Using an Accelerated Longitudinal Design to Study Outcomes for Children Who Are Hard of Hearing Considerations in selecting a design and "telling the story" of findings in a longitudinal study. Behind the Science
Research Design and Methods > Group Study Designs    >   Behind the Science
Free
Behind the Science  |   November 01, 2013
Using an Accelerated Longitudinal Design to Study Outcomes for Children Who Are Hard of Hearing
Author Affiliations & Notes
  • Mary Pat Moeller
    Boys Town National Research Hospital
  • The content of this page is based on selected clips from a video interview conducted at the ASHA National Office.
    The content of this page is based on selected clips from a video interview conducted at the ASHA National Office. ×
Article Information
Hearing Disorders / Research Issues, Methods & Evidence-Based Practice / Research Design and Methods / Group Study Designs
Behind the Science   |   November 01, 2013
Using an Accelerated Longitudinal Design to Study Outcomes for Children Who Are Hard of Hearing
CREd Library, November 2013, doi:10.1044/cred-gsd-bts-001
CREd Library, November 2013, doi:10.1044/cred-gsd-bts-001

I'll start with just a little bit of background. It's interesting that in contrast to a large number of studies that have been done with children who have cochlear implants, there really is a major gap in the evidence on children who are hard of hearing and wear hearing aids. So we know less about those children and, in fact, in the '70s Dr. Julia Davis wrote a book where she described them as "Our Forgotten Children." And so many years later, we still have sort of forgotten these children. There's not a lot of evidence to guide our practice, and there are many questions about their needs. They are a new generation of children in a lot of ways because since those earlier days, now we have universal newborn hearing screening. Most of these hard of hearing children are identified at birth, and they receive early access to amplification. So we have an opportunity for research to understand: how are these service innovations actually changing children's outcomes? Are they changing children's outcomes? And in what ways? And how should that be guiding our practice?

So that's sort of a backdrop of what we're doing. We're interested in the outcomes of children who are hard of hearing who had these advantages of being picked up at birth and fit with hearing aids. The conventional wisdom is that if we do catch them early and fit them with hearing aids, the children will progress as typically as possible. So our overarching questions are: Is that happening? And what factors contribute to promoting children's success? What factors might be barriers to success? Which children are struggling, and why?

What considerations factored into the development of your research design?

I will credit my colleague Dr. Bruce Tomblin -- we're co-PIs on this grant -- and, after many conversations about how to approach it, two issues of design were really important.

One was that we needed an epidemiological study. If we looked at the extant literature, a major problem was small Ns across most of the studies. Often the studies combined hard of hearing children with deaf children, making it difficult to draw decisions about the hard of hearing group. So one design decision was that we want children in a very specific decibel category, so that we can look at children who are hard of hearing who are not candidates for cochlear implants. This was quite a focused study. And we wanted a much larger N that would allow us to do multivariate analyses.

First of all, it is epidemiological -- that means we need a large N, lots of sites. We want to try to recruit a diverse group of children across diverse circumstances. So that was a goal.

Second, we're going to do a longitudinal study. Now, there's one potential approach, and that is go out and find all the babies you can and follow them prospectively.

Fortunately, Bruce had the wisdom to see that's going to make it complicated to get a large sample, if we rely only on babies in a tight age range. Instead, we chose what's called an accelerated longitudinal design. In this design, in the first two years of the study, we worked to recruit every participant who met our criterion who was between the ages of six months and six years. Then wherever they entered the study, we would follow them prospectively for at least three years, and gather retrospective records. So it was a combination of retrospective and prospective. What's interesting about that design is that as the children age into new categories, they are contributing data in every cell. So the sample sizes get larger and larger as the children age into these categories.

We were successful in recruiting 316 children across these age ranges. We always test them around their birthday, which allows you to make these nice cross-sectional analyses at every birth year. And also, the design allows you to look longitudinally across the prospective data set.

It's been a design that has really served us well, and in hindsight, I look at that and recognize how brilliant Bruce was to choose that design, because I know that it would have been really complicated to get the sample size had we only started with infants.

What are some challenges with implementing an accelerated longitudinal design?

It is a complicated design. It's probably easier to do the cross-sectional analyses. We're still working out exactly how we want to look at the longitudinal sample.

There are challenges related to missing data. You may have attrition on the longitudinal side. Although we worked very hard to prevent attrition.

No design is perfect, but for our purposes, I think this did serve us better than some other designs may have. It is complicated statistically, although there are ways to deal with missing data in the set. I would say one of the challenges is statistical, and so it's really important to have a very strong statistician on the team to support us in juggling both these cross-sectional looks and then the longitudinal approach.

The longitudinal approach is missing a lot in our field, and it's important to invest in, and so we're very motivated to begin to look at the longitudinal sample, which we've only recently finished gathering.

That's a problem with a longitudinal study: It takes a very long time to get to the end of the story. Because you need to gather all the data before you can truly tell the story. So we're just now at the threshold of being able to look at that aspect of the story.

What advice do you have for scientists working on studies that span several years?

Well, I know you're interested in encouraging young scientists -- I would say that this is not the type of study you want to start your career with. You see that I am older. If you're more experienced in the field, you certainly can invest in this type of study, more realistically.

It was a five-year study, and we just wrapped up that five-year study. It took us a full year, just in developing our measures and training all of the sites to conduct the science. A full year invested. And then four years in gathering the data across all of our sites.

Our children actually come from 17 different states, so we travel to do some of the data collection. It's complicated in terms of juggling all of that data collection. Fortunately, we have a very good infrastructure, which is essential when you're trying to work across multiple sites.

But I would say the fortunate thing is we just submitted our renewal application and it was successful, so all the work invested in recruiting this large cohort is paying off in that we'll continue to follow these children into their school-age years.

It's hard to do longitudinal work where you're not going to have the answer, because the grantsmanship part of this is, you need to be publishing data, and the data aren't fully matured yet. Yet, what the reviewers are going to want to see is a list of publications on your work.

So, we had to really be clever about what aspects of this project we can talk about now, before the datasets fully mature. And that took a lot of teamwork and creative thinking. Because again, you don't want to scoop yourself. You don't want to tell the end of the story and then realize "Oh, that isn't the end of the story, the story has changed." So we had to be wise about a term my colleague Kim Oller says, "Find the lower hanging fruit." In other words, what are nice aspects of this study that tell an important story.

One of our first publications was more of an epidemiological paper. It described the population itself. We looked at how quickly are families doing the follow-up stages after newborn hearing screening, and what factors affect that. In fact, we discovered that socioeconomic status played a role in how quickly families were following up after newborn hearing screening. That suggests to us we may need to devise resources and supports for families with less maternal education, for example, in creative and different ways. We aren't reaching that group, perhaps, in ways that we think we are. That was an important finding. But there's an example of thinking ahead.

We were always in discussion about what manuscripts we could begin to tackle that allow us to tell part of the story, when the story is not fully matured yet.

I am so enthused about this opportunity to follow our children for five more years. It's like this chance to see then end of the story. How do their preschool abilities influence their ability to learn language and literacy in the school years? So many interesting questions about school age environments and language and how it affects learning. So I just feel this tremendous eagerness to be able to address those questions and work to answer them. I feel like I'm in a very fortunate position to be doing research in an area where there's a very large research gap. People are enthused about the work because it's going to help us inform clinical practice. And that matters to me a lot.