Narayan Sharma successfully defended his dissertation on 15 September 2021

Narayan

Narayan Sharma’s dissertation seeks to improve patient’s safety by applying causal inference perspectives for time-varying variables, like capacity utilization, patient turnover and patient clinical complexity level in an observation study for Swiss general hospitals.

Narayan Sharma’s dissertation addresses the time-varying variables of hospital care demand and time-fixed variables to hypothesized data generating mechanism via directed acyclic graphs (DAGs) and to estimate the causal effect of capacity utilization on in-hospital mortality and is embedded in a nation-wide routine data of INVEST study: “INVEstigating Safety Tipping points in Swiss hospitals”.

It includes a systematic and iterative approach to exploring longitudinally the time-varying care demand measures to identify variations daily, days of the week, weekly and seasons of a year. This prepared time-varying variables for a causal model. Moreover, the important patient-level variable i.e., newly weighted Swiss comorbidity score was derived, validated and compared with existing comorbidity weights to predict in-hospital mortality. And was further utilized for risk adjustment in a causal model.  Utilizing both time-varying and time-fixed variables, the data generating mechanism was hypothesized via DAGs to estimate the causal effect of capacity utilization on 14-days in-hospital mortality. The safety tipping points were identified on the hospital level at 85th percentile of capacity utilization distribution, which was the treatment-exposure strategy for exposure to high capacity utilization. The inverse probability of treatment weight (IPTW) for marginal structural model (MSM) stabilized the weight between the exposure groups and eliminate the bias, i.e., treatment-confounder feedback (TCF) to estimate the causal effect. MSM incorporating IPTW with each additional day of a patient’s exposure to high capacity utilization the odds of 14-days in-hospital mortality increase by 2%.

Narayan Sharma’s dissertation bridges the evidence gap utilizing time-varying variables of hospitals in supporting the development of causal model in health service research. By using principles of causal inference, it contributes to the literature and practice by evaluating the safety tipping points in hospitals to better understand the nature of problems in patient safety, quality of care and the situations during pandemic situations by planning and managing hospital care demand.

The oral examination for the award of the doctoral degree was conducted virtually and onsite at the Kollegienhaus of the University of Basel, followed by a small reception for celebration.

With his doctorate, Narayan Sharma receives the dignity of a PhD (Dr. sc. med.) in Nursing Science from the University of Basel.