National Survey of Children’s Health

NSCH at a Glance

Iris Registry
National Survey of Children’s Health
Data type National self-report survey
Sample Nationally and State representative ample of children aged birth to 17
VEHSS Topics included Visual Function
Years Analyzed 2016-17 (merged)
Approximate size 95,000 people

The National Survey of Children’s Health (NSCH), sponsored by the Maternal and Child Health Bureau of the Health Resources and Services Administration, examines the physical and emotional health of children from birth to 17 years. The NSCH was designed to produce nationally- and state-representative estimates. NSCH was selected for inclusion in VEHSS in order to capture demographic and state-level variation of visual health among children, a population that may be underrepresented in other data sources selected for inclusion in VEHSS.

Sample Design

The NSCH went under a redesign in 2015 for the 2016 Survey and has since been conducted by the U.S. Census Bureau. The redesign was a result of declining response rates coupled with the fact that fewer households have landline telephones. The underlying sampling frame went from using the State and Local Area Integrated Telephone Survey (SLAITS) sampling frame based on telephone numbers originally developed for the National Immunization Survey (NIS)[1] to household addresses. The 2016 and 2017 NSCH used an address-based sample covering the 50 states and the District of Columbia. Household addresses were randomly drawn from the Census Master Address File (MAF). The sample file was selected from the Census MAF and supplemented with an administrative records-based flag identifying households likely to include children. This child-presence indicator allowed the Census Bureau to oversample households that were more likely to have children. The 2016 sample was developed to select a roughly equal number of addresses within each state and the 2017 sample was developed to produce roughly equal number of responses by state. In households with both children with special health care needs (CSHCN) and non-CSHCN, there was an 80% oversample of CSHCNs. A 60% oversample of young children, ages 0-5, was applied in households with either all or no CSHCNs.

Data Collection Procedures

Under the new design the survey uses a two‐phase multimode data collection design that combines the former NSCH and the National Survey of CSHCN (NS‐CSHCN) into the NSCH. It now consists of two questionnaires: (1) a household screener to determine whether the household includes a child(ren) and to select the target child within the household, and (2) a topical questionnaire tailored to three age groups ― 0-5 years, 6-11 years, and 12-17 years. The respondent is a parent or guardian who knew about the child’s health and health care needs.

The process differed a little between the 2016 and 2017 surveys. In general, all sampled addresses received an initial invitation letter with instructions to participate by web. After a certain point if a household did not complete their survey they were mailed a follow‐up letter. This letter again included instructions for responding via web. Addresses with the lowest probability of completing a web survey received a paper screener with the second follow‐up mailing (for 2017, paper screeners were sent with the initial mailing). All nonresponding addresses received paper screeners in the third and subsequent follow‐up mailings. Those who responded via web completed both the screener and topical questionnaires in one instrument, while those who returned the paper screener via mail were sent a paper topical instrument to complete. The NSCH paper and web instruments were available in both English and Spanish. Data were collected from June 2016 through February 2017 and from August 2017 through February 2018 for the 2016 and 2017 Surveys, respectively.

Analysis Overview

We estimated the prevalence rate and sample size for each survey instrument selected for inclusion.  We merged samples from the 2016 and 2017 rounds for analysis in order to maximize the available sample sizes at more detailed levels of stratification. We did not include data prior to 2016 because the different modes of data collection and the different sampling frame make the data not compatible.

For binary response questions included in the analysis, prevalence rate was defined as the number of persons who gave an affirmative response to the question divided by the total number of respondents who gave an affirmative or negative response and then multiplied by 100 for presentation in percentage format. For questions with a mark all that apply option, the data value is the proportion of respondents that selected that response option divided by the total number of respondents who either marked or did not mark that response option; and all responses may not sum to 100%. We estimated upper and lower confidence intervals and the relative standard error of the prevalence estimate using the Clopper-Pearson method with the smaller of the effective sample size and the sample size. The respondent sample size was reported for each response.

