2) *	15 (10.9)	35 (10.2)	2950 (7.8)	0.2892	16 (7.7)	19 (11)

Data are given as n or n (%). *, P < 0.05; **, P < 0.01; ***, P < 0.001; #, P = 0.05-0.09. Bolded P-value, P < 0.05.

Abbreviations: CAH, congenital adrenal hyperplasia; op, osteoporotic.

aMajor osteoporotic fracture is defined as any fracture of the spine, forearm, hip, or shoulder.

In those patients with CAH born after the introduction of neonatal screening, the difference was not significantly increased compared to controls (17.9% vs 14.6%, OR 1.28, 95% CI 0.98-1.67), and similarly nonsignificant results were seen in both sexes (females: 14% vs 12.2%, OR 1.17, 95% CI 0.70-1.95; males: 22.7% vs 17.4%, OR 1.39, 95% CI 0.97-1.99). Comparable nonsignificant results were also seen for any major osteoporotic fracture (Table 5).

Specific Fractures

In all patients with CAH fractures in ribs, shoulder and upper arm, humerus, hand, and lower leg including ankle were more common than in controls (Table 2). Although not significant, fractures in the lower end of radius were increased in all patients with CAH. Similar fractures were seen in females with CAH while men had significantly more fractures in the ribs and spine, but fractures of the lower end of radius and hand did not quite reach statistically significant levels (Table 2). In the SW group, fractures in the shoulder and upper arms, humerus, and hands were increased, which were similarly seen in females while men had no significant increase in any of the specific fracture types (Table 3). In the SV group, fractures in ribs, lower end of radius, hands and lower legs including ankle were increased. Again, this was seen in females while in men only fractures in the ribs were more prevalent (Table 3). In the NC group, only more rib and foot fractures in females were seen. In the genotype groups fractures differed slightly between the groups (Table 4). In the null genotype group, more humerus, hand, lower leg including ankle, and foot fractures were seen in females while this was not found in men. In the I2 splice genotype group, only rib fractures were increased, especially in females. In the I172N genotype group, many fracture types did not quite reach significant levels but fracture of the ribs, lower end of the radius, and lower end of leg including ankle did (Table 4). In the P30L group, fractures in ribs and hands occurred more in females.

Type of Trauma

A transport accident, especially as a car passenger or driver were more common in patients with CAH, both sexes, than in controls (Table 2). Any fall, including falls on the same level, was more common in all CAH patients and in both sexes. In the phenotype groups, both females and males with SW had more accidents as car occupants, and falls were more common in females with SW and in both sexes with SV (Table 3). In the genotype groups, transport accidents were more common in females with I172N genotype and as a car occupant in the null and male I2 splice group (Table 4). Any fall as well as fall on the same level was more common in females with null genotype group. Falls on the same level were more common in females with I2 splice. Any fall was more common in both sexes in the I172N group, as well as falls on the same level, but this was not significant when stratified for the different sexes, although with a tendency. In the P30L group, any fall as well as fall on the same level were more common, especially in females, compared to controls (Table 4).

Discussion

This is, to date, the largest study investigating fractures in patients with CAH, and it included all patients diagnosed with 21OHD in Sweden. Moreover, the fracture rates were compared to matched controls. Fractures in women and men, different age groups, and before and after introduction of neonatal screening as well as the different pheno- and genotypes were also studied separately.

The odds of any fractures were 1.61 higher in all patients with CAH compared to controls, and this was similar in both sexes. Similar findings were seen if 2 or more different fractures were analyzed or only major osteoporotic fractures were included. Fractures were significantly increased in patients before the introduction of neonatal screening but not in those born afterwards. Transport accidents as a car occupant or fall on same level were more common in patients with CAH. When the pheno- and genotypes were analyzed separately, SW and SV in addition to null, I172N, and P30L (only females significant) groups had increased prevalence of any fractures while major osteoporotic fractures were only significantly more common in SV females. Patients with NC phenotype or I2 splice genotype did not demonstrate any increased risk of any fractures at all. It should be noted that the average age was only around 30 years, and most osteoporotic fractures occur in older age. Nonetheless, there was a clear increased frequency of osteoporotic fractures in all patients with CAH. It should be noted that SV phenotype and I172N genotype groups were the oldest, which may explain why they had most fractures.

