Mapping and modeling of physician collaboration network

Shahadat Uddin, Jafar Hamra, Liaquat Hossain

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Effective provisioning of healthcare services during patient hospitalization requires collaboration involving a set of interdependent complex tasks, which needs to be carried out in a synergistic manner. Improved patients' outcome during and after hospitalization has been attributed to how effective different health services provisioning groups carry out their tasks in a coordinated manner. Previous studies have documented the underlying relationships between collaboration among physicians on the effective outcome in delivering health services for improved patient outcomes. However, there are very few systematic empirical studies with a focus on the effect of collaboration networks among healthcare professionals and patients' medical condition. On the basis of the fact that collaboration evolves among physicians when they visit a common hospitalized patient, in this study, we first propose an approach to map collaboration network among physicians from their visiting information to patients. We termed this network as physician collaboration network (PCN). Then, we use exponential random graph (ERG) models to explore the microlevel network structures of PCNs and their impact on hospitalization cost and hospital readmission rate. ERG models are probabilistic models that are presented by locally determined explanatory variables and can effectively identify structural properties of networks such as PCN. It simplifies a complex structure down to a combination of basic parameters such as 2-star, 3-star, and triangle. By applying our proposed mapping approach and ERG modeling technique to the electronic health insurance claims dataset of a very large Australian health insurance organization, we construct and model PCNs. We notice that the 2-star (subset of 3 nodes in which 1 node is connected to each of the other 2 nodes) parameter of ERG has significant impact on hospitalization cost. Further, we identify that triangle (subset of 3 nodes in which each node is connected to the rest 2 nodes), alternative k-star (subset of k nodes in which 1 node is connected to each of other k-1 nodes), and alternative k-2 path (subset of k nodes in which, between a specific pair of nodes, there exists k-2 paths of length 2) parameters of ERG have impact on the hospital readmission rate. Our findings can have implications for healthcare administrators or managers who could potentially improve the practice cultures in their organizations by following these outcomes.

Original languageEnglish
Pages (from-to)3539-3551
Number of pages13
JournalStatistics in Medicine
Volume32
Issue number20
DOIs
Publication statusPublished - 10 Sep 2013
Externally publishedYes

Fingerprint

Physicians
Hospitalization
Vertex of a graph
Pregnenolone Carbonitrile
Modeling
Random Graphs
Patient Readmission
Health Insurance
Delivery of Health Care
Star
Health Services
Healthcare
Organizations
Subset
Costs and Cost Analysis
Graph Model
Statistical Models
Insurance
Administrative Personnel
Triangle

Cite this

Uddin, Shahadat ; Hamra, Jafar ; Hossain, Liaquat. / Mapping and modeling of physician collaboration network. In: Statistics in Medicine. 2013 ; Vol. 32, No. 20. pp. 3539-3551.
@article{26240f8533cc446dbf71798b5d301c38,
title = "Mapping and modeling of physician collaboration network",
abstract = "Effective provisioning of healthcare services during patient hospitalization requires collaboration involving a set of interdependent complex tasks, which needs to be carried out in a synergistic manner. Improved patients' outcome during and after hospitalization has been attributed to how effective different health services provisioning groups carry out their tasks in a coordinated manner. Previous studies have documented the underlying relationships between collaboration among physicians on the effective outcome in delivering health services for improved patient outcomes. However, there are very few systematic empirical studies with a focus on the effect of collaboration networks among healthcare professionals and patients' medical condition. On the basis of the fact that collaboration evolves among physicians when they visit a common hospitalized patient, in this study, we first propose an approach to map collaboration network among physicians from their visiting information to patients. We termed this network as physician collaboration network (PCN). Then, we use exponential random graph (ERG) models to explore the microlevel network structures of PCNs and their impact on hospitalization cost and hospital readmission rate. ERG models are probabilistic models that are presented by locally determined explanatory variables and can effectively identify structural properties of networks such as PCN. It simplifies a complex structure down to a combination of basic parameters such as 2-star, 3-star, and triangle. By applying our proposed mapping approach and ERG modeling technique to the electronic health insurance claims dataset of a very large Australian health insurance organization, we construct and model PCNs. We notice that the 2-star (subset of 3 nodes in which 1 node is connected to each of the other 2 nodes) parameter of ERG has significant impact on hospitalization cost. Further, we identify that triangle (subset of 3 nodes in which each node is connected to the rest 2 nodes), alternative k-star (subset of k nodes in which 1 node is connected to each of other k-1 nodes), and alternative k-2 path (subset of k nodes in which, between a specific pair of nodes, there exists k-2 paths of length 2) parameters of ERG have impact on the hospital readmission rate. Our findings can have implications for healthcare administrators or managers who could potentially improve the practice cultures in their organizations by following these outcomes.",
keywords = "Exponential random graph, Hospitalization cost, Physician collaboration network, Readmission rate",
author = "Shahadat Uddin and Jafar Hamra and Liaquat Hossain",
year = "2013",
month = "9",
day = "10",
doi = "10.1002/sim.5770",
language = "English",
volume = "32",
pages = "3539--3551",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley & Sons",
number = "20",

}

Mapping and modeling of physician collaboration network. / Uddin, Shahadat; Hamra, Jafar; Hossain, Liaquat.

