Collaborative Involvement: Empirical Evidence from the Field

Today's organizations access expertise through formal and informal relationships as work is becoming more interdependent and collaborative in nature [22]. The IOM 2002 report concluded that collaborations are becoming more common in public health practice [2]. A number of studies suggest that collaborations are a viable strategy as LHDs are actively partnering with different entities/stakeholders in their communities to address public health issues [6, 8, 23–25].

The study by Mays et al. [26] reported that about 33% of the public health effort was contributed by community stakeholders and about half of the studied LHDs maintained support and communication networks with major community players. Another study found that diverse management teams were associated with higher LHD performance as they tended to have more extensive interactions in the community [15]. Erwin provided empirical evidence that LHD performance depended on how LHDs and their communities worked together [27]. Stoto et al identified a need for more effective public health collaboration including communication and coordination between LHDs and other community organizations engaged in crisis response to disease outbreaks [28]. Thus, there should be continued engagement of community stakeholders including private business and health care, conducting and monitoring their activities over time as well as across different levels (city, state) [29].

Boards of Health

Boards of Health are often a major contributor to local public health effort. Scutchfield et al. reported that the presence of the Boards of Health was associated with higher performance of LHDs in mobilizing communities to address public health issues [7]. Mays at al. found that perceived effectiveness of public health initiatives was 14% higher in communities that had Boards of Health than in communities without them [26]. Savoia et al. found that the presence of a Board of Health was positively associated with public health emergency preparedness activities such as conducting drills, exercises and training [30].

Elected Officials

Elected officials are often responsible for making political and public health decisions, thus, their role and influence on public health activities may be important [31]. Collaboration between elected officials and LHDs was found extremely important in making decisions regarding pharmacy sales of syringes in California [32]. Kennedy reported that higher LHD performance was associated with the perceived support by elected officials [33]. Based on the results of an empirical study, Mays et al. recommended building and maintaining a strong governmental involvement in public health systems [26].

Community-Based Organizations

Community-based organizations are typically non-profit health and social service organizations located in the communities they service. They are familiar with their communities and, as a result, represent a crucial component of the public health system for identifying community public health issues, developing responses, and assessing results [34, 35]. Cali et al. demonstrated positive outcomes of collaboration between LHDs and academic institutions and, specifically, their combined efforts at investigating and addressing local environmental health hazards [36].

Physician and Hospital Providers

One empirical study found that medical providers were selected as major contributors to health care priority setting because of their deep understanding of community needs and expert knowledge necessary for decision-making [19]. Gadomski et al. described a number of positive initiatives resulting from collaboration between a LHD and a rural non-profit health system of community hospitals focused on preventive services [37]. Effective collaboration of LHDs with medical providers is critical because of the dynamic and quite unpredictable nature of disease outbreaks. It is important that medical providers help with disease surveillance before an outbreak occurs, and the information they provide is a critical part of outbreak identification and response [28].

Community and Patient Representatives

A number of studies found that the general public had much to contribute to public health decision-making to supplement the contribution of health care professionals and recommended intensive community involvement in public health activities [4, 19, 38, 39]. For instance, projects focusing on health care information needs of Hispanic populations reported that involving Hispanic community leaders in the projects from the beginning was extremely important for the success of the projects [40]. However, evaluations of public participation in decision-making and resource allocation in public health are practically non-existent despite the increasing demands for a greater representation of communities in priority setting [41]. Communities are sometimes involved late into the project or they are merely used as informants or research subjects [2].

Measures and sample

North Carolina's 86 LHDs include large and small health departments, public health authorities, a community health alliance and district health departments which represent multiple counties. The LHDs function with the support of the state level Division of Public Health for administrative services, generalized nursing consultation, information services provided by the State Center for Health Statistics, and accreditation. As of July 2011, 61 of the LHDs have completed an accreditation process which is based upon their capacity to perform the core functions and essential services as detailed in the National Public Health Performance Standards Program. NC LHDs have interacted with boards of health, state agencies, schools, non-profit organizations, hospitals and other organizations across a variety of issues [42]. No study, however, has addressed the nature and extent of collaboration in the process of determining community health status priorities.

