Trends in Incidence of Adolescent Idiopathic Scoliosis: A Modern U.S. Population-Based Study

Introduction

Defined as scoliosis with no known cause in patients presenting between 10 and 18 years of age, adolescent idiopathic scoliosis (AIS) is the most common spinal deformity among children. The incidence of the disease is previously reported between 0.47% and 5.2% [1–4]. However, methodology and inclusion criteria vary across studies [2, 3, 5, 6]. Factors such as access to medical care, age at presentation, and specific patient population genetics may also affect incidence [7].

The most recent paper reporting the population-based incidence of pediatric AIS in the United States was published in 1978 [8]. Additionally, there are very few studies reporting how incidence rates have changed over time. One study from Sweden reported a peak in scoliosis rates in the 1970s followed by lower, but stable rates in the 1980s and 1990s [9]. Such studies are useful to track the temporal trends in AIS.

Further, modern data on the incidence of scoliosis in U.S. adolescents may provide further insight regarding the role for scoliosis screening, which is currently debated [2–4, 6, 10, 11]. While many U.S school districts adopted a practice of universal scoliosis school screening, budgetary pressures and the lack of data proving the efficacy of brace treatment led to the discontinuation of school-based scoliosis screening programs in many regions. Recent studies conclusively demonstrate that bracing is an effective method of preventing curve progression and reducing risk of needing surgery; thus, there is renewed interest in the value of school screening for scoliosis [12, 13].

Our study evaluates the population-based incidence of scoliosis over a 20-year period (1994–2013) in a single Midwestern U.S. county using a comprehensive database. In 2004, 4 out of 5 public school districts including the main population center and all private schools in the county discontinued school scoliosis screening. We used this year as a cut-off point to compare incidence rates of scoliosis from 1994–2003, during school screening, and 2004–2013, after school screening was discontinued.

Thus, the primary purpose of this study is to determine trends in incidence of AIS. Additionally, we aim to observe any change in curve magnitude stratified by > 10° and >20° at diagnosis and AIS treatment prescribed at the first specialty visit. We hypothesized that incidence rates would not change significantly over the study period despite the discontinuation of school screening.

Materials and Methods

This study was designed as a retrospective, population-based study in Olmsted County, Minnesota using the resources of the Rochester Epidemiology Project. IRB approval was obtained for all aspects of this study. According to 2010 Olmsted County census, the total population is 144,248, and 86% of individuals are white. The socioeconomic characteristics of the source population are similar to US whites generally, except that a higher proportion of residents are more highly educated [14]. The Rochester Epidemiology Project is a diagnostic indexing and medical records linkage system and its potential for population-based studies has been described previously in detail [15, 16]. Population-based, epidemiologic research in this community is enhanced due to relative geographic isolation from urban centers. In addition, nearly all medical care is delivered to local residents by a small number of health care providers, including all medical providers in the county. All medical and surgical diagnoses and other key information are abstracted and entered into computerized indices to facilitate case identification. This unique population-based data resource makes it unlikely that diagnosis and follow-up of clinically diagnosed cases of AIS in the county occurred outside of the geographic community. Furthermore, data from all providers are contained in a single, comprehensive medical records linkage system. Availability of the original medical records, including all radiographs allows identification and characterization of all AIS cases in this population.

Using the resources of the Rochester Epidemiology Project, we identified all adolescents (10–18 years) with a diagnosis suggestive of AIS from January 1, 1994 to December 31, 2013. The complete medical records (from all healthcare providers) were identified and reviewed using a standardized data abstraction form. Data collection included first date seen for scoliosis, gender, age, curve magnitude at first visit, radiographic data if radiographs were taken, and any treatment prescribed at the first visit. Patients with a neuromuscular or syndromic diagnosis were excluded, as were patients presenting prior to age 10 or after age 18 and patients with congenital scoliosis. Patients who moved to the county area with known pre-existing scoliosis (i.e., prevalent cases) and treatment already provided at another center were similarly excluded. Radiographs if performed by treating provider were reviewed to confirm AIS diagnosis and verify treatment plan. Initial prescribed treatment was recorded, including bracing, surgery, observation, or no further follow-up.