All estimates were calculated using SAS® PROC SURVEY FREQ procedure. Suppression was determined using the National Center for Health Statistics Data Presentation Standards for Proportions released in August 2017[1]

A detailed description of the analytical steps is described in the report “VEHSS Survey Analysis Plan pdf icon[PDF – 480 KB]external icon.”  Full analysis documentation is included in the “VEHSS NSCH Data Report pdf icon[PDF – 592 KB]external icon.”

Variables Analyzed in VEHSS

Starting in 2016 the NSCH contains four questions related to VEHSS indicators and were included for analysis.  These questions were fielded in both the 2016 and 2017 surveys. One question was categorized under the ‘Visual Function’ Topic and ‘Difficulty Seeing with Glasses’ Category. The other 3 questions were under the ‘Service Utilization’ Topic and covered ‘Need’, ‘Screening’, and ‘Provider Type’ categories. Two of the questions were sub-questions and were asked if the respondent responded Yes to the main question. Table 1 presents these four questions and includes the VEHSS Topic and Category, the NSCH variable name, the year(s) survey data are available, the survey question, and the response options.

Table 1. Overview of NSCH Variables Included in VEHSS

VEHSS Indicator Topic VEHSS Indicator Category NSCH Variable Name Years Available
Question Response Options Included
NSCH eye health questions and response options
Visual Function Difficulty Seeing with Glasses Blindness 2016, 2017 Does this child have any of the following? Blindness or problems with seeing, even when wearing glasses 1 Yes
2 No
Service Utilization Need K4Q27, K4Q28X03 2016, 2017 During the past 12 months, was there any time when this child needed health care but it was not received? By health care, we mean medical care as well as other kinds of care like dental care, vision care, and mental health services. If yes, which types of care were not received? Mark ALL that apply. 1 Medical Care
2 Dental Care
3 Vision Care
4 Hearing Care
5 Mental Health Services
6 Other, specify
L Legitimate skip
M Missing in error
Service Utilization Screening K4Q31_R 2016, 2017 Has this child ever (0-5 years)/ During the past 2 years (6-17 years) had his or her vision tested with pictures, shapes, or letters? 1 Yes
2 No
Service Utilization Provider Type






2016, 2017 If yes, what kind of place or places did this child have his or her vision tested? Mark ALL that apply. 1 Eye doctor or eye specialist
2 Pediatrician or other general doctor’s office
3 Clinic or health center
4 School
5 Other, specify
L Legitimate skip
M Missing in error

The four questions included in the analysis come from Section A: This Child’s Health (visual function question) and Section C: Health Care Services (service utilization questions) for both the 2016 and 2017 NSCH.

The 2016 and 2017 surveys used the same question wording for all four questions. The two surveys were merged and new weights were created by dividing the original weights by 2 according to survey guidance [3].

Outcome variables for all questions were coded such that bivariate responses were coded into Yes/Selected, No/Not Selected, and Missing (including Legitimate skip and Missing in error), although only the prevalence of Yes/Selected responses are reported below. The sample population for analysis (i.e., denominator) was also limited to those coded as Yes or No. Though K4Q27 was not analyzed, the frequencies to K4Q27 ― During the past 12 months, was there any time when this child needed health care but it was not received? By health care, we mean medical care as well as other kinds of care like dental care, vision care, and mental health services. ― were used to determine the sample who should have responded to K4Q28X03― If yes, which types of care were not received? ―selected Vision Care.

Tables 2a, b, and c present the sample size for analysis by coded response option for the eight eye health variables. As demonstrated by the table, K4Q28X03 is one of the follow-up response options to those respondents who said Yes to K4Q27. K4Q23X01 – K4Q23X05 are the possible response options to a follow-up question to those respondents who said Yes to K4Q31_R.

