Characteristics of young children with developmental delays and their trends over 14 years in Taiwan: a population-based nationwide study.

OBJECTIVES: To describe the epidemiological characteristics of developmental delays (DD) at the national level and assess the changes over time.
DESIGN: We calculated the incidence rates of DD from 2003 to 2016 and assessed the trends over time.
SETTING: As dictated by law, local governments in Taiwan are required to register children with DD and provide services. The central government has constructed a national registry with the data from local centres. We analysed the national registry data. PARTICIPANTS: We included children who were under 6 years old, and this population ranged from 1 164 150 to 1 577 443 per year during the study period. All registered cases were certified through a process set forth by law. PRIMARY AND SECONDARY OUTCOME MEASURES: We calculated annual incidence rates by age, sex and geographical area, and assessed trends over the study period.
RESULTS: The incidence of DD in children under 6 years old displayed an increasing trend over the study period, ranging from 7.0 to 16.3 per 1000 person-years. Boys had higher incidence throughout all 14 years, and the boy-to-girl rate ratios had an increasing trend over time with some fluctuations, ranging from 1.84 (95% CI 1.77 to 1.92) to 1.99 (95% CI 1.93 to 2.06). In addition, rural areas had higher incidence rates, and the rural to urban rate ratios ranged from 0.98 (95% CI 0.94 to 1.03) to 2.00 (95% CI 1.94 to 2.06) without apparent time trends. Girls had a higher proportion of early reporting (<3 years) throughout all years, but the differences in the proportion of early reporting between rural and urban areas were not consistent.
CONCLUSIONS: Male sex appeared to be a risk factor for DD, which is unlikely to be due to more attention received by boys because girls had a higher proportion of early reporting. We also found that children lived in rural areas appeared to have a higher risk of having DD. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

The Taiwanese registry of children with developmental delays includes only new cases, which provides incidence data that are rarely available at the national level.

The law in Taiwan mandates the reporting of cases, which minimises under-reporting.

Using nationwide registry data, this study has a large sample size, with a population of more than one million each year.

The duration of data collection in this study was 14 years, and the assessment of time trends over such a long period of time at the national level has rarely been achieved in previous studies.

The major limitation of our study is that the government does not release data on individual cases, which make it impossible to conduct the analysis in further detail.

Introduction

Developmental delays (DD) are generally defined as a child does not reach expected developmental milestones at the age.1 Although DD are prevalent among children, data on its incidence at the national level are limited. For the identification of risk factors, incidence data are generally more useful than prevalence data. However, the majority of previous studies on DD used prevalence data, most likely because incidence data were more difficult to obtain.2 As a result, changes in epidemiological characteristics over time are seldom assessed, because such studies would require incidence data.

A child’s development is affected by many biological and environmental factors.1 Among biological ones, male sex is a strong risk factor.3 Many studies found that the prevalence rate for DD was higher in boys than in girls.3 4 Also, residents of rural areas have reported more health-related problems than those of urban areas in general,5 and this disparity might also be true for DD. However, studies on this topic are limited and have reported inconsistent results. For example, the Multiple Indicator Cluster Survey conducted by the United Nations Children’s Fund in countries with low and middle incomes did not observe consistent results in terms of differences between rural and urban areas across various countries.6

Under the Children Welfare Law, local governments in Taiwan have established Early Intervention Reporting and Referral Centers (EIRRCs) to register and provide services to children with DD.7 The Children and Youth Welfare Law also mandates that institutions of welfare, education and medicine report all children with suspected DD to the EIRRC.8 Using data collected at the EIRRCs, the central government has constructed a national registry of children with DD. This registry provides researchers with a unique resource with which to conduct nationwide epidemiological studies. In a previous study, we analysed the national registry data from 2003 to 2008.2 However, since it is difficult to precisely evaluate the trends of incidence with only 6 years’ worth of data, we extended the time range from 2003 to 2008 to a 14-year period until 2016 as more data became available. This better suited our purposes of studying epidemiological characteristics of incident cases of DD and assessing time trends.