In 2004, a formal school screening program run by the county health department was discontinued. Prior to 2004, screening for scoliosis throughout the county was performed by a school public health nurse (LPN level or higher). Students were screened in grades 5, 7, and 8 (approximately age 11, 13 and 14). If the student’s examination suggested scoliosis but did not meet screening parameters (such as a scoliometer reading greater than 6 degrees), a second nurse examined the child. If scoliosis criteria were met, the patient’s parent was called and a letter was sent. The school requested communication as to whether the scoliosis had been assessed by a medical professional. Due to budgetary constraints, screening in 4 out of 5 public school districts and all the private schools was discontinued in 2004. The remaining public school district did not have complete records, but stated that no screening had been done after 2007. Thus, county patients with a new diagnosis between 1994 and August 2004 were considered to have taken place during school screening. Patients presenting with a new diagnosis between September 2004 and September 2014 were considered to present after school screening.

During the study period, a pediatric orthopaedic practice with little employee turnover existed in the county with four surgeons, several nurse practitioners, and one physician assistant providing consistent care. No other orthopaedic surgeons or neurosurgeons are known to offer pediatric orthopaedic care within a 70 mile radius. Standards for initiating bracing (skeletally immaturity, curve > 20–25 degrees), and surgery (curve > 45–50 degrees for skeletally immature patient and > 50 degrees for skeletally mature patient) were consistent over the study period.

We also assessed whether other healthcare providers were managing scoliosis bracing or surgery in the region. Within a 70 mile radius, there was only one orthotics office which manufactured scoliosis braces during the study period. They had not received a prescription for a scoliosis brace for anyone other than from the pediatric orthopaedic providers included in this analysis. We surveyed 13 regional chiropractors to see if they had prescribed braces for adolescent idiopathic scoliosis treatment. Many stated that they would treat children with scoliosis but would refer for moderate or severe scoliosis and would not prescribe a brace.

Previous work has confirmed that 99% of county patients receive all medical care at centers captured in the database and that this county’s experience provides an accurate estimation for incidence rates of most diseases in the United States [15]. A survey of county residents indicated that 95% or more receive their medical care at one of the two health care providers, making it feasible to track trends in disease occurrence and treatment initiation [17].

Age- and sex-specific incidence rates were calculated by using the number of incident AIS cases as the numerator and population estimates based on decennial census counts as the denominator with linear interpolation between census years. Only AIS subjects who were residents at initial AIS diagnosis were included in the incidence calculations. Overall incidence rates were age- and sex-adjusted to the 2010 United States population. Poisson regression models were used to examine incidence trends by age, sex, and calendar period. Ninety-five percent confidence intervals (95% CIs) for the incidence rates were constructed using the assumption that the number of incident cases per year follows a Poisson distribution.

Results

Between 1994 and 2014, 1782 adolescents (aged 10–18 years) had a new diagnosis of AIS made by a medical practitioner. The highest incidence was among males and females between ages 13–15. The overall age and sex-adjusted annual incidence of a new AIS diagnosis was 522.5 (95% CI, 498.2, 546.8) per 100,000 person-years, or 0.52% (Table 1). Incidence was higher in females than males (717.7 versus 336.5 per 100, 000, p<0.05, Table 1). Incidence of a new diagnosis of scoliosis decreased after discontinuation of school screening (Table 2). This was particularly significant in females age 10–15, and males age 13–15 years of age, which corresponds to age at which screening occurred in the schools and the sex-specific adolescent growth spurt.

Table 1.

Incidence of Scoliosis Diagnosis in Olmsted County^{^}

	Number of Cases			Incidence Rate (per 100,000)
Age group	Female	Male	Total	Female	Male	Total
10–12	427	99	526	746.7	171.1	457.2
13–15	630	340	970	1150.2	588.2	861.7
16–18	147	139	286	280.2	252.9	266.2
Total (95% CI)	1204	578	1782	732.3 (691.5–774.9)	338.8 (311.7–367.5)	531.9 (507.5–557.2)
Total (95% CI)	1204	578	1782	717.7 (677.1–758.2)*	336.5 (309.0–363.9)*	522.5 (498.2,546.8)†

^{^}Include 10 years with school screening present and 10 years with school screening absent.

^*Age-adjusted to the US Total 2010 population

^†Age- and sex-adjusted to the US Total 2010 population

Table 2.

