WHY USE MIXED METHODS TO DEVELOP AND EVALUATE BEHAVIORAL INTERVENTIONS?
Mixed methods can bridge the gap between evidence generated from “ideal” intervention conditions (e.g., careful selection of participants after excluding “complex” cases with multiple comorbidities, high level of training and experience among interventionists) and the adoption of evidence-based practices for diverse populations in real-life settings. Although numerous novel behavioral interventions are funded and tested every year, knowledge gained from research is often slow or ineffective in resolving real-life problems, involving poor access to health services, engagement, and outcomes of health services. Quantitative approaches alone cannot fully describe the attitudes toward, and dissatisfaction with, aspects of current models of health care services by patients, families, clinicians, and administrators or areas of needed improvement (Becker & Newsom, 2003; Rossler, 2012). An in-depth qualitative analysis that assessed dissatisfaction in health care among African American patients with chronic illness found that low-income patients compared to middle-income patients were particularly dissatisfied with health care associated with lower resources available at care sites serving low- income patients, dealing with more bureaucracy in health care, and lack of health insurance (Becker & Newsom, 2003). Such depth of knowledge on dissatisfaction with health care among low-income African Americans with chronic illness would not have been available with a standard quantitative study that adjusts for socioeconomic status. In addition to quantitatively assessing the clinical, functional, and behavioral outcomes of interventions, qualitative methods take into account the contexts in which interventions are deployed at individual, social, and organizational levels.
Epidemiologic investigations involve statistical approaches to identify causes of diseases and environmental or social conditions/trends, to measure exposure, to count cases, and to guide treatment or prevention (Goodman, Buehler, & Koplan, 1990). Quantitative methods provide ways to reliably assess intervention outcomes, but not necessarily how or why an intervention worked or failed, or why uptake of an intervention was poor. Qualitative methods can fill in such important information by obtaining the participant’s perspective. As the leading causes of mortality have shifted from communicable diseases to noncommunicable diseases, multiple methods are required to understand complex relationships, often tied to a specific context or culture, that influence the onset and persistence of behavioral health problems (Murray, Vos, & Lozano, 2012). While investigators have used some form of “mixed” methods (e.g., combining epidemiologic and anthropologic approaches; Trostle, 1986, 2005) for some time, the use of “mixed methods” has recently emerged as a cohesive set of strategies to address individual motivational factors as well as contextual and environmental factors that contribute to disease burden (Creswell & Plano Clark, 2011; Tashakkori & Teddlie, 2003).
In order to flexibly respond to a changing landscape of behavioral health, we need effective interventions that are successfully implemented in community and clinical settings (Midgley, 2006). Without consideration of the contextual factors that influence uptake by patients, practitioners, and organizations, even effective interventions are unlikely to lead to substantial change in public health (The National Academy of Medicine, formerly known as the Institute of Medicine [IOM], 2006; National Institute of Mental Health, 2008; U.S. Department of Health and Human Services, 2001). The IOM Quality Chasm report called attention to the need for “outside the box” thinking related to the redesign of health care, including a strong focus on preferences and patient (person)-centered care, and evidence-based clinical decision making (IOM, 2006). Obtaining the “insider” perspective seeks to understand the patient’s point of view, employing methods that are designed to elicit the patient’s cultural model of illness and health. Understanding how providers view and adapt an intervention also is critical to translate interventions along the “pipeline” from research into practice.
The value of mixed methods has also been recognized by the National Institutes of Health (NIH). In 2011, the NIH Office of Behavioral and Social Science Research (OBSSR) convened a work group to develop guidelines for mixed methods proposals in the health sciences (Creswell & Plano Clark, 2011). NIH OBSSR published and widely disseminated the “Best Practices for Mixed Methods Research in Health Sciences” to aid investigators using mixed methods to prepare for competitive funding applications and to assist reviewers and staff to properly evaluate mixed methods applications and papers (Creswell, Klassen, Plano Clark, Clegg Smith, & Meisser, 2011). In 2014, several NIH Institutes funded a Mixed Methods Research Training Program for the Health Sciences (R25MH104660; Joseph J. Gallo, Principal Investigator). The need for training in mixed methods was clear because many investigators are seeking to gain insight on how context and culture influence the adoption and adaptation of behavioral interventions (Creswell et al., 2011). An increase in proposals submitted to NIH using mixed methods reflects the growing awareness of the importance of this approach in addressing population and behavioral health (Plano Clark, 2010) in fields such as nursing (Morse & Niehaus, 2009), medicine (Albright, Gechter, & Kempe, 2013; Creswell, Fetters, & Ivankova, 2004), mental health (Wittink, Barg, & Gallo, 2006), cardiovascular health (Curry, Nembhard, & Bradley, 2009), palliative care (Farquhar, Ewing, & Booth, 2011), public health (Curry, Shield, & Wetle, 2006), global health (Bass et al., 2013; Betancourt et al., 2011; Nastasi et al., 2007), implementation science (Bradley et al., 2009; Greenhalgh et al., 2010), health policy (Brannen & Moss, 2012), and health disparities (Apesoa-Varano & Hinton, 2013; Stewart, Makwarimba, Barnfather, Letourneau, & Neufeld, 2008).