Multivariate Approaches

The article by Talbot et al. (2020) entails a study that examined the associations between
telebehavioral health (TBH) use and Medicaid telehealth policies among beneficiaries of fee-for-
service (FFS) in rural areas, specifically those with behavioral needs. The study also assessed the
relationships between TBH use and characteristics among the beneficiaries (Talbot et al., 2020).
The study involved explanatory variables and outcome variables. The outcome variable in the
study was TBH use. There were two categories of beneficiaries. Those with any TBH claims
were categorized as TBH users, while those with no such claims were grouped as TBH nonusers.
The explanatory variables were telehealth policies in State Medicaid programs. The researchers
classified beneficiaries as either those that were enrolled in Medicaid programs that explicitly
had a requirement for the informed consent process (Talbot et al., 2020). The other category of
beneficiaries was the beneficiaries that enrolled in programs where the originating site was
granted the payment of a facility fee (Talbot et al., 2020). Further, during the study, the
researchers created a four-level variable that classified beneficiaries into four groups. First, the
beneficiaries that were enrolled in programs with a requirement for informed consent but without
a facility fee policy, those in programs with a facility fee policy but without a requirement for
informed consent, beneficiaries with both policies, and those with neither policy.
The use of various types of methods to collect data in a study fosters the reliability and
validity of the data collected (Mohajan, 2017). In the study, Talbot et al. (2020) used various data
sources and methods to obtain the data, and this helped enhance the reliability and validity of the
data used. For instance, data was extracted from secondary sources such as the 2011 Medicaid
Analytic extract. The researchers also conducted a survey on Medicaid telehealth policies at the
state level. During the multivariate analysis, the researchers used the generalized estimating

equations, and these were instrumental in examining the odds of TBH use and how these were
related to telehealth policies such as facility fees and informed consent.
There are threats to the internal validity of the study. First, there is the issue of history or
time. There are events unrelated to the variables studied that affect the outcomes. For instance,
the study is based on data collected in 2011. However, since then, several events have taken
place that has affected the generalizability of the results as well as the outcomes, albeit
indirectly. For instance, the Covid-19 pandemic has acted as an incentive for states to support the
expansion of the use of telehealth services. This is the case, particularly among residents of rural
areas. While the data in the study reflects a strong FFS environment, this may have existed in
2011 but is no longer the case after the pandemic. Researchers can only use the data as a point of
reference since a lot has changed. The covid-19 pandemic thus affected the internal validity of
the study.
The multivariate model used in the study allowed researchers to examine the
relationships between several variables in an overarching manner, consequently helping them
understand the association among all the variables used. In the study, the researchers used
generalized estimating equations at the multivariate level (Talbot et al., 2020). These enabled the
researchers to examine the odds of using TBH and the connection between this and other
interrelated variables such as facility fee payment, informed consent, and how the variables
interact after adjusting for covariates. The researchers were able to study the many
interrelationships because they used a multivariate model.
One of the limitations of the multivariate models used in the study is that they used some
variables that have been changing over the years, consequently affecting the implementation of
the findings. Specifically, the study combined Medicaid programs that operate within Managed

Care Organizations (MCO) and FFS environments (Talbot et al., 2020). It then mixed them with
TBH policies that have been implemented within these regions. However, so many changes have
taken place within these environments since 2011, which is the year when the data analyzed in
the study was collected. First, there has been a decline in the state Medicaid programs that
operate within FFS environments. In 2011, it was at 72%, but by 2019, the number had gone
down to 20% (Talbot et al., 2020). Further, states are increasingly adopting a wide variety of
approaches when it comes to implementing facility fees, informed consent, and parity policy
levers covered in the study. Thus, including multiple variables in the study is a limitation
because some are changing at a faster rate than others, and this affects the dynamics of the study.
Another limitation is the interpretability of the independent variable. The study did not include
some participants, and such information gaps could affect the generalizability of the findings in
cases where certain groups within the population were omitted during data collection.

References

Mohajan, H. K. (2017). Two criteria for good measurements in research: Validity and
reliability. Annals of Spiru Haret University. Economic Series, 17(4), 59-82.
Talbot, J. A., Jonk, Y. C., Burgess, A. R., Thayer, D., Ziller, E., Paluso, N., & Coburn, A. F.
(2020). Telebehavioral health (TBH) use among rural Medicaid beneficiaries:
Relationships with telehealth policies. Journal of Rural Mental Health, 44(4),