Epidemiological data

From: Jonathan Bashor
Date: 10/3/99
Time: 9:37:00 AM
Remote Name:


What kinds of organizations employ epidemiologists ? What kinds of organizations use the data that
epidemiologists collect ? What is the accepted estimate of stuttering incidence in the USA ?

Re: Epidemiological data

From: Bobbie Lubker
Date: 11/3/99
Time: 6:33:34 PM
Remote Name:


What kinds of organizations employ epidemiologists? 

Every state health department in the US employs epidemiologists. Many city and county health departments
also employ their own epidemiologists, and all local health departments have access to state epid services.
NIH, FDA and the Centers for Disease Control and Prevention have large cadres of these scientists. The CDC
epid program monitors endemic and epidemic rates of certain reportable infectious diseases (rabies, rubella,
measles, HIV,etc). CEC also monitors rates of congenital anomalies. Pharmaceutical companies have
epidemiologists whose expertise lies in design of projects such as clinical trials for new treatments. 

All large hospitals and medical centers employ epidemiologists, and every hospital has access to epid consult.
Epidemiologists in these settings monitor nosocomial infections (look it up!), utilization of ER facilities, birth
and death rates, make-up of catchment areas and a host of population characteristics. 

Many industries employ epidemiologists to study absenteeism, employee health, insurance plans suitable for
particular workforces, and controls for accident prevention. WHO, Pan-American Health Organization,
various European , Asian and other health organizations around the world generate and utilize epid data for
health planning and policy development. 

Epidemiologists are found wherever questions are asked about increased or decreased risk, measurement of
risk (SLPs and As toss the words *at risk* around much too casually. At least four different measures of risk
are used routinely in epid. See new ASHA Prevention Manual), documentation of prevention effects, and
population distributions of health events. I am unusual in that my primary employment is in a school of
education with a joint appointment in speech and hearing sciences in a medical school. I am director of the
Center for Educational Management of Chronically Ill Children and Adolescents. My research is on the
changing epidemiology of language-learning disorders in chronically ill children, administrative models and
services (including speech-language-hearing services) in hospital schools, and educational transitions for
children with language-learning disorders deriving from iatrogenesis (look it up!) - true incidence! 

What kinds of organizations use the data that epidemiologists collect? 

ASHA, for one. Herb Baum at ASHA has expertise (doctorate in public health sciences from Hopkins) in
population research design, and ASHA now has an epidemiologist named Rob Mullen. And it*s about time.
In the past ASHA undertook some research projects without having the epidemiological expertise required for
such investigations. 

Survey research, which can be incredibly badly done by the naive and uninformed, is a major epid research
method. The ASHA Omnibus survey follows epid principles. 

The Human Genome Project is an excellent example of an organization that utilizes science ranging from
population trends (epid) to the molecular. 

The Red Cross, FEMA, March of Dimes, AARP, ACLD, ARC, all the disease-specific foundations,
professional associations (AMA, APA, ANA, ADA) and scores of others use epid data that measure risk,
survival rates, recovery rates, primary health conditions, secondary health conditions (e.g., communication
disorders),treatment efficacy...just about anything about health that anybody can think of measuring...from
birthweight and gestational age to the incidence ( yes, incidence) of deafness in Costa Rica. 

Much of what you read in the paper about health events is based on epid. For example, Three Mile Island
outcomes, the Gulf War Syndrome, new medications, the HIV epidemic, responses to vaccines, influenza
outbreaks in Hong Kong, resurgence of interest in thalidomide, eboli outbreaks, leukemia following the
Chernobyl meltdown, changing rates of birth incidence of Down Syndrome, Alzheimer*s (age of onset,
race-sex distributions)..and on and on...these are epid data. The geographic plotting of HIV occurrence in
Africa follows the building of highways. Epid is a tool to inform policy and principles on which practice may
be based. It is not usually intended to inform the treatment of a particular individual 

What is the accepted estimate of stuttering incidence in the USA? 

I have no idea. Well, actually I have an idea, but I want to make a point or two. Having an "accepted" estimate
of stuttering incidence and an "accurate" estimate of stuttering incidence may be two totally different

Go back and look at the article. As far as I know, no true incidence data have ever been generated. Again, as
far as I know, no appropriate strategies for collecting incidence data have ever been employed. Mansson
(1997) reported a proportion of about 5% for stuttering in a longitudinal study of three year olds...but
longitudinal design does not necessarily mean that incidence data are obtained. Mansson and a team screened
all the children in a contained population, but the important point is this: They reported stuttering after onset.
They will have missed any children who started to stutter and stopped prior to the screening (NEW CASES!).
This is not likely to be a large proportion, but from a theoretical perspective, they have not obtained incidence
data. Yairi and Ambrose (1999) have begun to report data on remission rates in children identified very near
the onset of stuttering, but these authors seem not to have reported population denominators that are absolutely
essential for incidence rate calculation. Further, participants responding to *recruiters* may be different from
those who do not choose to participate...and on and on. 