Methods

Case definition

The Enforcement Rules of the Children and Youth Welfare Law of Taiwan9 defined DD as ‘allegedly or expected abnormal development in respect of cognitive development, physiological development, language and communication development, psycho-social development or self-governing skills that have been judged and confirmed by the accredited medical institutes under health authority.’ In order to be registered as a case of DD, the child requires an assessment performed by a physician at an accredited hospital or by a trained social worker in an EIRRC,10 who generally used the Taipei City Developmental Checklist for Preschoolers, Second Version as the assessment tool.11

Data collection

The EIRRCs collect data on newly reported cases and report such cases to the central government, which maintains a registry of cases and publishes a yearbook each year. Before the reorganisation of the government in 2013, the registry was maintained by the Ministry of the Interior. After 2013, this responsibility fell to the Ministry of Health and Welfare. This yearbook includes the total numbers of newly reported cases by age at reporting, sex and area.12 We analysed the data from 2003 to 2016.

We divided the cases into two groups, <3 years and 3–5 years, according to the age at the time of reporting and defined the group <3 years as ‘early reporting’.13 Although the yearbook includes statistics on an age group ‘≥6 years,’ we did not include this group in our analyses because it only includes children who have not yet entered the primary school, not all children who are 6 years of age or older.

In the stratified analyses by geographical area, we divided the administrative regions in Taiwan into two groups: ‘urban areas’ and ‘rural areas.’ An urban area is one in which more than 50% of the population live in metropolitan areas (as defined by the Directorate-General of Budget, Accounting and Statistics of Taiwan).14 Consequently, a rural area is one in which 50% or less of the population live in metropolitan areas. In a reorganisation of administrative regions in 2011, the original Tainan City (an urban area) and Tainan County (a rural area) were merged into the new Tainan City. Data on the two original administrative regions have been unavailable since this merging, so they were both excluded from this part of the analyses.

To calculate incidence rates, we obtained data on the numbers of boys, girls and the total population in each age group for each administrative region from the Monthly Bulletin of Interior Statistics.15

Data analysis

We estimated age-specific incidence rates in a given year by dividing the number of newly reported cases by the number of individuals in a specific age group. Likewise, the sex-specific incidence rates of each year were estimated by dividing the number of newly reported cases by the number of individuals of each sex. We evaluated the differences between the two sexes by using the incidence rate ratio calculated through dividing the incidence rate in boys by the incidence rate in girls. To evaluate the geographical differences, we estimated the incidence rate in rural or urban areas in each year by first dividing the number of newly reported cases by the number of individuals in each type of area and then obtaining the incidence rate ratio by dividing the incidence rate in rural areas by the incidence rate in urban areas. A 95% CI was constructed for each rate ratio to evaluate its statistical significance.

To evaluate the difference in early reporting between boys and girls, we calculated the proportion of early reporting in boys or girls each year by dividing the number of cases under 3 years of age by the number of cases under 6 years of age in each sex. We then obtained the boy-to-girl proportion ratio of early reporting by dividing the proportion in boys by the proportion in girls. A 95% CI was constructed for each proportion ratio to evaluate its statistical significance.

Likewise, to evaluate the difference of early reporting between rural and urban areas, we calculated the proportion of early reporting in rural or urban areas in each year by dividing the number of cases under 3 years of age by the number of cases under 6 years of age in each area. We then obtained the rural to urban proportion ratio of early reporting by dividing the proportion in rural areas by the proportion in urban areas. A 95% CI was constructed for each proportion ratio to evaluate its statistical significance.

We reported descriptive statistics as numbers or percentages and used linear regressions to evaluate the trends of changes in incidence rates, incidence rate ratios and proportion ratios over time. All the analyses were conducted using SAS V.9.1, and all the statistical tests were performed at the two-tailed significance level of 0.05.

