Recommendations by Age

Consider the age of respondents when selecting a measure.

Measures are developed and tested on specific age groups and respondents. PROMIS, for example, includes measures for:

Early Childhood Parent-Report (ages 1-5)
Pediatric self-report (ages 8-17)
Parent Proxy report (ages 5-17)
Adult self-report (ages 18+)

These measures are on different metrics. This means scores from the same domain (e.g., anxiety) but from measures developed for different age groups/respondents (e.g., pediatric and parent proxy) are NOT comparable. They should be analyzed separately. This can present a challenge for studies using longitudinal designs or studies sampling diverse age groups. Deciding what measures to use requires evaluating your specific circumstances and applying your judgement. Here we provide initial guidelines.

Factors to Consider in Selecting Measures

Self-report versus proxy report.
- Pediatric self-report (i.e., PROMIS Pediatric) is preferred for measuring outcomes among children. Use self-report when possible.
The number of participants at age transition points.
- When only a small number are affected, use of a consistent measure is preferred over administering a separate measure to the small group.
How far are respondents outside of the recommended age band?
- As the gap between the respondent’s age and the measure’s targeted age becomes bigger, the stronger the rationale for transitioning to the age-matched measure.
Individual change versus group comparisons.
- If you are interested in evaluating how an individual changes over time, it is preferred to use the same measure (if reasonable) versus using a different measure and converting it to a different metric.

Assessing 1- to 5-year-old Children

Use the PROMIS Early Childhood Parent-Report measures.

Assessing 5-year-old Children

Both the PROMIS Parent Proxy and PROMIS Early Childhood Parent-Report measure can be used with 5 year-olds. Select either one.
- For longitudinal research and/or on-going clinical follow-up, use the measure that aligns with the majority of the time frame with which the child will be studied. For example, if the measure is administered at child age 1 year through child age 5 years, use the PROMIS Early Childhood Parent-Report measure.
- If the child will be studied across in both early childhood (ages 1-5) and beyond age 5, switching to the PROMIS Parent Proxy measure is necessary. Unfortunately, crosswalks have yet to be established to link scores from PROMIS Parent Proxy measures to PROMIS Early Childhood Parent-Report measures. In other words, each measure will still produce a T-score comparing the child to the general population in that measure’s tested age group, but these measures have different distributions. Thus, a child’s T-score on each measure is still informative but not equivalent across the two measures. Please keep this in mind when designing studies and analyzing results that utilize both PROMIS Early Childhood Parent Report and PROMIS Parent Proxy.

Assessing 5- to 8-year-old Children at Single or Multiple Points in Time

Use PROMIS Parent Proxy measures with 5- to 8-year-old children.
For longitudinal research and/or on-going clinical follow-up, children will “age out” of the PROMIS Parent Proxy measure. Pediatric self-report (i.e., PROMIS Pediatric) is optimal for measuring outcomes among children ages 8 to 17 who can self-report. Consequently, it is desirable to transition to the Pediatric measure. Unfortunately, it is not possible to compare scores from PROMIS Parent Proxy and PROMIS Pediatric measures. Each measure will produce a T-score comparing the child to the general population in that measure’s tested age group, but these measures have different respondents and different score distributions. Thus, a child’s T-score on each measure is still informative but not equivalent across the two measures. Please keep this in mind when designing studies and analyzing results that utilize both PROMIS Parent Proxy and PROMIS Pediatric.
If feasible, collect both the Pediatric self-report and Parent Proxy report measures for children 8 years old and older. Their perspectives may be independently informative.

Assessing 8- to 17-year-old Children at Single or Multiple Points in Time

Use PROMIS Pediatric measures with 8- to 17-year-old children.

Assessing Adolescents/Young Adults/Emerging Adults at Single or Multiple Points in Time

Pediatric respondents will “age out” of the PROMIS Pediatric measures at age 18. If your respondents are aged 8-17 but you also include a small number of very young adults (e.g., 18-20-year-olds), use the pediatric measures for all respondents.
- This prioritizes keeping all respondents on the same metric and increases the likelihood that the items are understood by all respondents. For example, Reeve found that 14- and 15-year-olds did not always have sufficient literacy to complete the Adult measures.
- However, some of the measure content may not be relevant for young adults. For example, a pediatric measure may reference “kids my age”, ”school”, and “playing” which may be difficult for a young/emerging adult to answer.
If your respondents include 16- and 17-year-old adolescents and respondents 18 and older, use the Adult self-report measures for all respondents.
- This prioritizes keeping all respondents on the same metric.
- Unfortunately, there is limited information on how well individuals under 18 interpret and answer items from a specific adult measure. Some of this work was conducted by Reeve who found that 16- and 17-year-olds could complete adult measures appropriately. Ideally, you should assess the performance of adult measures you are using with adolescents.
- An alternative and less desirable approach is to use the Pediatric self-report measures for 16- to 17-year-olds and transform their scores to the Adult metric using linking tables (e.g., Reeve et al., 2016) for analyses. Linking tables provide estimated Adult scores from the Pediatric scores of 16- and 17-year-olds. However, using a linking table introduces additional measurement error (i.e., reduces score precision) and linking tables are not available for all domains. We discourage the use of linked scores to interpret individual change over time.
If your respondents include children/adolescents under 16-years old and adults (e.g., sample age range 14 – 30), we recommend using the Pediatric measures with 8- to 17-year-olds and Adult measures for those age 18 and up.
- Use linking tables to transform Pediatric measure scores to the Adult metric or vice versa (e.g., Reeve et al 2016). This enables all scores to be on the same metric for analyses.
- Using a measure designed for a specific age group increases the likelihood that the measure content will be understood and relevant for all respondents.
- However, linking tables only provide estimated scores. Using a linking table introduces additional measurement error (i.e., reduces score precision). Linking tables are not available for all domains. Linked scores are most appropriate when scores will be interpreted at the group level (e.g., group average) versus at the individual level.
- If feasible, consider collecting both the Pediatric and Adult measures from all respondents.
  - This enables comparisons between measures such as evaluating the similarity between self- and proxy-reports of anxiety or the difference between a self-report measure score and a score converted to a different metric using a linking table.
  - The drawback is that this doubles the assessment burden on respondents.

Reeve, B. B., Thissen, D., DeWalt, D. A., Huang, I. C., Liu, Y., Magnus, B., Quinn, H., Gross, H. E., Kisala, P. A., Ni, P., Haley, S., Mulcahey, M. J., Charlifue, S., Hanks, R. A., Slavin, M., Jette, A., & Tulsky, D. S. (2016). Linkage between the PROMIS^® Pediatric and Adult Emotional Distress Measures. Quality of Life Research, 25(4), 823–833. doi.org/10.1007/s11136-015-1143-z