If (and it's a big "if"), the prevalence rate in adults is (for ease of calculation) 1%, and the recovery rate (for
ease of calculation) is about 80%, simple arithmetic supports an incidence rate of about 5%....but none of
these estimates has been calculated from pure epid models. 

Probably my most important point: Let me say also that it probably doesn't matter whether a national incidence
rate is identified to the fourth decimal place. Such gross rates are not very useful information except to get a
handle on the magnitude of the problem. What may matter more is that we come close enough with
population-specific measures (e.g. males and females, laterality groups) so that statistically significant
differences can be measured between clearly defined smaller populations. What else matters is that we are
sophisticated enough to recognize the impact of methodological procedures on whatever rates we are trying to
measure. Does all that make sense? It's clear to me! 

See also my comments to Melissa Bolling about normal functioning, randomness, sampling and probability
theory below.

Compliment to Author

From: Fran Freeman
Date: 10/4/99
Time: 12:35:37 PM
Remote Name:


Greatly enjoyed the article. I have printed it out so I can study it carefully for questions and comments. This
p.m. I am assigning the article to my students who will also be reading Susan Felsenfeld's chapter on
Epidemiology and genetics of stuttering. Good Work. 

Re: Compliment to Author

From: Bobbie Lubker
Date: 11/3/99
Time: 6:38:57 PM
Remote Name:


Thank you your enthusiastic comments. I hope the article's rather breezy, cheery and casual tone does not
prohibit its being taken seriously as a forum for scientific method. 

I have not read the Falsenfeld chapter on genetics and the epidemiology of stuttering. Could you please send
me the reference? I've been away from stuttering and into other aspects of epid for many years. 

You will want to look at the work of Falsenfeld and Plomen in the reference presented below. See what you
think about their assertions about incidence data. What have you learned from my article? 

Falsenfeld, S. & Plomin, R. (1997). Epidemiological and offspring analyses of developmental speech
disorders using data from the Colorado adoption project. Journal of Speech, Language and Hearing Research,
40, 778-791. 

I have written a letter to the ASHA Editorial Board about the spate of inaccurate epid concepts being accepted
for publication in our most prestigious journals. 

Incidence studies

From: Melisa Bolling (East Carolina University)
Date: 10/6/99
Time: 7:56:53 AM
Remote Name:


Bobbie Lubker, 

Hello. I'm a first year graduate student at East Carolina University this year. I enjoyed reading your article
"Who Stutters? When and Where as Clues to Why" and wanted to ask you a question. As an epidemiologist,
you say that there is no incidence data for stuttering. I can understand why because it's difficult for me to even
think of a way researchers could study incidence as you define it in your article. Perhaps one could perform a
long-term study where they select a representative group of normally functioning children (it seems as though
the group would have to be very large) at a prelingual stage and follow them throughout the years.
Periodically, the subjects would be assessed for any type of stuttering behaviors. However, it seems you
could choose 1000 children to participate, and at the end of your study, none of them were found to stutter -
that would be discouraging! Do you have any ideas about how one could possibly come up with this incident
data? Thank you for your time! 

Melisa Bolling

Re: Incidence studies

From: Bobbie Lubker
Date: 11/3/99
Time: 6:42:47 PM
Remote Name:


Ms. Bolling, you have raised some very sophisticated epidemiological issues. Let's look at several of the
issues in the order in which you have presented them. 

1. Little semantics here: incidence "...as you define it in your article". I hope it's clear that this is not just
my definition. It is an internationally accepted definition in public health. If SLPs and As want to "speak
epidemiology"...I firmly believe that accurate use of the lexicon indicates basic understanding...we will
use the vocabulary in ways that are understood by other professions. Right now we don't do that. I'm
working on it. 

2. "Perhaps one could perform a long-term study where they (sic) select a representative group of normally
functioning children (it seems as though the group would have to be very large) at a prelingual stage and
follow them throughout the years. Periodically the subjects would be assessed for any type of stuttering

You are exactly right. The only way to generate true incidence data is to start with a pre-lingual population
free from stuttering and follow it over time. Those who develop fluency disorders are called "incident
cases". Be careful of the study admission criterion of "normally functioning". Do you want to know about
the incidence in children with Down Syndrome compared with children who have no known chromosomal
anomalies? The outcome variable will be "stuttering" with severity strata defined. We will also
control...not in subject selection and matching the way clinicians often do...but via analysis for race, sex,
birthweight, cognitive development, and a lot of other good stuff. The importance of this research method
is that we retain denominator data, the population in which the stuttering group "lives" and develops.
Denominators are the heart beat of epid. Another article! 