Results

During the study period of 2003–2016, 211 661 new cases of DD under 6 years of age were registered, and the overall incidence rate was from 7.0 to 16.3 per 1000 person-years. The rate was from 5.7 to 15.3 per 1000 person-years in children under 3 years of age and 7.9 to 18.0 per 1000 person-years in children 3–5 years of age. We observed a general increasing trend over time in all three age groups (p<0.01 for all) (table 1).

Table 1

Overall incidence rate of developmental delays in children under 6 years of age

Year	Reported number of cases		Population		Incidence (per 1000 person-years)
	<3 years	3–5 years	<3 years	3–5 years	<3 years	3–5 years	<6 years
2003	4139	7557	720 897	856 546	5.7	8.8	7.4
2004	4058	6682	680 737	846 130	6.0	7.9	7.0
2005	4512	6672	641 095	809 663	7.0	8.2	7.7
2006	5098	6798	619 354	730 819	8.2	9.3	8.8
2007	6201	7273	606 840	692 164	10.2	10.5	10.4
2008	6655	7439	599 674	654 179	11.1	11.4	11.2
2009	6103	8789	587 219	633 676	10.4	13.9	12.2
2010	6957	9122	551 334	621 318	12.6	14.7	13.7
2011	5850	8816	551 707	612 443	10.6	14.4	12.6
2012	6270	9921	589 053	600 984	10.6	16.5	13.6
2013	7474	9644	622 140	564 244	12.0	17.1	14.4
2014	9618	9711	634 646	566 056	15.2	17.2	16.1
2015	9372	10 319	613 735	608 119	15.3	17.0	16.1
2016	9160	11 451	627 470	636 735	14.6	18.0	16.3

With data on 14 years, we were able to compare the incidence among four mutually exclusive birth cohorts. Using the 2001–2003 cohort (those who were in the <3 years group in 2003) as the baseline, we observed an increasing trend in the estimated overall incidence rates in children under 6 years old. Specifically, the estimated overall incidence rate before 6 years of age was 7.5 per 1000 in the 2001–2003 birth cohort, 11.1 per 1000 in the 2004–2006 birth cohort, 13.5 per 1000 in the 2007–2009 birth cohort and 13.9 per 1000 in the 2010–2012 birth cohort. The trend was similar to that observed in the year-by-year overall incidence rates.

The overall incidence rate ranged from 9.0 to 21.3 per 1000 person-years in boys and from 4.9 to 11.0 per 1000 person-years in girls. The rate was constantly higher in boys than in girls, and the boy-to-girl rate ratios had an increasing trend over time (p<0.01) with some fluctuations, ranging from 1.84 to 1.99 (table 2). All the rate ratios were statistically significant.

Table 2

Overall incidence rate and rate ratio of developmental delays in children under 6 years of age by sex

Year	Reported number of cases		Population		Incidence (per 1000 person-years)		Rate ratio (95% CI)
	Boy	Girl	Boy	Girl	Boy	Girl
2003	7837	3859	823 435	754 008	9.5	5.1	1.86 (1.79 to 1.93)*
2004	7183	3557	797 949	728 918	9.0	4.9	1.84 (1.77 to 1.92)*
2005	7550	3634	758 014	692 744	10.0	5.2	1.90 (1.83 to 1.98)*
2006	7998	3898	705 549	644 624	11.3	6.0	1.87 (1.80 to 1.95)*
2007	9053	4421	679 395	619 609	13.3	7.1	1.87 (1.80 to 1.94)*
2008	9463	4631	655 736	598 117	14.4	7.7	1.86 (1.80 to 1.93)*
2009	9981	4911	637 773	583 122	15.6	8.4	1.86 (1.80 to 1.92)*
2010	10 968	5111	612 072	560 580	17.9	9.1	1.97 (1.90 to 2.03)*
2011	9969	4697	607 034	557 116	16.4	8.4	1.95 (1.88 to 2.02)*
2012	11 069	5122	619 396	570 641	17.9	9.0	1.99 (1.93 to 2.06)*
2013	11 578	5540	616 451	569 933	18.8	9.7	1.93 (1.87 to 1.99)*
2014	13 122	6207	622 626	578 076	21.1	10.7	1.96 (1.90 to 2.02)*
2015	13 419	6272	633 460	588 394	21.2	10.7	1.99 (1.93 to 2.05)*
2016	13 939	6672	654 986	609 219	21.3	11.0	1.94 (1.89 to 2.00)*

*P<0.05.

