Trends in the Incidence of Central Precocious Puberty and Normal Variant Puberty Among Children in Denmark, 1998 to 2017

Key Points Question Have the annual incidence rates of central precocious puberty (CPP) and normal variant puberty (ie, premature thelarche and premature adrenarche) increased during the past 20 years in Denmark? Findings This cohort study of 8596 children in Denmark who were diagnosed with CPP, premature thelarche, or premature adrenarche between 1998 and 2017 found that the annual incidence of CPP and normal variant puberty has increased substantially in Denmark during the last 20 years. Meaning The findings of this study have implications for short-term and long-term health and potentially for the international classification of the reference age at puberty.


Introduction
Puberty marks the dynamic transition from childhood to the mature adult phenotype with full reproductive capacity. In girls, puberty most commonly starts with breast development (thelarche) and is followed by the first menstrual bleeding (menarche) approximately 3 years later. Timing of puberty exhibits wide interindividual variation (age 8-13 years in girls; age 9-14 years in boys), and known factors regulating timing and tempo of puberty include familial history and genetics, lifestyle, nutrition, and environmental exposures. 1 Understanding changes in the timing of puberty as well as increasing referrals of children with suspected premature puberty is important because earlier age at puberty may be associated with psychosocial difficulties, is a risk factor for younger age at first sexual intercourse, and carries negative implications for long-term health, including increased risk of type 2 diabetes, 2,3 greater adiposity, weight gain, and obesity, 4,5 cardiovascular disease, 2 depression, 6 and premature death. 7 Large-scale genetic data suggest that early puberty is associated with increased risk of breast cancer, 8 consistent with extensive epidemiological evidence that early menarche increases the risk of breast cancer. For men, early puberty may increase the risk of prostate cancer. 9 During recent decades, there have been global reports of a secular trend toward earlier onset of puberty in the general population, first reported among girls in the United States in the 1990s [10][11][12][13] and subsequently in boys. 13,14 European data from the same period did not replicate these findings, 15,16 but 20 years later, European data (in 2005-2006) suggested a similar trend toward earlier onset of puberty in girls 17,18 and, to a lesser extent, in boys. 19 This trend toward earlier onset of puberty in the general population has been paralleled by an increase in the number of children (mainly girls) referred for evaluation of suspected precocious puberty (PP). 20 Precocious puberty is defined as the development of secondary sexual characteristics before age 8 years for girls and 9 years for boys 21 and is more commonly diagnosed in girls than boys. 22 In girls, central (or true) precocious puberty (CPP) is generally idiopathic without lesions of the central nervous system, whereas boys are more likely to have an underlying pathological etiology. 23 Classic CPP is defined by premature development of secondary sex characteristics, acceleration of linear growth, advanced bone age, and a pubertal response to a gonadotropin-releasing hormone (GnRH) test. By contrast, premature thelarche (PT) is defined as a self-limiting, isolated breast development associated with a normal growth rate, bone age corresponding to chronological age, and prepubertal response to a stimulation GnRH test. There are 2 forms of PT. The most common form occurs in infancy following the marked sex steroid production in minipuberty and the other at approximately age 5 to 8 years in girls. Premature adrenarche (PA) is a condition characterized by premature pubic hair development (pubarche), normal growth rate, and mild elevations of adrenal androgens after exclusion of serious adrenal disorders (ie, congenital adrenal hyperplasia or adrenal carcinoma). 22 Both PT and PA are benign forms of PP and are associated with normal height potential and age at menarche. To date, very few epidemiological studies of which we are aware have reported the national incidence of PP (only 5 previous studies 22,[24][25][26][27], and no studies to our knowledge have specifically reported incidences of PT and PA ( Table 1). Our previous national Danish study 22 found no secular upward trend in the annual incidence of PP during the 9-year period from 1993 to 2001. However,   more recent European 24,25 and Korean 26,27 studies based on number of prescriptions, medication reimbursements, or data from health insurance agencies have suggested a possible upward trend in the incidence of CPP. Furthermore, a 2020 meta-analysis 28 reported substantial worldwide changes in age at breast development among healthy girls. Early breast development is associated with adiposity and may not constitute full activation of the hypothalamo-pituitary-gonadal hormone axis, which results in progressive central puberty. Thus, age at menarche is also showing a trend toward earlier age of onset but less substantial than that observed for earlier thelarche. Thus, we do not know if the observed worldwide trend in earlier age at thelarche is associated with an increase in girls being evaluated, diagnosed with, and treated for CPP. Therefore, we set out to establish valid epidemiological data on the 20-year nationwide incidence of CPP and normal variant puberty, including PT and PA, in Denmark.