Trends in Age- and Sex-specific Incidence of Scoliosis Diagnosis

		Time Period
		1994–2003 (School Screening)		2004–2013 (No School Screening)
Gender	Age Group	Number	Incidence Rate	Number	Incidence Rate	p-value
Male	10–12	52	182.5	47	160.0	0.514
	13–15	208	740.8	132	444.1	<0.001
	16–18	80	299.2	59	209.1	0.037
	Total	340	408.1	238	272.6	<0.001
Female	10–12	251	891.0	176	606.7	<0.001
	13–15	380	1428.7	250	887.3	<0.001
	16–18	68	266.5	79	293.3	0.563
	Total	699	870.6	505	600.3	<0.001

Of the 1782 AIS patients, 1408 (89%) had radiographs taken to confirm the diagnosis in addition to a physical exam (Table 3). The incidence of a radiograph showing a major curve greater than 10° was 181.7 (95% CI) per 100,000 person-years (Table 4). Notably, the incidence of major curve >10° decreased in number over the years (215.6 vs 149.4, Figure 1), as did the incidence of bracing and surgery at initial diagnosis (Table 4). Although the incidence of recommendation for fusion surgery for scoliosis at initial visit dropped from 2.9 to 1.1 cases per 100,000 and incidence of bracing decreased from 20.5 to 12.9 per 100,000, again, these trends did not reach statistical significance.

Table 3.

Scoliosis with curve > 10° ^*

		Time Period
		1994–2003		2004–2013
Gender	Age	n/N†	%	n/N†	%	p-value
Male	10–12	11/44	25.0%	9/36	25.0%	>0.999
	13–15	49/151	32.5%	34/104	32.7%	0.968
	16–18	13/50	26.0%	12/45	26.7%	0.941
	Total	73/245	29.8%	55/185	29.7%	0.988
Female	10–12	107/206	51.9%	67/143	46.9%	0.351
	13–15	161/304	53.0%	118/218	54.1%	0.792
	16–18	21/49	42.9%	22/58	37.9%	0.605
	Total	289/559	51.7%	207/419	49.4%	0.477
Overall	10–12	118/250	47.2%	76/179	42.5%	0.331
	13–15	210/455	46.2%	152/322	47.2%	0.772
	16–18	34/99	34.3%	34/103	33.0%	0.841
	Total	362/804	45.0%	262/604	43.4%	0.538

^*Among the 1,408 Scoliosis patients with x-ray confirmation of curve > 10°

^†n is the number of patients with curve < 10°, and N is number of patients in this group with a spine x-ray.

Table 4.

Incidence of Scoliosis in Olmsted County age 10–18, in 1994–2003^*

	Overall	1994–2003	2004–2013	p-value
Any Scoliosis	524.1 (499.8, 548.5)	623.5 (585.6, 661.5)	426.5 (395.8, 457.2)	P<0.001
Initial Bracing	16.6 (12.3, 20.9)	20.5 (13.7,27.3)	12.9 (7.6,18.1)	P>0.05
Initial Surgery	2.0 (0.5,3.4)	2.9 (0.4,5.4)	1.1 (0,2.7)	P>0.05
Curve > 10°	181.7 (167.5,196.0)	215.6 (193.4,237.8)	149.4 (131.3,167.6)	P<0.001
Curve > 20°		85.5 (71.5, 99.5)	57.3 (46.1,68.6)	P=0.0002

^*Rates expressed per 100,000, age- and sex-adjusted to US Total 2010 population, with 95% CI

Figure 1.

Incidence of scoliosis in males and females between 1994 and 2013. County-wide school screening was discontinued in 2004.

Discussion

Our study found that the average yearly incidence of AIS (new cases per year) was around 1 in 200, or 0.5%, which is similar to incidence rates reported in studies in Japan, Brazil, China, Turkey, Greece and Korea (<2%) [18–23]. Scoliosis is more common in girls compared to boys. The incidence of scoliosis may differ by patient population, genetics, environmental factors, access to healthcare, vitamin D, sex, age, and threshold for curve diagnosis [1, 24] [20, 24–27].

