Using social network analysis methods to identify networks of physicians responsible for the care of specific patient populations.

BACKGROUND: Coordinating health care within and among sectors is crucial to improving quality of care and avoiding undesirable negative health outcomes, such as avoidable hospitalizations. Quality circles are one approach to strengthening collaboration among health care providers and improving the continuity of care. However, identifying and including the right health professionals in such meetings is challenging, especially in settings with no predefined patient pathways. Based on the Accountable Care in Germany (ACD) project, our study presents a framework for and investigates the feasibility of applying social network analysis (SNA) to routine data in order to identify networks of ambulatory physicians who can be considered responsible for the care of specific patients. METHODS: The ACD study objectives predefined the characteristics of the networks. SNA provides a methodology to identify physicians who have patients in common and ensure that they are involved in health care provision. An expert panel consisting of physicians, health services researchers, and data specialists examined the concept of network construction through informed decisions. The procedure was structured by five steps and was applied to routine data from three German states. RESULTS: In total, 510 networks of ambulatory physicians met our predefined inclusion criteria. The networks had between 20 and 120 physicians, and 72% included at least ten different medical specialties. Overall, general practitioners accounted for the largest proportion of physicians in the networks (45%), followed by gynecologists (10%), orthopedists, and ophthalmologists (5%). The specialties were distributed similarly across the majority of networks. The number of patients this study allocated to the networks varied between 95 and 45,268 depending on the number and specialization of physicians per network. CONCLUSIONS: The networks were constructed according to the predefined characteristics following the ACD study objectives, e.g., size of and specialization composition in the networks. This study shows that it is feasible to apply SNA to routine data in order to identify groups of ambulatory physicians who are involved in the treatment of a specific patient population. Whether these doctors are also mainly responsible for care and if their active collaboration can improve the quality of care still needs to be examined.

Patients at high risk for a severe clinical course of COVID-19 - small-area data in support of vaccination and other population-based interventions in Germany.

BACKGROUND: Research has shown that the risk for a severe course of COVID-19 is increased in the elderly population and among patients with chronic conditions. The aim of this study was to provide estimates of the size of vulnerable populations at high risk for a severe COVID-19 course in Germany based on the currently available risk factor data. METHODS: We used nationwide outpatient claims data from the years 2010 to 2019 collected according to § 295 of the Code of Social Law V, covering data for all statutory health insurees (SHI) which is nearly 87% of the entire German population. We considered 15 chronic disorders based on the current state of knowledge about clinically relevant risk factors. Three risk groups for a severe COVID-19 course were defined: 1. individuals in the age group 15 to 59 years with at least two comorbid disorders; 2. individuals aged 60 to 79 years with at least one disorder and 3. all individuals 80 years and older irrespective of the presence of chronic conditions. Regional analysis was conducted at the level of administrative districts (n = 401). RESULTS: Overall, 26% of individuals over 15 years were at high risk for a severe COVID-19 course in 2019 amounting to a total number of nearly 18.5 million individuals in Germany. This included 3.8 million individuals in risk group 1, 9.2 million in risk group 2, and 5.4 million in risk group 3, corresponding to 8, 50 and 100% of German inhabitants in the respective age groups. On the level of the 17 administrative regions formed by the Association of SHI Physicians (ASHIP regions), the proportion of individuals at high risk ranged between 21% in Hamburg and 35% in Saxony-Anhalt. Small-area estimates varied between 18% in Freiburg (Baden-Württemberg) and 39% in the district Elbe-Elster (Brandenburg). CONCLUSIONS: The present study provides small-area estimates of populations at high risk for a severe COVID-19 course. These data are of particular importance for planning of preventive measures such as vaccination. TRIAL REGISTRATION: not applicable.

[Patient-sharing networks : New approaches in the analysis and transformation of geographic variation in healthcare]. / Virtuelle Behandlernetzwerke : Neue Ansätze zur Analyse und Veränderung räumlicher Versorgungsunterschiede.

The analysis of geographic variations has spurred arguments that area of residence determines access to and quality of healthcare. In this paper we argue that unwarranted geographic variations can be traced back to actions of individual patients and their healthcare providers (doctors, hospitals). These actors interact in a complicated web of shared responsibilities. Designing effective interventions to reduce unwarranted geographic variations may therefore depend on methods to identify these interactions and communities of providers with a shared accountability. In the US, Canada, and Germany, routine data have been used to identify self-organized informal or virtual networks of physicians and hospitals, so-called patient-sharing networks (PSNs). This is an emerging field of analysis. We attempt to provide a brief report on the state of work in progress. It can be shown that variation between PSNs in a given area is effectively greater than variation between regions. While this suggests that reducing unwarranted variation needs to start at the level of PSN, methods to identify PSNs still vary widely. We compare epidemiological approaches and approaches based on graph theory and social network analysis. We also present some preliminary findings of exploratory analyses based on comprehensive claims data of physician practices in Germany. Defining PSNs based on usual provider relationships helps to create distinctive patient populations while PSNs may not be mutually exclusive. Social network analysis, on the other hand, appears better equipped to differentiate between provider communities with stronger and weaker ties; it does not yield distinctive patient populations. To achieve accountability and to support change management, analytic methods to describe PSNs still need refinement. There are first projects in Germany which use PSNs as an intervention platform in order to achieve improved cooperation and reduce unwarranted variation in their care processes.