Androgen excess in CAH may contribute to fewer fractures due to its protective role, but this effect of androgens is difficult to tease out due to the interplay of many additional confounding factors. Poor glucocorticoid adherence increases androgen concentrations, and this may increase BMD, while good adherence may result in low androgen concentrations (42,43) and impaired BMD (25). Increased adrenal androgen concentrations in males with CAH, on the other hand, may decrease the total testosterone concentration due to inhibition of gonadotropins (44).We did not find an increased fracture rate in patients with the NC phenotype, which may be explained by prolonged androgen exposure before diagnosis, at least in females. Although this could at least in part be an issue of power since there were fewer patients in the NC than SW or SV phenotype groups. In the study by El-Maouche et al, the nontraumatic fracture rate was higher in classic than in NC CAH (18), which resembles our findings. Why patients with the I2 splice genotype did not have an increased fracture rate is unclear, but since many of them were born before the introduction of neonatal screening in Sweden 1986 (45), some were probably not diagnosed in the neonatal period and could have been exposed to a prolong period of androgen excess. However, although some patients with I172N genotype were probably diagnosed late, they still had increased fracture rate. Prior to the introduction of neonatal screening, especially boys but also some girls with the SV phenotype were diagnosed later during childhood and thus were not treated with replacement doses of glucocorticoids from the neonatal period (42,46). Having said that, patients born before the introduction of neonatal screening had significantly increased fracture prevalence compared to controls while this was not seen in those born after the introduction of neonatal screening. It could be speculated that early diagnosis and modern management of CAH may have had a larger effect on bone health than prolonged androgen exposure. However, those neonatally screened were younger than those not screened, and fracture prevalence is lower in younger age groups. A possible increase in fracture rate may become significant compared to their controls once they become older.

Both the glucocorticoid dose and regimen may affect BMD (29,35,47). Prednisolone seems to impair BMD more than hydrocortisone through multiple different mechanisms (25,29). However, whether this results in higher fracture rate is unclear, and the current study unfortunately could not illuminate this since we neither had the glucocorticoid doses nor regimens available, but this would be interesting to investigate in future studies. Since patients with classic CAH require glucocorticoid for survival (1,2), it is plausible to assume that the majority of patients in the current cohort were on long-term glucocorticoid replacement. The exceptions are some individuals with the milder mutations such as P30L and V281L, the latter typically consistent with NC CAH (5).

Females with CAH have more gender-atypical behavior and interests with higher likelihood of having male-dominant occupations, participation in rough sports activities, and motor interests (48), and this is associated with the severity of CAH. Thus, females with the most severe pheno- and genotype may be exposed to more trauma, which, in turn, may result in more fractures. In contrast, males with CAH may even have less interest in sport activities than controls (49). In a study of 61 adult females with CAH, 30% had had a fracture compared to 3% of matched controls (24), while in 30 adult males with CAH, 53% had experienced at least 1 fracture compared to 44% of matched controls (25). It should be noted that all these fractures were self-reported, and many were minor fractures such as finger and toe fractures. It could be argued that the fracture rate of females with CAH were more similar to the male controls than their own matched controls and that it may be related to a different way of life. However, in the current study, fall on the same level, which may indicate a low trauma (ie, fragility) fracture was more common in patients with CAH. Yet, we cannot exclude that many fractures may have occurred during sport activities.

Different studies have used different methods to detect fractures, including self-reports, X-ray screening, and ICD codes (36). Some fractures, such as vertebral fractures, especially if asymptomatic, may not be identified if the spine is not regularly imaged. Moreover, studies relying on self-report may include many minor fractures (24,25), such as finger and toe fractures, that may not be of importance to identify osteoporosis. In contrast, ICD codes may underestimate the fracture prevalence but should identify major symptomatic fractures, especially if treated.

There are several important clinical implications of the current study. Since fractures are common in CAH, the clinicians need to be vigilant in collecting information on previous fractures and symptoms and signs that may indicate undiagnosed fractures, such as vertebral fractures when reviewing the patients. Fracture risk assessment should be performed, but dual-energy X-ray absorptiometry is not recommended by the Endocrine Society guidelines to be performed regularly (3).

The major limitation of the present study was that all outcome data were derived from national registries; thus, we did not have data on therapy, hormone concentrations, and imaging including dual-energy X-ray absorptiometry. The fracture prevalence may have been underestimated since minor fractures and vertebral fractures may not have been diagnosed with an ICD code. Moreover, the mean age of the included patients was low, and the risk of having an osteoporotic fracture increase with age. If we were to repeat the study in a few decades, the associations with osteoporotic fractures may be more obvious. Since a prerequisite for obtaining ethical approval was that all subjects included were anonymized to protect their integrity, we could not compare the study results with information from medical files. Furthermore, despite the large CAH cohort and the number of patients in the different age groups, pheno- and genotypes were limited, so the results from the subgroup analyses must be interpreted with caution due to the low sample sizes. In addition, we only studied 21OHD and no other variants of CAH, but the situation could be suspected to be similar for all forms of CAH since glucocorticoids is the main therapy. In contrast, the strength of this study is the unique national CAH registry with very high coverage of all patients diagnosed in Sweden, with both geno- and phenotype available for the vast majority. By including the patients identified via the National Patient Register, we were able to obtain virtually complete coverage. Moreover, 100 matched controls for each 21OHD case made the analyses robust.

In conclusion, patients with CAH had an increased prevalence of both any fracture and major osteoporotic fracture. Both sexes were affected. Patients with the SV phenotype and the I172N genotype had most fractures. However, patient born after the introduction of neonatal screening did not have significantly increased prevalence compared to controls so it could be speculated that this may be less of an issue in the future when all patients with classic CAH have been diagnosed neonatally and received modern management. Fracture risk assessment should be performed regularly. Supraphysiological glucocorticoid replacement dosing in combination with low androgen concentrations is the most probable cause for the increased risk of fractures. Glucocorticoid therapy should be optimized, together with lifestyle interventions, to improve bone health.