With the exclusion of Tainan City and County, there were 6 cities and 17 counties in Taiwan. All six cities and 5 of the 17 counties were classified as urban areas, and the remaining 12 counties were classified as rural areas. The overall incidence rate ranged from 7.2 to 25.5 per 1000 person-years in rural areas and from 6.1 to 14.7 per 1000 person-years in urban areas. The incidence rates were generally higher in rural than in urban areas, except for the year 2003, in which the rural to urban rate ratio was not statistically significant (table 3). In other years, the ratio fluctuated from 1.10 to 2.00 and all reached statistical significance. There was no apparent trend in the changes over time.

Table 3

Overall incidence rate and rate ratio of developmental delays in children under 6 years of age by area†

Year	Reported number of cases		Population		Incidence (per 1000 person-years)		Rate ratio (95% CI)
	Rural	Urban	Rural	Urban	Rural	Urban
2003	3016	7651	417 597	1 041 508	7.2	7.3	0.98 (0.94 to 1.03)
2004	3499	6161	401 360	1 010 972	8.7	6.1	1.43 (1.37 to 1.49)*
2005	3275	7031	379 954	962 324	8.6	7.3	1.18 (1.13 to 1.23)*
2006	3401	7580	350 937	898 852	9.7	8.4	1.15 (1.10 to 1.20)*
2007	3688	8794	331 562	871 125	11.1	10.1	1.10 (1.06 to 1.14)*
2008	4843	8177	312 880	847 893	15.5	9.6	1.61 (1.55 to 1.66)*
2009	4306	9515	298 839	831 537	14.4	11.4	1.26 (1.22 to 1.31)*
2010	5108	9881	280 108	806 265	18.2	12.3	1.49 (1.44 to 1.54)*
2011	5074	8779	268 296	808 844	18.9	10.9	1.74 (1.68 to 1.80)*
2012	5829	9336	266 682	833 418	21.9	11.2	1.95 (1.89 to 2.02)*
2013	5339	10 124	261 442	835 199	20.4	12.1	1.68 (1.63 to 1.74)*
2014	6629	10 838	259 828	850 302	25.5	12.7	2.00 (1.94 to 2.06)*
2015	5531	12 576	257 152	872 878	21.5	14.4	1.49 (1.45 to 1.54)*
2016	5591	13 365	261 055	908 526	21.4	14.7	1.46 (1.41 to 1.50)*

*P<0.05.

†Tainan City and Tainan County were excluded in the analyses due to a reorganisation of administrative regions in 2011.

The proportion of early reporting had an increased trend from 2003 to 2008 and then showed fluctuations afterwards (table 4). Girls had a higher proportion of early reporting than boys throughout all years, with the boy-to-girl proportion ratios ranging from 0.85 to 0.94. All the proportion ratios were statistically significant, but there was no apparent trend in the changes over time. The difference in the proportion of early reporting between rural and urban areas fluctuated over time. The rural to urban proportion ratios ranged from 0.79 to 1.19, and there was no apparent trend in the changes over the years (table 5).