Size of the group. This is an issue of probability and sampling theory. I offer the reference below.
Incidence studies are cohort studies. Case-control studies are those most familiar to clinicians.
Schlesselman (and others in epid and biostat) discuss some sampling strategies that I hope will become of
interest to clinicians....such as use of multiple controls to increase informativeness. 

Schlesselman, J. J. (1974). Sample size requirements in cohort and case-control studies. American Journal
of Epidemiology, 99, 6. 

Periodic assessment. Read Mansson (1997) and some of the new work published by Yairi and his
colleagues in JSLHR. They toss and mis-toss epid concepts around rather too freely, but their emphasis on
evaluations over time is right on. They do not report incidence data. 

3. "However, it seems you could choose 1000 children to participate, and at the end of your study, none of
them were found to stutter - that would be discouraging!" 

Go back to your own comment about having a "representative group". I suggest that you go to the ECU
department of statistics and talk with somebody about probability and sampling theory. That's what your
question is actually about. A couple of issues are hidden in your comment. You won't just "choose a 1000
children to participate". If, in your study, you employ accepted principles of probability to draw a sample
(random sampling, systematic sampling, etc.), and if our estimates of incidence over the years are
anywhere near accurate, it would be most unusual to draw a sample of prelingual children in whom no
stuttering appears. That would not be a representative sample. If our 5% incidence estimates are in the
ballpark, then we would expect 50 children in a probability sample to become dysfluent enough to fall into
the stuttering category. If probability samplings shows subgroups in which the 5% incidence estimate does
not hold true, then we are on to something. 

Also look at an oldie but a goodie: 

Slonim, M. J. (1960). Sampling. New York:Simon and Schuster 

Lots of more recent books about probability and sampling are out there. This remains my favorite for
beginners. There may even be newer editions of this one. I hang on to my dog-eared copy. 

Now, several other complex epidemiological issues raise their tousled heads. It is not necessary and may
not even be desirable to have "children" as the sampled unit. Read what Bruce Tomblin and his colleagues
have written about designing a sampling frame comprised of schools as the sampling unit in studies of
specific language disability. In your study, you might want to use day care centers as the sampled units in
your probability sample! 

Another complex concern is that stuttering probably does not occur randomly in the population. If
stuttering is, as Fran Freeman suggests a) more than one disorder (and epid distributions tend to indicate
that it is more than one disorder!!) and further, b) genetically determined in some groups, then it for sure is
not randomly distributed. We would expect family clusters in non-mobile societies and other distributions
in mobile societies. The issues of incidence rates and prevalence rates, their association with risk factors,
and meaningful risk calculations are subjects for another day. 

4. "Do you have any ideas about how one could possibly come up with this (sic) incident (sic) data?" Yes,
I have lots of ideas...see above...but not in a one-shot deal on the Internet! Is that too blunt? Such an
approach is not fair to you as a learner. The late Dr. John Cassel ( He was to epid as Van Riper was to
communication disorders.) used to say, "The less one knows about something, the greater the tendency to
over-simplify; the more one knows about something, the harder it is to make it simple." Hie yourself over
to the ECU Medical School. Charm them with your knowledge, interests and good questions. Push to
enroll in (or at least sit in on) lectures that they are sure to have on epid for physicians. I warn you that
such lectures are likely to be dull as dishwater, but you can make it interesting by applying the knowledge
to your areas of clinical interest. Don't fall into the unfortunate pattern that is evolving in speech and
hearing sciences. Lots of people are publishing inaccurate epid data generated from inappropriate research
methods....and they don't seem to recognize that a theoretical base, a knowledge base and specific kinds of
research design expertise are required. They don't know what their information gaps are. These are the
same people who get all bent out of shape when other folks mess about in communication disorders
without having what our profession has established as knowledge and skill foundations. 

5. Will you be at NCSHLA in the spring? I'm a past president. Maybe we can "speak epid" with each other
there. We can talk more about how to wangle your way into an epid course in the Medical School. 

geographical areas

From: LaVonne Reed
Date: 10/7/99
Time: 2:43:06 PM
Remote Name:


Thank you, I have a much clearer picture of what epidemiology is excatly, and a better understanding of the
difference between incidence and prevalence. You mentioned there is data that indicates population clusters of
people who stutter that exist in geographical regions. Are you aware of any regions in the United States that
have a higher population of people who stutter? 