Other frequently cited studies report on prevalence, or number of patients with scoliosis in the entire population at a single time point. Prevalence is equivalent to incidence of a condition times the duration of the disease. As scoliosis is for the most part a lifelong condition, terminology regarding incidence and prevalence have been used loosely at times in the orthopaedic literature. In Tokyo, the prevalence of a 10° curve in children aged 7–15 years was 0.25% for boys and 1.77% for girls [19]. In a 2011 follow-up study, researchers evaluated 255,875 children aged 11–14, and found the overall prevalence of 0.87% for of a 10° scoliotic curve was (1.60% in girls, 0.14% in boys). Thus, the overall prevalence was stable over time. Other studies show increasing prevalence of scoliosis over time [3, 28, 29]. Hong Kong researchers evaluated 394,301 students enrolled in 5^th grade from 1995–2000. For curves >10°, prevalence rose from 2.3% in 1995 to 4.7% in 2000, and the prevalence of curves >20° increased from 1.3% to 2.2%. The authors posit that increased urbanization has been tied to greater prevalence of AIS [30, 31]. They have 78% participation rate for a completely volunteer-participation screening program embedded in the school system where students are followed until age 19 [32]. Compared to community screening programs, prevalence of scoliosis in a spine referral practice is much higher. The rate of scoliosis within a spine referral practice in a country without school screening showed that patients presented with a mean curve magnitude of 36° for females and 33° for males, and that 33% were eligible for brace treatment at presentation [4]. Similarly, a Canadian study consisting of 831 children aged 10–18 referred for a scoliosis consult in 5 pediatric orthopaedic clinics noted that 517 patients (62%) had curves greater than <10° [33].

The most recent paper on the U.S. rates of scoliosis was published by Rogala et al. in 1978. In that paper, 26,947 students were screened. Data were obtained for 1,122 of the total group of students. The prevalence was 4.5% [8]. A 1976 study in Minnesota reported a statewide prevalence rate of 4.0% [34]. As expected, the incidence rates reported in our study are lower than the prevalence rates reported in 1978 and 1976, at only 426.5 new cases per 100,000 individuals per year, or around 0.4%.

There are several limitations to our study. 374 patients with AIS did not have a radiograph available for review and were diagnosed clinically. It is possible that these patients were followed clinically by their primary care provider, and it was never deemed necessary to see an orthopaedic surgeon for evaluation. This could also indicate that radiographs were not taken due to only mild findings on physical exam, and that the majority of these patients had only spinal asymmetry. This group could include a large number of children who had curves over 10 or even 20 degrees. However, it is not possible to ascertain the reasoning that this group of patients diagnosed with AIS did not get a radiograph. For this reason, we have reported both overall incidence of scoliosis diagnosis as well as incidence of radiologically confirmed scoliosis and scoliosis potentially requiring brace treatment (curve ≥ 10º and ≥ 20º). In addition, the size of our patient population is significantly smaller than several of the cited studies, and thus our study may be underpowered to detect a difference in rates of surgery and bracing at initial presentation. There was one school that still had some school screenings from 2004–2007 that could potentially skew our results, though it is impossible to know how many children included in our study would have been screened at that school.

Our study provides a population-based estimate for the incidence of scoliosis in a pediatric population. This database has been used to produce estimates for incidence of a variety of medical conditions, such as breast cancer, hip fractures, and cardiovascular disease with similar results to nationwide databases (35–37). However, there are some differences from the study population and the general U.S. population, including decreased ethnic diversity, higher levels of education, and increased median household income [14]. County residents of the county have more ready access to medical care than is typical in much of the world, which could provide results reflecting a higher incidence of scoliosis than other communities without ready access to healthcare which may have higher rates of undiagnosed disease. However, the database has been used to determine the incidence of other pediatric orthopedic conditions including SCFE [38], ACL tear [39], tarsal coalition [40] and osteochondritis dissecans [41]. Further, our study evaluates the incidence of scoliosis over time.

In summary, our study showed that the population-based incidence of scoliosis diagnosis was 522.5 per 100,000 individuals. Interestingly, the overall incidence as well as incidence of curves > 10° decreased after ending school screening. The reported incidence prior to 2004 is comprehensive, with likely all children in the county undergoing nurse screening. The reported incidence after 2004 represents patients whose scoliosis was detected typically either at home or by a primary care provider. Likely, fewer patients with mild scoliosis were diagnosed after 2004 due to the discontinuation of school screening. Because our study is underpowered, it is unclear whether some children with severe scoliosis have not been diagnosed due to the discontinuation of school screening. The incidence of bracing and surgery at initial presentation decreased after discontinuing school screening, but this was not a statistically significant result. The strength of this study is that it is a U.S. population-based study regarding the incidence of scoliosis, which has not been reported upon in several decades. Updated data regarding the incidence of scoliosis in a modern U.S. population may help determine the role of broader scoliosis screening programs.