Methods
Research was conducted in accordance with principles of the Declaration of Helsinki. 29 The present registry-based study was approved by the Danish Data Protection Agency  a The generally accepted maximum diagnostic age limit for precocious puberty remains 8 years for girls and 9 years for boys. There is an established 6-month to 12-month lag time between the first observation of signs of puberty reported by parents and the establishment of a record (ie, diagnosis by a pediatrician); this lag was included in 3 of 6 studies (50.0%).
b Data in the present study were analyzed in main analyses with the 1-year lag for diagnosis (ie, at-risk girls aged 0-9 years; at-risk boys aged 0-10 years) and without the 1-year lag for diagnosis (ie, at-risk girls aged 0-8 years; at-risk boys aged 0-9 years). c Incidence was dependent on geographical location (proposed exposure to endocrine-and Related Health Problems, Tenth Revision (ICD-10) is recorded for each patient-hospital contact.
Because health care is free in Denmark, complete case ascertainment is expected in these highquality, validated registries, implying minimal bias. 31 In Denmark, children suspected of PP are evaluated in 1 of 18 pediatric departments according to national guidelines. We identified all hospital patient records of first-time diagnoses for boys and girls for our main area of interest, CPP (ICD-10 codes E22.8 and E30.1) and normal variant puberty, including PT (ICD-10 code E30.8) and PA (ICD-10 code E27.0), in the DNPR from 1998 to 2017 to measure changes in incidence during this time period. The generally accepted maximum diagnostic age limit for PP remains 8 years for girls and 9 years for boys and has not changed since the 1950s.
Hence, the present diagnostic criteria for PP include onset of puberty before age 8 years for girls and age 9 year for boys. However, there is an established 6-month to 12-month lag time between the first observation of signs of puberty reported by parents and the establishment of a record (ie, diagnosis by a pediatrician) in the DNPR. This is explained by the time taken for families to notice and seek medical attention from their general practitioner, for the general practitioner to make a pediatric referral, and for the specialist appointment to occur. The clinical and biochemical evaluation of each child is electronically recorded in the patient record file and automatically transferred to DNPR (ie, as final diagnosis). This lag has previously been reported in a Belgian study 32 and is in accordance with our previous experience, in which we reported a mean lag time of 7 months (range, 0-66 months) in girls and 4 months (range, 0-12 months) in boys. 22 Thus, to account for this lag in our main analyses, we included first-time diagnoses if they were recorded in the DNPR between age 0 to 9 years for girls and 0 to 10 years for boys. In sensitivity analyses and to address potential overestimation of annual incidences due to the introduction of a 12-month lag for diagnosis, we also assessed trends only considering first-time diagnoses if they were recorded in the DNPR between the ages of 0 to 8 years for girls and 0 to 9 years for boys. The age of diagnosis was defined on the date of the respective incident diagnosis.
To determine the potential association of genetics and/or early life environmental exposures of children emigrating to Denmark with the outcomes studied, we stratified our findings into 3 groups

Statistical Analysis
First, the crude distribution of incident cases according to sex, immigration status, and diagnostic subgroup was recorded. The mean incidences of each disorder estimated by computing an annual mean of the total number of children in the reported age group in Denmark during the entire 20-year period and in 5-year subperiods by sex and immigration group, within the entire study period from 1998 to 2017.
Second, the trend in yearly incidence rates were estimated for our primary outcome, CPP, then for benign forms of PP, namely PT and PA, separately and for the sum of all diagnosis subgroups.
Data are visualized as a function of calendar year.  Danish origins compared with boys with Danish origin (data not shown).

