Statistical approaches rooted in econometric methodology, so farforeign to psychosomatic medicine, have provided dynamic psycho-somatic models on brittle diabetes in yielding to Meissner s (2006) and others call for a more integrated view on mind-body relationships with no a priori cause and effect assignments to interacting variables. The conception and portrayal of such models is the focus of this work, in addition to the clinical findings they provide on the case study otherwise primarily serving as an example for their application. Over 120 days, this structured diary time series case study explored the mutual interactions of individual affect states in a classic three person family with its type 1 diabetic adolescent s daily blood glucose variability and vice versa. Glycemic variability was measured through daily standard deviations of blood glucose recordings (at least three per day). For the same period of time, affect states were captured individually utilizing the self-assessment manikin (Lang, 1980; Bradley & Lang, 1994) on affective valence (positive negative), arousal (high low), and dominance (sense of being absent sense of being present). Auto- and cross-correlating the stationary absolute (level) values of the mutuallyinteracting parallel time series data sets through standard vector autoregression (VAR, Lütkepohl, 2005; Lütkepohl & Krätzig, 2004) and a newly conceived Optimized Multivariate Lag Order Selection Process (Winker, 1995, 2000; Savin & Winker, 2013)allowed for the formulation of threepredominantly consistent models. In the two standard VAR models Cholesky Impulse Response Analysis was applied at a 95 per cent confidence level, cumulatively evidencing for an adolescent being happy, calm, and experiencing high dominance to exhibit less glycemic variability andhence diabetic derailment. A non-dominating mother and a happy father also seemed to reduce glycemic variability. Random external shocks to the two VAR models increasing glycemic variability affected onlythe adolescent and her father: In one model, the male parent exhibited high dominance; in the other, hecalmed down while his daughter turned sad. All effects lasted for lessthan four full days. In the third model based on the Optimized Multivariate Lag Selection Process, more specific temporal relations between affect states and the biological marker of glycemic variability were isolated at statistical significance: Low glycemic variability and therefore good diabetic control correlated with high glycemic variability four days earlier, an excited mother three days earlier, a calm mother seven days earlier, a non-dominating mother four days earlier, a happy father both five and six days earlier, an excited father both three and seven days earlier, and a non-dominating father both two and five days earlier. Low glycemic variability also correlated with a happy child six days later, a calm mother three days later, and a non-dominating father one day later.Graphical representations were proposed for all three models the intention being a demonstration of avenues for clinically oriented presentations of arguably rather abstract quantitative findings. Additionally, a multiply regressive approach to the data with interval-valued variables and a qualitative case vignette were presented, to complement these highly quantitative models. Extant literatureon brittle diabetes in children andadolescents and the family dynamics complementing it was reviewed in light of all findings. The recurring correlation between maternal dominance and poor glycemic control was recognized. In addition, the prospects and contingencies arising from applying econometric theory to psychosomatic questionswere discussed throughout this thesis. The value and limitations of qualitative and quantitative research on brittle diabetes in general, as well as pertaining to this study, received particular attention.
Verknüpfung zu Publikationen oder weiteren Datensätzen