Table 4

Proportion of early reporting of developmental delays in children under 6 years of age by sex

Year	Reported number of cases									Boy-to-girl proportion ratio of early reporting (95% CI)
	Total			Boy			Girl
	<3 years	<6 years	<3 years (%)	<3 years	<6 years	<3years (%)	<3 years	<6 years	<3 years (%)
2003	4139	11 696	35.4	2660	7837	33.9	1479	3859	38.3	0.89 (0.84 to 0.93)*
2004	4058	10 740	37.8	2648	7183	36.9	1410	3557	39.6	0.93 (0.88 to 0.98)*
2005	4512	11 184	40.3	2881	7550	38.2	1631	3634	44.9	0.85 (0.81 to 0.89)*
2006	5098	11 896	42.9	3321	7998	41.5	1777	3898	45.6	0.91 (0.87 to 0.95)*
2007	6201	13 474	46.0	3953	9053	43.7	2248	4421	50.8	0.86 (0.83 to 0.89)*
2008	6655	14 094	47.2	4310	9463	45.5	2345	4631	50.6	0.90 (0.87 to 0.93)*
2009	6103	14 892	41.0	3995	9981	40.0	2108	4911	42.9	0.93 (0.90 to 0.97)*
2010	6957	16 079	43.3	4627	10 968	42.2	2330	5111	45.6	0.93 (0.89 to 0.96)*
2011	5850	14 666	39.9	3896	9969	39.1	1954	4697	41.6	0.94 (0.90 to 0.98)*
2012	6270	16 191	38.7	4153	11 069	37.5	2117	5122	41.3	0.91 (0.87 to 0.95)*
2013	7474	17 118	43.7	4818	11 578	41.6	2656	5540	47.9	0.87 (0.84 to 0.90)*
2014	9618	19 329	49.8	6320	13 122	48.2	3298	6207	53.1	0.91 (0.88 to 0.93)*
2015	9372	19 691	47.6	6066	13 419	45.2	3306	6272	52.7	0.86 (0.83 to 0.88)*
2016	9160	20 611	44.4	5944	13 939	42.6	3216	6672	48.2	0.88 (0.86 to 0.91)*

*P<0.05.

Table 5

Proportion of early reporting of developmental delays in children under 6 years of age by area†

Year	Reported number of cases						Rural-to-urban proportion ratio of early reporting (95% CI)
	Rural			Urban
	<3 years	<6 years	<3 years (%)	<3 years	<6 years	<3 years (%)
2003	1159	3016	38.4	2475	7651	32.3	1.19 (1.12 to 1.26)*
2004	1215	3499	34.7	2393	6161	38.8	0.89 (0.85 to 0.94)*
2005	1140	3275	34.8	2959	7031	42.1	0.83 (0.78 to 0.87)*
2006	1353	3401	39.8	3415	7580	45.1	0.88 (0.84 to 0.93)*
2007	1423	3688	38.6	4294	8794	48.8	0.79 (0.75 to 0.83)*
2008	2472	4843	51.0	3578	8177	43.8	1.17 (1.12 to 1.21)*
2009	1845	4306	42.8	3646	9515	38.3	1.12 (1.07 to 1.17)*
2010	2211	5108	43.3	4152	9881	42.0	1.03 (0.99 to 1.07)
2011	1888	5074	37.2	3555	8779	40.5	0.92 (0.88 to 0.96)*
2012	2054	5829	35.2	3715	9336	39.8	0.89 (0.85 to 0.92)*
2013	2271	5339	42.5	4366	10 124	43.1	0.99 (0.95 to 1.02)
2014	3318	6629	50.1	5175	10 838	47.7	1.05 (1.02 to 1.08)*
2015	2772	5531	50.1	5712	12 576	45.4	1.10 (1.07 to 1.14)*
2016	2482	5591	44.4	5730	13 365	42.9	1.04 (1.00 to 1.07)

*P<0.05.

†Tainan City and Tainan County were excluded in the analyses due to a reorganisation of administrative regions in 2011.