JAMA Network Open | Pediatrics
The trends in age-specific incidence of CPP, PT, and PA among girls with Danish origin are shown in Figure 2. The incidence of CPP was less than 5 per 10 000 girls with Danish origin aged 5 years or younger and is highest at age 8 years, with an age-specific incidence of 34 per 10 000 ( Figure 2). For PA, age-specific incidence is highest at ages 7 to 8 years, while for PT, age-specific incidence is highest between infancy and 2 years (Figure 2). For boys with Danish origin, the incidence of CPP was much lower (ie, <1 per 10 000) and increased to 2 to 3 per 10 000 in boys older than 8 years (data not shown). Table 2 also shows the mean annual incidence of CPP, PT, and PA according to first-generation and second-generation immigrant groups during this 20-year period. The 20-year mean incidence of CPP and PA among girls in the first-generation and second-generation immigrant groups were greater than that of girls with Danish origin. The incidence rate for CPP per 10 000 girls in the firstgeneration and second-generation groups were 13.7 (95% CI, 9.3 to 18.2) and 14.2 (95% CI, 4.6 to 23.9), respectively, the incidence rate for PA per 10 000 girls in the first-generation and secondgeneration groups were 2.0 (95% CI, 0.3 to 3.6) and 1.5 (95% CI, −1.6 to 4.7), respectively. The annual incidence of thelarche was similar among all girls (Table 2). No marked differences between immigrant groups were observed among boys (Table 2). When assessing the time trends in mean 5-year incidences of each disorder, a steady increase was observed for all children (boys and girls) with Danish origin (eTable in the Supplement).   Abbreviation: NA, not applicable. a The generally accepted maximum diagnostic age limit for precocious puberty remains 8 years for girls and 9 years for boys. There is an established 6-month to 12-month lag between the first observation of signs of puberty reported by parents and the establishment of a record (ie, diagnosis by a pediatrician), applied here.