The paper survey was administered to NC LHD directors or their representatives during regular monthly meetings of NC health officers in the fall of 2007. The institutional review board (IRB) of the University of North Carolina, Charlotte, approved the project protocol and survey instrument prior to data collection. Respondents consented to the use of the data, including publication of the findings, with the promise of anonymity and confidentiality. Fifty-three respondents completed the surveys, representing 62 percent of all NC local public health directors. Respondents were given a list of eight types of organizations/individuals as follows: Health Department Staff, Local Boards of Health, Patient Representatives, Community Advisory Board, Community Health Professionals (physicians, hospital administrators), Elected Officials, Community-based Organizations (United Way), and Experts. Respondents were asked to rate the level of involvement with each of the organizations/individuals on a scale from 1 (to a very little extent) to 7 (to a very great extent).

Analysis

The analysis had three components. First, an average score was calculated for each of the eight collaborator (individuals/organizations) categories. This score was the accumulated ratings (maximum possible 7 × 53 = 371) of involvement levels (1-7) for each collaborator category divided by the number of respondents (maximum possible = 7). Second, a total involvement score was calculated for each of the 53 local public health officials. This score was the sum of all collaborator ratings (maximum possible 7 × 8 = 56). Health official involvement scores were segmented into two groups at the approximate midpoint: low involvers (n = 26) and high involvers (n = 27) and their scores were distributed among three categories of involvement extent responses (scores 1-2, scores 3-5, scores 6-7).

Third, the technique of Self-Organizing Maps (SOM) was applied to the data. SOM is a data visualization technique which reduces the dimensions in a data set and displays similarities among the data objects as groups or clusters. Essentially, we were probing to identify clusters of health officials in regard to their involvement choices and levels; and, clusters of organizations/individuals collaborators in regard to their patterns of selection by health officials. The SOM is a stochastic clustering technique that distributes data across a grid (3 × 3 cells in this case) so that neighboring instances within a grid cell are likely to be more similar to each other than to instances that are associated with other grid cells [43]. The grid cell definitions adapt over the course of many iterations (training) and every instance is compared to all grid cells and a best match identified. Over the course of iterations the learning rate, how much each instance is allowed to alter its best match, is lowered, allowing the system to settle into an equilibrium, thus producing the clusters.

Involvement levels: Organizations/Individuals

Involvement levels: Health officers

Low involvers selected low involvement scores (scores 1 and 2) more than three times as frequently as High involvers (36% vs. 11%, respectively), but High involvers selected high involvement scores (scores 6 and 7) nearly four times as frequently as Low involvers (50% vs. 13%). Not even a single High involver selected a low involvement score for Community-Based Organizations or Community Health Professionals even though Low involvers selected those response categories 50% and 27% of the time, respectively. By contrast, no Low involver selected a high involvement score for Community Health Professionals or Elected Officials, while High involvers did so 70% and 19% of the time, respectively. The high involvement responses (scores 6 and 7) represented only 13% of the Low involvers responses but 50% of the High involvers responses. These Low/High comparisons indicated a greater tendency for a generalized response among highly involved health officers than for the low involvement group. In other words, high involvement is more generalized, low involvement is more selective.

SOMs

The SOM technique produced three clusters of health officers whose arrangement appears to correspond generally to the overall involvement scores: upper left cell grid location; high involvement, n = 19; upper center cell grid location, medium involvement, n = 14; and, lower right cell grid location, low involvement, n = 20. The relative positions of these three clusters are displayed in Figure 1.

The three clusters exhibit a characteristic shape in regard to organization/individual collaborations. The least amount of difference among the three clusters is at the extremes; i.e. where there is the highest involvement (Health Department Staff) and the lowest involvement (Experts). For only a single organizational category (i.e. Community-based Organizations) do we see a relatively higher level of involvement for the medium involvement group than for the high involvement group, and the difference is very small. The greatest apparent distance between the high and medium involvement groups is for collaboration with the Local Board of Health.

The SOM technique also produced four clusters from the eight organizational/individual categories identified in the survey (Figure 2). In this case, the data are used to identify the similarities between the organizational types and to reduce the number of types in an attempt to understand the patterns of collaboration. The four clusters were composed as follows: Community Advisory Board and Health Department Staff; Community-based Organizations and Community Health Professionals; Local Boards of Health; and, Experts, Elected Officials and Patient Representatives.

Note that individual respondents in Figure 2 are organized in a clockwise fashion from lowest to highest total involvement. For most of the range of responses, Health Department Staff and Community Advisory Board are the most prominent collaborators in priority setting. The Local Board of Health collaboration, although not quite as prominent overall, is strongest in the low medium ranges of total involvement, occasionally approaching or eclipsing that of the Health Department Staff. Community-Based Organizations and Community Health Professionals are more prominently involved with medium and high level involved public health officials. Experts, Elected Officials, and Patient Representatives were the least involved across the complete spectrum of total involvement.