Re: geographical areas

From: Bobbie Lubker
Date: 11/3/99
Time: 6:46:15 PM
Remote Name:


Thank you for your comments. It is gratifying for readers to acknowledge having obtained new
knowledge! Be cautioned, though, that what I elected to present for this article just scratches the surface of

"You mentioned there is data that indicates population clusters of people who stutter that exist in
geographical regions. Are you aware of any regions in the US that have a higher population of people who

The simple, simplistic answer to your question is, no, I don't know of regions in the US with increased
prevalence of stuttering, but....I think appropriate questions have often not been posed about this. We have
tip-toed very gingerly around some issues in stuttering research. 

Let me approach your question from a different perspective....not just regional differences but region as a
proxy for some other population variables. 

I don't have it in front of me...it's the kind of information that doesn't get into speech and hearing
reference lists....but a man named Brophy (and somebody whose name I can't remember in a journal I can
find for you!) estimated some years back that the prevalence of stuttering among adults in western Africa is
as high as 5%. I'm sure others have written on this too. Anecdotal data: A young Black woman exchange
student from Liberia lived in my home for a while some years ago. She had, by any observational
measure, moderate, sometimes severe stuttering. She is one of 16 children. Her mother and, if I remember
correctly, 11 of her siblings stutter. 

Now, if incidence rates, recovery rates and prevalence rates are, in fact, different among the Black
populations of western Africa from those in other parts of the world, could we form the hypothesis that
regions settled by immigrants from western Africa would have rates of stuttering different from those in
groups from other racial pools even in the same region? Or...another question might be about whether
stuttering rates are different among west Africans who immigrated and those who stayed at home. 

Another approach to your question is whether there are likely to be clusters of people who stutter in
communities that have good health services....including communication disorders services. They come for
treatment, and they stay....for all kinds of reasons. Certainly we know this phenomenon with cancer,
cystic fibrosis, and other major health conditions. Major medical centers and their specializations distort
epidemiologic prevalence data. 

My bias as an epidemiologist about Wendell Johnson's work on stuttering among Native Americans
several professional generations ago is that he was actually asking genetic questions, not cultural
questions! Notice that I acknowledge my bias. 

Have I told you more than you care to know? 

Is it only in the stutterer's mouth?

From: Jeff Knox
Date: 10/15/99
Time: 12:18:53 PM
Remote Name:


Very interesting paper. I was trained at U of Iowa, so the diagnosgenic approach to stuttering has been of
interest to me for several decades now. Ehud Yairi and I were students together with Dean Williams as our
mentor. There must be something to the thought that perception is 90% of fact--you think you can't hit a golf
ball and you may not be able to, etc. Also there must be something in the speech, especially the fluency
characteristics of the speech of stutterers, which calls attention to listeners that this is different. I applaud
everyone trying to do studies, so we can identify just what those characteristics are-- 


Re: Is it only in the stutterer's mouth?

From: Bobbie Lubker
Date: 11/3/99
Time: 6:49:07 PM
Remote Name:


Thanks so much for identifying yourself as an "old timer" who has been around for several decades. I'm
delighted that the article seemed to resonate with you. I may be one of the few people out here in the world
who met Wendell Johnson and Charles Van Riper and Bryng Bryngelson and Lee Travis and the second
generation of Dean Williams and Hugh Morris, who is, of course, still around. I even met Ollie Backus and
Jane Beasley who were doing child language therapy and calling it something else. 

Ehud Yairi and his colleagues are breaking out of the mold in their excellent work on recovery rates and the
characteristics of those who recover and those who don't. They need a good epidemiology consultant because
they are making some insupportable statements, and they apparently are not collecting denominator data, but
they are pioneering. My quarrel is, in part, with the reviewers of manuscripts who pretend to expertise they
don't have and accept some epid misinterpretations that should not find their way into print. 

I'll have to solve that another day. 

"There must be something to the thought that perception is 90% of fact--you think you can't hit a golf ball and
you may not be able to, etc." Whose perception? According to the dx theory, it's other people's perceptions
that cause the speaker to develop the problem. According to your golf ball example, it's one's own
perceptions that cause the problem. Or is it one's own perceptions about others' perceptions that cause the
problem? A major flaw in the dx theory is this causal conceptual leap: my perceptions can have such a
powerful impact on your perceptions and subsequently on your behavior. The dx theory meets none of the
epidemiological causal criteria (which is another article in itself!). It does not meet the temporal sequence
criterion: cause always comes before effect. 