Discussion
In this 20-year nationwide registry-based cohort study, we detected a substantial upward trend in the annual incidence of CPP and normal variant puberty (ie, PT and PA) from 1998 to 2017 in girls with Danish origin. Annual incidences were several-fold lower for boys with Danish origin, but a similar substantial upward secular trend was detected. The annual incidence of CPP and PA was substantially greater in girls with non-Danish origins compared with girls with Danish origin.
While national data for CPP are sparse, 5 other studies (3 in Europe 22,24,25 and 2 in Korea 26,27 ) have reported the incidence of CPP (Table 1). In contrast to our previous study, 22 in which no upward trend in the annual incidence of CPP was detected during a 9-year period from 1993 to 2001 in girls or boys, 22 our new data during the 20-year period from 1998 to 2017 demonstrate a 3-fold and 2-fold increased incidence of CPP in girls and boys, respectively. These trends are deeply concerning for several reasons. First, there has been considerable debate about whether current criteria for the diagnosis of PP, defined as any sign of puberty in girls before age 8 years and boys before 9 years, are appropriate in contemporary populations. If indeed there has been a downward trend in the reference age of pubertal onset, this may have led to more girls falsely diagnosed with PP, reflecting outdated reference ranges. In addition to the potential adverse psychosocial implications of this diagnosis, this may also include unjustified hospitalization, brain magnetic resonance imaging (potentially under general anesthesia) to rule out central nervous system pathologies, and a treatment with GnRH agonists for several years that may not have been needed. Second, an increasing incidence of premature sexual development in younger girls potentially is associated with  Our registry-based study does not have access to these individual data, so these mechanisms could not be explored.
The potential influence of environmental exposure to endocrine disruptors has been addressed in a 2018 French national study 25 in which the incidence of CPP during a 3-year period from 2011 to 2013 was evaluated by geographical location. The authors reported that CPP was greater in children living in agricultural areas, suggesting that exposure to agricultural endocrine disruptors may be associated with CPP. 25 That study used similar inclusion criteria and diagnostic criteria as our study, with comparable annual incidences for CPP. 25 Two Korean studies 26 26,27 The mean incidence of CPP was several fold higher in these Korean studies compared with our data. Given that these the studies did not use the same inclusion criteria, direct comparisons can only be made with caution, but these differences may potentially indicate the importance of ethnicity in this condition.
A Spanish observational study of 34 pediatric endocrinology units (including 250 patients) 24 estimated the annual incidence of CPP at 0.06 per 10 000 children in 2008 to 2009, substantially lower than other European studies. Unlike in Denmark, health care is not free in Spain, which may lead to a noncomplete case ascertainment and underestimation of cases in that study.
To our knowledge, this is the first study to report the national incidence of PA and PT. Our findings of an increased incidence of PT is consistent with a global reduction of 3 months per decade in the reference age at pubertal onset, as measured by thelarche, the first clinical sign of puberty.
This was summarized in our 2020 systematic review and meta-analysis 28 of 30 studies including more than 145 000 girls from 7 regions of the world (eg, the United States, Europe, Asia). 28 Although we find the upward trends in incidence of CPP concerning, there has also been considerable debate regarding whether present guidelines of CPP are appropriate in contemporary populations. There may be basis for revision of reference ranges for pubertal onset based on new thorough clinical investigations of CPP diagnosis and interrelation with bone age advancement and hormonal changes in the general population.
Screening programs identifying early puberty could lead to preventative health measures given that early pubertal onset is consistently associated with adverse short-term and long-term health outcomes. If there is an international shift in the reference ranges for puberty in the general population, this carries implications for the current diagnostic criteria, which may need to be updated to reflect this change.

Strengths and Limitations
In Denmark, individual-level information on disease has been collected in nationwide registries for several decades. Because health care is free, these high-quality registries have complete disease ascertainment with minimal bias in data collection owing to socioeconomic status. Denmark has a genetically homogenous population (>98% White individuals); thus, the results of the study would be generalizable to other similar populations.
We were unable to identify children who were adopted (approximately 0.8% of Danish children), so approximately 59 cases identified as having Danish origin could have been adopted and belonged to a different ethnic group. However, we do not expect this number to skew results substantially.
Our study was registry-based, and although Danish registries are known for their completeness and validity, there are still some limitations. First, none of the ICD-10 codes were validated by investigation of hospital records, which may have affected the number of cases included. Thus, it is not known to what degree patients with registration of the ICD-10 codes E22.8 and E30.1 met all diagnostic criteria for progressive CPP. Second, there is a chance that the initial diagnosis coding of a child suspected of having CPP was not updated appropriately when subsequent investigation did not confirm the diagnosis. However, in our previous epidemiological study, 22 we performed a rigorous validation of registered diagnoses and showed that 96% of registered cases were classified correctly as a disorder of pubertal development including CPP, PA, and PT. In addition, any potential minor variations in registration of ICD-10 codes would be random and would not affect the overall trends we observed.
During recent decades, there has been growing parental awareness of and concerns regarding earlier puberty due to media coverage in Denmark. This may have increased the number of children referred to our specialist service and other public and private Danish hospitals.

Conclusions
This nationwide, registry-based cohort study during 20 years demonstrated a substantial increase not only in PT and PA but also in true CPP. These findings have implications for short-term and longterm health and potentially for the international reference classification of age at puberty.