Discussion

The overall incidence rate of DD in children under 6 years of age generally increased during the study period, which is consistent with the observation in our preliminary study.2 We believe the increase cannot be attributed to an increase in the number of cases, because the major risk factors for DD such as genetic constructs, medical care and socioeconomic status1 16 did not undergo remarkable changes over the 14 year period. In fact, there were no changes in the diagnostic and reporting criteria of DD, neither. Therefore, the increased incidence rate of DD is more likely to be due to the improvement in reporting of cases, which may be in turn attributable to the improvement in the awareness of the developmental abnormalities in children, lower discrimination against the patients, better service of the related agencies, more willingness demonstrated by parents and guardians to register their children and so on.17 18 These were generally gradual changes, and therefore we did not observe any abrupt changes in the incidence of DD over time, indicating the lack of an obvious period effect.

In addition to cases confirmed by physicians, the registry also includes cases identified by trained social workers in the EIRRCs. This would lead to overestimation of the incidence rates. However, the registry ascertains cases through reporting rather than thorough active screenings in the communities or schools, and therefore the candidates who receive the screening generally have shown certain symptoms or signs of DD which draw the attention of caregivers. Consequently, the proportion of confirmed cases in the registry is very high. Although the government never provides the statistics on the proportion of cases excluded after reporting, from our experience of running the EIRRC in the Chia-Yi City (an urban area) and Chia-Yi County (a rural area), we notice that the proportion of cases confirmed by physicians at reporting has been increasing. Therefore, the effect of overestimating the incidence should have been decreasing, and thus would not affect our conclusion of an increasing trend over time.

Since the critical period of treatment for DD is before the child reaches 3 years of age, an early diagnosis is crucial.19 Therefore, the Taiwanese government has been promoting the reporting and intervention in cases of children with DD under 3 years of age,2 13 when intervention procedures are most cost-effective. In fact, it is generally believed that the earlier the intervention, the better the prognosis.20 Although we observed an increased trend in early reporting from 2003 to 2008, the trend did not persist after 2008. This might indicate that the existing measures of promotion have reached their limits of effectiveness, additional efforts and approaches are required, especially since the proportion of early reporting has never risen beyond 50%.

Previously, we used the data from the national disability registry of Taiwan, a different nationwide source of information than that used in the current study, to conduct a series of studies on major developmental disabilities, including autism spectrum disorders (ASD), intellectual disability, and speech and language disability.17 18 21 We found that the prevalence rates were consistently higher in boys than in girls. Those developmental disabilities are all major risk factors for DD. Therefore, the finding of a higher incidence of DD in boys in the current study is compatible with the findings of our previous studies. Many theories have been proposed to explain the sex difference, including both biological and sociological factors. Biologically, the higher risk in boys is often attributed to hereditary factors such as X-linked conditions.22 23 The central nervous system of young boys is also found to be more susceptible to damage.22 In terms of social factors, in Asian countries, families generally display a male gender preference, and thus pay more attention to boys.22 23 In addition, boys are more likely to demonstrate impulsive and aggressive behaviours than girls and therefore are more likely to be referred for diagnosis.23

In this study, however, we found that girls had a higher proportion of early reporting than boys. This does not support the theory that the higher level of attention received by boys is a major factor contributing to the higher incidence rate of DD. In fact, a review of literature concluded that boys are more likely to have DD, but when such conditions arise in girls, a more severe form is usually manifest.24 Our finding of a higher proportion of early reporting in girls is compatible with this conclusion. Furthermore, as the birth rate has been decreasing in Taiwan,15 the difference in the attention received by boys and girls should be decreasing, but we found the sex difference had an increasing trend instead of a decreasing trend. Therefore, we believe biological factors contribute more than social factors to the sex difference in DD in early childhood.

In the preliminary study, we failed to identify the time trend in the sex difference (no significant trend in boy-to-girl rate ratio).2 The current study has the advantage of a longer follow-up period. In a separate study, which used data from the national disability registry, we found that there was a 497.30% increase in the prevalence rate of childhood ASD from 2000 to 2011 in Taiwan, which is much higher than the 54.40% increase in all reported disabilities combined.25 In addition, the boy-to-girl ratio of ASD was found to be the highest among the major conditions contributing to DD.4 17 Therefore, we believe the increasing trend in the boy-to-girl incidence rate ratio of DD in Taiwan can largely be attributable to the increasing trend in the number of cases of ASD.

