Sexually Transmitted Diseases  >  Program Guidelines  >  Partner Services

Program Operations Guidelines for STD Prevention
Partner Services

Previous Page

Table of Contents

Next Page

Appendix PS-K

TOOLS FOR NETWORK ANALYSIS

The lot system, developed in the 1960s to assist in the analysis of syphilis transmission, places person who are connected to each other in the same case folder (for example, if A and C are both interviewed, and both name B, they are thereby connected, and can be placed in the same lot). Thus, the notion of examining networks of persons involved in sexual transmission is a traditional part of syphilis epidemiology. In the past few years, a number of software tools have been developed to assist in network visualization and analysis. These tools, once implemented as part of a case management approach, can provide useful insights into the nature of the groups at risk for STD transmission, and may provide a program manager with guidance on how to proceed with the investigation of both endemic transmission and STD outbreaks.

The traditional syphilis and gonorrhea interviewing forms and the contact investigation form contain all of the information necessary to perform a network analysis. Program managers may choose as well to gather information not included in these forms (or the versions used in program areas) such as places of aggregation, household information, history of incarceration, etc.; such additional information can be systematically added to the record. Most managers will probably select a subset of the variables that are collected on the routine interview forms. For network analysis, the single critical variable is the unique identification number (ID). Together, the information on a patient and on one of his/her partners is the basic observational unit for network analysis. To create a file for network analysis, the manager will want a single observation for each patient-partner pair. Thus, if a man names four women, he will generate four observations in the data set. Each of the observations will contain all of his information and all the information on a contact. (The repetition of his information four times is simply a convenience; he will not be counted four times.)

The data set can be created directly from the STD*MIS, if it is used, or by entering the data from the interview record into a data base manager, such as Epi Info, or a spreadsheet. Once entered, the data are converted into an ASCII, or "flat" file. The total number of observations in the file will be equal to the total number of partners named by all of the persons interviewed. The preservation of a unique ID for each individual identified is crucial. For example, if a person is named as a partner, he or she will be assigned a unique ID. That unique ID should be carried with him if he is then interviewed. This procedure permits the network program to create the connections between people. If the STD*MIS currently in use in a program area uses multiple identifiers for the same person, this procedure will not work. Creating a unique ID for each individual, whether identified as a patient or a partner, may be the major source of extra work that has to be performed.

Once a flat file has been constructed, it is read by a program that converts it to a "DL" file that can be read by the network analysis program, UCINET V. The DL file is then used by the network analysis program to construct a file that can be used for network analysis. Program managers may or may not want to calculate the properties of the network they are examining. More likely, they will want to visualize the network. UCINET V contains a module that will create another file to be read by KRACKPLOT, a program that permits visualization of all the connections in the network, and permits the manager to move the nodes (patients and partners) on the screen to create a visually informative display. The combination of these two programs permits attributes to be assigned to nodes (such as infected or not infected, male or female, sex partner or nonsex partner, etc.). Specific attributes of nodes are represented by the types of boxes in which the node identifiers are placed, and specific types of connections can be identified by the color of connecting lines.

As an investigation unfolds, and cases with partners are added, the diagrams can be frequently redrawn to visualize the direction and intensity of the outbreak or of endemic spread. By manipulating the visualization, managers can develop a sense of the "loose ends" and the persons or groups that might be most important to revisit. Such visualization can also provide a graphic sense of the boundaries of the epidemic. For additional information on programs for network mapping, consult: http://www.insna.org/INSNA/soft_inf.html.

 

 

Previous Page

Table of Contents

Next Page