Tables 2a-c. Frequency of coded response options for vison-related questions

Sample Size Variable – Blindness
Table 2a
Yes 894
No 70,582
Missing 335
Sample Size K4Q27 K4Q28X03
Table 2b
Yes 1,660 Yes – 290
No – 1,349
Missing – 21
No 69,897
Missing 254
Sample Size Variables
K4Q31_R K4Q32X01 K4Q32X02 K4Q32X03 K4Q32X04 K4Q32X05
Table 2c
Yes 51,929 Yes – 30,648
No – 21,072
Missing – 209
Yes – 20,542
No – 31,178
Missing – 209
Yes – 1,727
No – 49,993
Missing – 209
Yes – 11,792
No – 39,928
Missing – 209
Yes – 522
No – 51,198
Missing – 209
No 19,657
Missing 225
Table 3. Stratification variable frequencies
Variables Frequency
0 -17 years 71,811
50 States, D.C. 71,811
Male 36,800
Female 35,011
Hispanic 7,993
Non-Hispanic White 50,219
Non-Hispanic Black 4,236
Non-Hispanic Other 9,363

Stratification Levels Included in the Full Analysis

The full analysis includes additional stratifications beyond those included in this data brief report.  NSCH only includes one VEHSS-defined age group. Based on the sample sizes and rates of suppression, we included 0, 1, and 2-level stratification at the national level, but only included 0-level and 1-level stratification at the state level.  We found high rates of missing/suppression when attempting to analyze further levels of stratification. Table 4 reports the stratification levels included in the data table, which equate to the variables available for selection in the data visualization application.

National State
Table 4. Stratification factor combinations included in full results
0-level All participants All participants
1-level Race/ethnicity Race/ethnicity
Sex Sex
2-level Race/ethnicity*sex

The vision health-related questions in the NSCH have not been used by researchers in previously published papers. This is probably largely due to new vision health questions replacing the older ones in 2016. There was one abstract[1] at the Association for Research in Vision and Ophthalmology (ARVO) Annual Meeting that presented findings from the 2016 NSCH vision question on screening. Prior to 2018, the DRC and a paper[2] looking at the relationship between ADHD and children ages 4-17 years with vision problems were the only sources found that used the NSCH vision health questions. The question wording for blindness – “Does this child have any of the following? Blindness or problems with seeing, even when wearing glasses” – is nearly identical to the wording used in the ACS and BRFSS, though BRFSS does not include children.

The NSCH provides national- and state-level estimates of visual function, as reported by a parent or guardian. The level of detail of results is limited by sample size. At the state level, we were only able to identify prevalence estimates at a single level of stratification. We currently plan to include full stratification at the national level. Unlike other data sets included in VEHSS, NSCH only includes one VEHSS-defined age group, which means that full stratification of results is at the 2nd level (race*sex).

Diabetes was the only VEHSS-defined risk factor included in the survey. However, due to the relatively small number of children with diabetes, diabetes was not included in the analysis for NSCH as nearly all values were suppressed in the analysis of previous years of NSCH.

Additional information about NSCH can be found on the Data Resource Center for Child & Adolescent Healthexternal icon website and the Design and Operation of the National Survey of Children’s Health, 2007 pdf icon[PDF – 640 KB] Vital and Health Statistics report.

See the VEHSS NSCH Data Analysis report for more details on our analysis and validation of the VEHSS analysis of the NSCH.

[1] Bramlett MD, B. S. (2017, August 8). Design and operation of the National Survey of Children’s Health, 2011–2012. Hyattsville, Maryland: National Center for Health. Retrieved from National Center for Health Statistics:

[2] Parker JD, Talih M, Malec DJ, et al. National Center for Health Statistics Data Presentation Standards for Proportions. National Center for Health Statistics. Vital Health Stat 2(175). 2017.


[4] Sandra S Block, Kira Baldonado; Results from 2016 National Survey of Children’s Health (NSCH). Invest. Ophthalmol. Vis. Sci. 2018;59 (9):163.

[5] DeCarlo DK, Swanson M, McGwin G, Visscher K, Owsley C. ADHD and Vision Problems in the National Survey of Children’s Health. Optometry and vision science : official publication of the American Academy of Optometry. 2016;93(5):459-465. doi:10.1097/OPX.0000000000000823.