In: Statistics in Medicine, Vol. 32, No. 20, 10.09.2013, p. 3539-3551.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Mapping and modeling of physician collaboration network

AU - Uddin, Shahadat

AU - Hamra, Jafar

AU - Hossain, Liaquat

PY - 2013/9/10

Y1 - 2013/9/10

N2 - Effective provisioning of healthcare services during patient hospitalization requires collaboration involving a set of interdependent complex tasks, which needs to be carried out in a synergistic manner. Improved patients' outcome during and after hospitalization has been attributed to how effective different health services provisioning groups carry out their tasks in a coordinated manner. Previous studies have documented the underlying relationships between collaboration among physicians on the effective outcome in delivering health services for improved patient outcomes. However, there are very few systematic empirical studies with a focus on the effect of collaboration networks among healthcare professionals and patients' medical condition. On the basis of the fact that collaboration evolves among physicians when they visit a common hospitalized patient, in this study, we first propose an approach to map collaboration network among physicians from their visiting information to patients. We termed this network as physician collaboration network (PCN). Then, we use exponential random graph (ERG) models to explore the microlevel network structures of PCNs and their impact on hospitalization cost and hospital readmission rate. ERG models are probabilistic models that are presented by locally determined explanatory variables and can effectively identify structural properties of networks such as PCN. It simplifies a complex structure down to a combination of basic parameters such as 2-star, 3-star, and triangle. By applying our proposed mapping approach and ERG modeling technique to the electronic health insurance claims dataset of a very large Australian health insurance organization, we construct and model PCNs. We notice that the 2-star (subset of 3 nodes in which 1 node is connected to each of the other 2 nodes) parameter of ERG has significant impact on hospitalization cost. Further, we identify that triangle (subset of 3 nodes in which each node is connected to the rest 2 nodes), alternative k-star (subset of k nodes in which 1 node is connected to each of other k-1 nodes), and alternative k-2 path (subset of k nodes in which, between a specific pair of nodes, there exists k-2 paths of length 2) parameters of ERG have impact on the hospital readmission rate. Our findings can have implications for healthcare administrators or managers who could potentially improve the practice cultures in their organizations by following these outcomes.

AB - Effective provisioning of healthcare services during patient hospitalization requires collaboration involving a set of interdependent complex tasks, which needs to be carried out in a synergistic manner. Improved patients' outcome during and after hospitalization has been attributed to how effective different health services provisioning groups carry out their tasks in a coordinated manner. Previous studies have documented the underlying relationships between collaboration among physicians on the effective outcome in delivering health services for improved patient outcomes. However, there are very few systematic empirical studies with a focus on the effect of collaboration networks among healthcare professionals and patients' medical condition. On the basis of the fact that collaboration evolves among physicians when they visit a common hospitalized patient, in this study, we first propose an approach to map collaboration network among physicians from their visiting information to patients. We termed this network as physician collaboration network (PCN). Then, we use exponential random graph (ERG) models to explore the microlevel network structures of PCNs and their impact on hospitalization cost and hospital readmission rate. ERG models are probabilistic models that are presented by locally determined explanatory variables and can effectively identify structural properties of networks such as PCN. It simplifies a complex structure down to a combination of basic parameters such as 2-star, 3-star, and triangle. By applying our proposed mapping approach and ERG modeling technique to the electronic health insurance claims dataset of a very large Australian health insurance organization, we construct and model PCNs. We notice that the 2-star (subset of 3 nodes in which 1 node is connected to each of the other 2 nodes) parameter of ERG has significant impact on hospitalization cost. Further, we identify that triangle (subset of 3 nodes in which each node is connected to the rest 2 nodes), alternative k-star (subset of k nodes in which 1 node is connected to each of other k-1 nodes), and alternative k-2 path (subset of k nodes in which, between a specific pair of nodes, there exists k-2 paths of length 2) parameters of ERG have impact on the hospital readmission rate. Our findings can have implications for healthcare administrators or managers who could potentially improve the practice cultures in their organizations by following these outcomes.

KW - Exponential random graph

KW - Hospitalization cost

KW - Physician collaboration network

KW - Readmission rate

UR - http://www.scopus.com/inward/record.url?scp=84882258354&partnerID=8YFLogxK

U2 - 10.1002/sim.5770

DO - 10.1002/sim.5770

M3 - Article

VL - 32

SP - 3539

EP - 3551

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 20

ER -