S08: Using Observational Studies and Secondary Data Analyses for Research on Complementary and Integrative Healthcare (CIH)

Richard Nahin1, Annette Fitzpatrick2, Jeffery Dusek3, Ryan Bradley4

1NCCAM, NIH, USA; 2Department of Epidemiology, University of Washington, USA; 3Penny George Institute for Health and Healing, Abbott Northwestern Hospital, USA; 4Bastyr University, USA


Observational research offers a variety of study designs for gathering evidence from existing data sets regarding CIH interventions and disciplines.  Confounding causal inference in such studies is the possibility that patient characteristics, not balanced by random assignment of treatment, may be the true cause of any observed effects.  A variety of methods can be used to control for measured characteristics. This symposium,chaired and opened by NCCAM's Director, Josephine Briggs, presents examples where researchers are taking advantage of existing data sources to study outcomes associated with CIH.    Richard Nahin introduces a retrospective, longitudinal study examining the effects of relaxation techniques on mortality in individuals with hypertension using data from the 1999 National Health Interview Survey linked to the National Death Index.  To control for potentially confounding factors, propensity scores derived from logistic regression were used to match individuals using relaxation techniques to those not using such techniques.  Cox proportional hazard models were used to examine overall mortality.    Wendy Weber presents an analysis of CIH utilization data from the longitudinal cohort Mid-Life in the United States study (MIDUS).  The MIDUS data were analyzed using multivariate regression techniques to determine if different types of CIM therapies were associated with mortality during long-term follow-up of the cohort.  Annette Fitzpatrick and Carla Mercado employ data collected over 8 years on 3,068 adults enrolled in a clinical trial, the Ginkgo Evaluation of Memory Study, to investigate a question outside of the study”s primary aims.  Using Cox Proportional Hazards Regression and time-dependent risk factors, they evaluate whether use of dietary supplements, including non-vitamin non-mineral dietary supplements, increases or decreases risk for mortality and incidence of cardiovascular disease.  Jeffery Dusek highlights an ongoing prospective, observational, pre-post study of CIH at a large hospital-based CIH program.  Using data from an electronic medical record (EMR), OLS models will be used to explore potential effects of CIH on immediate pain change.  Growth curve modeling techniques will assess the effectiveness of one or more CIH therapies on repeated measures of self-reported pain status over a five-hour period after therapy. Ryan Bradley presents a mixed methods prospective cohort study assessed the impact of adjunctive naturopathic care (ANC) on self-care behaviors and clinical risk factors in patients with poorly controlled Type 2 diabetes. Observed changes in clinical risk and utilization of clinical services during ANC were compared to a usual care control group created from EMR data of eligible patients who were not invited to participate in the intervention group.