Gh. Gilbert et al., EVALUATION OF BIAS AND LOGISTICS IN A SURVEY OF ADULTS AT INCREASED RISK FOR ORAL HEALTH DECREMENTS, Journal of public health dentistry, 57(1), 1997, pp. 48-58
Citations number
46
Categorie Soggetti
Dentistry,Oral Surgery & Medicine","Public, Environmental & Occupation Heath
Objectives: Designing research to include sufficient respondents in gr
oups at highest risk for oral health decrements can present unique cha
llenges Our purpose was to evaluate bias and logistics in this survey
of adults at increased risk for oral health decrements. Methods: We us
ed a telephone survey methodology that employed both listed numbers an
d random digit dialing to identify dentate persons 45 years old or old
er and to oversample blacks, poor persons, and residents of nonmetropo
litan counties. At a second stage, a subsample of the respondents to t
he initial telephone screening was selected for further study, which c
onsisted of a baseline in-person interview and a clinical examination.
We assessed bias due to: (I) limiting the sample to households with t
elephones, (2) using predominantly listed numbers instead of random di
git dialing, and (3) nonresponse at two stages of data collection. Res
ults: While this approach apparently created some biases in the sample
, they were small in magnitude. Specifically, limiting the sample to h
ouseholds with telephones biased the sample overall toward more female
s, larger households, and fewer functionally impaired persons. Using p
redominantly listed numbers led to a modest bias toward selection of p
ersons more likely to be younger, healthier, female, have had a recent
dental visit, and reside in smaller households. Blacks who were selec
ted randomly at a second stage were more likely to participate in base
line data gathering than their white counterparts. Comparisons of the
data obtained in this survey with those from recent national surveys s
uggest that this methodology for sampling high-risk groups did not sub
stantively bias the sample with respect to two important dental parame
ters, prevalence of edentulousness and dental care use, nor were concl
usions about multivariate associations with dental care recency substa
ntively affected. Conclusion: This method of sampling persons at high
risk for oral health decrements resulted in only modest bias with resp
ect to the population of interest.