The investigators propose a pragmatic trial of the comparative effectiveness of two
levels of remote internet-based cognitive behavior therapy (eCBT) to treat major
depressive disorder (MDD) with and without comorbidities. The investigators intend to
recruit 3,360 patients receiving primary care MDD treatment throughout West Virginia (WV)
and Kentucky (KY). The investigators two main aims will be to use experimental methods to
evaluate the aggregate effects of these interventions on patient-centered outcomes and to
investigate predictors of heterogeneity of treatment effects. MDD treatment in WV & KY is
far from optimal, with rural patients especially underserved. WV is the 2nd most rural
state in the US, has the 2nd lowest per capita income and has the highest proportion of
residents covered by Medicaid. Need for services is high, as indicated by WV having the
highest suicide rate of any state east of the Mississippi River and the highest opioid
death rate in the country. Yet WV ranks only 42nd in overall mental health care. The vast
majority of MDD treatment in rural areas is in primary care and consists largely of
antidepressant medication (ADM). Electronic medical records (EMRs) show that 88% of
primary care MDD patients in WV's FQHCs are treated exclusively with ADM and that the
other 12% are treated with ADM plus psychotherapy. The limited number of patients who can
access psychotherapy usually must go on a waiting list (often 3+ months) and travel long
distances for treatment once available. Access to telephone or videoconference
psychotherapy is limited. Yet 75% of depressed primary care patients express a desire for
psychotherapy either alone (40%) or in combination with ADM (35%).
This mismatch between treatment availability and preference is important because MDD
remission increases substantially when patients are treated with their preferred type.
There is thus good reason to believe that providing access to eCBT will improve MDD
treatment outcomes in our trial. Indeed, prior controlled trials show that both types of
eCBT the investigators will randomize yield significantly better outcomes than waiting
list controls. Controlled trials also show that guided eCBT yields equivalent outcomes to
telephone and face-to-face CBT, but at much lower cost. Other controlled trials show that
combined CBT-ADM yield significantly better outcomes than either CBT-alone or ADM-alone,
although these comparisons have been made only for face-to-face CBT. These results
provide good reason to believe that offering eCBT in rural FQHCs throughout WV and KY
could improve MDD outcomes. Existing research on eCBT in rural areas, while promising,
has been limited, making the research the investigators propose important to provide
actionable information for patients and other stakeholders. Results intend to inform
decisions about whether to offer/use eCBT, with what level of guidance, and for whom.
Unguided eCBT is web-based CBT completed with computerized feedback but no clinician
involvement after an initial orientation meeting. Guided eCBT is web-based CBT completed
with a remote eCoach who communicates with the patient via email, text, and telephone.
eCoaches also provide elements of remote collaborative care case management, such as
encouraging ADM adherence, monitoring ADM side effects and treatment response,
coordinating with the primary care physician (PCP), and facilitating specialty referral.
Collaborative care is known to be highly effective in promoting MDD symptomatic
remission. In addition, a study in Arkansas FQHCs found that remote collaborative care
case management out-performed on-site case management in rural clinics too small to
justify having a dedicated mental health case manager on staff. However, remote
collaborative care case management often involves delivering telephone CBT. A major
constraint on expanding the collaborative care model for primary care MDD treatment,
which has been used in urban but not rural WV clinics, is lack of case managers who can
deliver telephone CBT. Thus, expanding eCBT in rural WV would allow offering a strongly
evidence-based form of patients' preferred treatment (psychotherapy) and a form of a
well-validated rural MDD care model (collaborative care case management with guided eCBT)
that cannot be offered currently because of limited clinical resources.
Given its documented efficacy and rapid spread, the investigators expect eCBT to become
widely available in rural WV as a result of our trial. But two real-life decisional
dilemmas will arise in that context. Primary care clinicians will be faced with the
decision about when to recommend eCBT and at what level of intensity. Patients will be
faced with the decision of whether to accept guided or unguided eCBT as part of their
treatment plan. These are non-trivial decisions, as eCBT incurs a time cost, and guided
eCBT incurs a financial cost and requires interactions with a supporter for patients who
desire independence and privacy. Further, eCBT has the potential to harm, as when lack of
engagement leads the patient to drop out of all treatment, including ADM, whereas that
patient would have remitted with ADM. Our heterogeneity of treatment effects (HTE)
analyses will examine which patients profit from guided eCBT, which do equally well or
better with unguided than guided eCBT, and which do as well or better with ADM in the
absence of eCBT. A good deal of research has been carried out on eCBT HTE, although not
in conjunction with ADM. This research suggests that the value of eCBT for MDD varies
considerably depending on diverse patient characteristics the investigators plan to
study. As with the comparative effectiveness evidence for eCBT vs other MDD therapies,
though, research on MDD HTE up to now has focused on narrowly-defined symptom outcomes.
In addition, although more than two dozen consistently significant baseline
patient-reported predictors of MDD HTE have been documented, no single study ever
considered more than a handful of these predictors. In addition, past MDD HTE studies
have been underpowered. Our analysis will be based on a sample of 3,360 patients powered
to detect HTE in the entire sample. Our patient and provider partners have indicated that
evidence about the prescriptive predictors of these differences will be of great value in
their treatment selection decisions. The causal model underlying the design is drawn from
previous studies reviewed here: that MDD remission of primary care patients can be
increased by adding eCBT to treatment-as-usual (TAU) via mechanisms that include
influencing cognitions and behaviors to promote psychological recovery and encouraging
increased ADM compliance. This model underlies all aspects of our design (selection of
population, interventions, measures, analytic methods, procedures for handling
confounding, time frame).