Studies on the differences in DD between rural and urban areas are limited. There are relatively more data on childhood disabilities. The Multiple Indicator Cluster Survey conducted by the United Nations Children’s Fund in low/middle-income countries did not observe a consistent result.6 Munro reviewed several studies of intellectual disability in the USA and also found results to be inconsistent.23 A nationwide study of childhood disability in China, however, found a higher prevalence rate in rural areas,26 which is consistent with our finding on DD in this study. Moreover, according to the national disability registry of Taiwan, some major childhood disabilities leading to DD, including intellectual disability and speech and language disability,18 21 as well as all disabilities combined27 were more prevalent in rural areas. While the lower socioeconomic status of rural residents in Taiwan should be a major factor,28 29 the difference might also be explained by environmental factors such as lower quality of child care, lower parental education, and limited educational and cultural opportunities.5 22 30

The registry of children with DD in Taiwan has some unique features. First of all, it included new cases only, which provides incidence data that are rarely available at the national level and are more reliable for identifying risk factors.31 Second, Taiwanese law mandates that cases be reported, and therefore, instances of under-reporting are rare. The registry also has the advantage of a large number (more than 200 000) of cases. Lastly, the duration of data collection is long, which makes assessing time trends possible.

The major limitation of our study is that the government does not release data on individual cases, and therefore we were unable to conduct more detailed analyses. For example, while we believe the increasing trend in the boy-to-girl incidence rate ratio of DD can largely be attributable to the increasing trend in the number of cases of ASD, we could not perform analyses to verify this. Likewise, we were unable to adjust for the geographical difference while evaluating the effects of sex or age, or study the interaction between sex and geographical difference. In addition, the government does not provide data on the prevalence of DD, and therefore we were unable to study the prevalence of DD directly. Nonetheless, a study that used the National Health Insurance Research Database of Taiwan to estimate the prevalence of DD in children aged 0–6 years and found that the prevalence gradually increased from 0.16% to 3.25% from 1997 to 2008.32 Even though the case definition was not exactly the same as in the current study and the source of information was different, the study also observed an increasing trend in the annual prevalence and a higher risk in boys than in girls. Geographical differences were not evaluated in that study. Our study is also limited by the range of age under observation. A likely reason of observing a significant increasing trend in the incidence without a remarkable actual increase in the number of cases is that there was a trend of cases being diagnosed (and thus reported) at an earlier age. Although this hypothesis can be tested by evaluating if there was a decreasing trend in the order population, such as 6–12 years old, we did not include cases diagnosed later than 6 years old because the registry only includes cases who had not yet entered the primary school, not all the cases. In the current study, we adopted the ‘administrative’ data from a national registry which obtains the information on all persons receiving or qualifying for the services, and such an approach was found to generally result in underestimation because it does not include persons who do not receive or not qualify for the services.33 However, the administrative data we used have the advantage of providing information on incidence. Although incidence data are preferable to prevalence for investigating risk and aetiology, such data on DD are rare, and therefore epidemiology of DD has generally to be inferred from cross-sectional data.31 It has been proposed that incidence data need to be obtained by longitudinal studies from conception or birth to death.31 Whereas our study did not apply an active case ascertainment approach such as a house-by-house screening survey, it is a longitudinal study that followed the population from birth and thus can provide the incidence data that some researchers once believed to be unattainable in any population.31

Conclusions

The incidence rates of young children with DD have been increasing in Taiwan, which is likely to be attributable to improvements in the reporting of cases. The incidence rates are higher in boys than in girls, which is unlikely to be due to more attention received by boys because girls had a higher proportion of early reporting. In addition, the boy-to-girl rate ratio has an increasing trend over the years, and we believe the increasing number of cases of ASD is an important factor. Rural areas had higher incidence than urban areas, which calls for further studies to identify the contributing factors.