It seems to me that you are right on target, and you are getting at this principle in your comments about
fluency characteristics that listeners respond to. These characteristics clearly, in my opinion, precede the
so-called cause in the dx theory. Caretakers are concerned about some characteristics and not about others. I
think the work of Yairi et al has marvelous potential to compare these behavioral events. 

I have forgotten (or repressed) a lot of what I used to know about the dx theory. The dx theory does not seem
to consider what is known as "the dose/response criterion": How much "dose" of calling attention does it take
to create stuttering (the response)? Does a little attention create mild stuttering? Does a lot of attention create
severe stuttering? 

The dx theory, in my opinion, kept us from a generation of other kinds of research on stuttering. Wendell
Johnson was a wonderful teacher and a charismatic man. I have one of the few copies of a film on stuttering
made at Iowa in the '50s. Jim Lubker (my husband's distant cousin), a speech science Iowa graduate of your
era, gave me the film. I had it repaired. Johnson was so convincing. Fransella and Beech wrote in the 1970s
that stuttering theory in the US was more like profession of faith than science. 

The Who and When of Stuttering

From: Terrence Lockett
Date: 10/20/99
Time: 8:43:54 PM
Remote Name:


I was very informed by what I read in your article. Through reading many articles and books I can understand
the link between stuttering and gender. My question is, does stuttering have a link to race and culture?

Re: The Who and When of Stuttering

From: Bobbie Lubker
Date: 11/3/99
Time: 6:49:50 PM
Remote Name:


Read what I wrote to LaVonne Reed about geographic distributions. That starts to get at race as a variable. My
informed epid opinion: I think there is absolutely no doubt that stuttering has different rates of occurrence and
possibly in different racial groups. That's what we have tip-toed around about. If stuttering occurs more often
in some races than in others, then offerings of early intervention and prevention must be informed by these
racial differences. Do we know whether the male to female distributions hold up in racial groups other than
white? I don't know that. Maybe somebody does. 

I acknowledge that my opinions about stuttering are very "biological/neurological". Again, what follows is an
epid perspective: yes, I think we know that there are cultural differences, but we must use research design and
statistical analyses (e.g., discriminate function analysis) that will permit us to control for race in asking
questions about culture. If we study only white boys in university communities, then all we will know about
is white boys in university communities. Bruce Tomblin and I have written a little bit about the epid selection
biases resulting from studying only clinical populations. People who find their way to clinical services are
quite different from the denominator population...and I believe race to be one of those differences. Thank you
for your forthright, objective question. 

Who, When, Where

From: Lynn K. Bender
Date: 10/21/99
Time: 4:09:54 PM
Remote Name:


This is a very eye opening article. I do beleive that looking at stuttering in a different light may reveal things
we never thought to consider. I would be interested in what a study would reveal when applied to
epidemiology standards. 

Re: Who, When, Where

From: Bobbie Lubker
Date: 11/3/99
Time: 6:50:33 PM
Remote Name:


Right on, Lynn Bender! I think we absolutely must start to look at the shared characteristics of the
sub-populations in which stuttering is and is not prevalent..in addition to the clinical characteristics of the

I have a couple of epid articles in mind that could be done relatively easily. Want to help with the lit

I moved into epid in part, as I wrote in the article, because of the '60s rubella epidemic, but my earliest
research interests were in stuttering, and epid has given me a way back into that with new and different ways
of thinking about populations, distributions and causality....and I hope, with some unique contributions to

Closing comments to all who posted

From: Bobbie Lubker
Date: 11/3/99
Time: 6:53:11 PM
Remote Name:


Apologies for my delayed response to one and all who read about the epidemiology of stuttering in the
cyberspace conference. I stopped all but the absolutely essential aspects of my life to work with contractors to
repair a flooded house and to care for my elderly mother who has aphasia and who is in sudden and
precipitous decline. 

I appreciate your interest in epidemiology. Judy Kuster, chief-honcho-in-charge of the Internet conference on
stuttering, was kind enough to forward your comments to me so that I could respond to them in an expanded
time frame. 

Your comments and questions confirm my suspicions that we really do need to use information on
epidemiology and its interpretation to inform policy, practice, research design, training and decision making in
speech, language and hearing sciences. I encourage each of you to find a good intro epid course. 'Good'
being the operative word. They can be mind-numbingly tedious with little understanding of what novice
learners need even to formulate questions and enter the discussion. Reading a book won't get it. Most
epidemiologists have never even considered communication disorders from epid perspectives; they are
surprised when I say that there are ninety thousand SLPs and As. Any way, your questions recharged my