Many of the essentials of adaptive management—modeling complex, dynamic problems; interacting with a wide range of stakeholders; and an evidence-based, iterative approach to decision making—are familiar to public health. The process is perhaps most akin to evidence-based medicine and its cousin, evidence-based public health (Brownson et al. 2009
; Eriksson 2000
). As with these approaches, embracing the entire paradigm confers several advantages over a disjointed approach.
Models are fundamental to adaptive management and can be relatively straightforward conceptual models that distill the system into key components or more complex computer-based models (Ebi 2011
). Integrated assessment models (IAMs) are examples of the latter and are often used to facilitate decision making and to assess impacts of potential interventions. These models draw from multiple disciplines to capture system behavior (Chan et al. 1999
). Such frameworks have been developed only for certain climate-sensitive health outcomes, and there are currently no IAMs for heat specifically, although some models of urban heat impacts are being developed (Dawson et al. 2009
). Team-based modeling efforts to organize and focus group thinking are also relevant (Vennix 1996
). Examples of these types of models, and other tools, are presented in .
Steps in the adaptive management cycle, central actions in each step, and tools useful for completing the central actions.
Despite the lack of an IAM for urban heat, we can outline an adaptive management process focused on EHEs and consider how this process might evolve iteratively as uncertainties regarding the climate system, health communications, exposure determinants, population susceptibility, and the response to various potential interventions are clarified.
Assessment is the first step of the adaptive management process (). This is one type of vulnerability assessment, for which multiple theoretical frameworks and methodologies are available. In the case of heat, several components of risk, from hazard frequency and severity to population exposure and susceptibility, must be assessed. EHE risk results from the interaction of various factors at multiple scales, as depicted in . Using the natural hazards risk formula to incorporate hazard probability, hazard exposure, and population susceptibility (Malilay et al. 1997
), taking care to incorporate social factors affecting vulnerability (Sullivan and Meigh 2005
), can help organize these components. A wide range of stakeholders should be engaged, from neighborhoods to emergency medical responders to city planners to electrical and water utilities, in order to assess dynamics affecting both exposure and response. Substantial literature provides insight into effective strategies for stakeholder engagement (Lim et al. 2005
Components of heat-related morbidity and mortality risk operative at various spatial scales. AC, air conditioning.
Planning prepares for real-world implementation and often uses IAMs. Response activities incorporated into the model should parallel exposures, that is, strategies to change land use and urban form at the mesoscale (Clarke 1972
; Golden 2004
; Shimoda 2003
); building materials, vegetation, and other factors affecting sensible heat at the neighborhood and street levels (Jenerette et al. 2007
; Silva et al. 2010
); home visitation and other social capital strategies at the neighborhood level (Luber and McGeehin 2008
; Wolf et al. 2010
); and strategies for changing the home and other environments and relocation of susceptible people (O’Neill et al. 2005
). Planning should also incorporate a range of possible futures and be tailored to stakeholder inputs. Improved forecasts that are downscaled to a finer geographic scale can help to limit uncertainty.
Certain tools allow practitioners to organize information on the hazard and population at risk in order to prioritize responses. Vulnerability mapping, for example, allows for visual rendering of relative population vulnerability in relation to hazards and response infrastructure (Li et al. 2010
; Morrow 1999
). The maps should be used to identify a range of possible interventions to incorporate into the IAM. Decision support tools, including software tools, documents, and work processes, are designed to help practitioners and policy makers evaluate decisions available to them and the potential impacts of those decisions across complex systems, but few tools for selecting adaptation options are available (Pyke et al. 2007
Stakeholders should also heavily influence the selection of adaptation options. Adaptation requires a new level of cross-sectoral planning, and other sectors are increasingly acknowledging the need to incorporate health (Kashyap 2004
) and vice versa (Cole et al. 2007
). In the case of extreme heat, electricity generation for air conditioning is a primary concern, and water and forestry are also important. Dynamic models to simulate such interconnected relationships have not been well developed in public health but are increasingly important. Adaptive management must consider scenarios in which other sectors that typically facilitate public health are not fully functional, and alternatives must be modeled and explored. Importantly, research has shown that the primary threat to such systems is the inability of managers to reorganize and recover from significant stressors (Bodin and Norberg 2005
; Bunce et al. 2009
), highlighting the role of intersectoral collaboration and communication in the planning process.
Implementation occurs at various time, geographic, and administrative scales. For instance, implementation of strategies focused on hard infrastructure (e.g., changes in the built environment) will occur at longer time scales than those focused on changes in vegetation, outreach programs, and implementation of early warning systems. From an administrative perspective, implementation will take place via established networks, although adaptive management should result in more interdisciplinary, transsectoral implementation efforts. Implementation will require integration of several dynamic information streams tracking exposures (Webster and Jian 2011), population response to early warnings (Basher 2006; Ebi and Schmier 2005
; Kashyap 2004
), and assets available for response. A wide variety of decisions must be made at different administrative levels (Luber and McGeehin 2008
), such as how predictions will be made, what variables will be tracked, how warnings will be conveyed, thresholds for triggering warning messages (Hajat et al. 2010b
; Metzger et al. 2010
), and strategies for acting on preparedness plans (Balbus et al. 2008
) and communicating warnings (Ebi 2007
Monitoring provides data fundamental to learning in adaptive management (Holling 1978
). Monitoring should be planned early in the process (Ebi 2011
) and capture relevant data. Monitoring for EHE management would ideally capture shifts in exposures and modifying factors at various levels, changes in demographics, urban form, and outcome, such as heat morbidity and mortality rates. Syndromic surveillance of symptoms of heat-related illness can be analyzed in real time, for instance, to detect significant increases in these symptoms even before diagnoses are confirmed and reported to public health agencies, facilitating earlier response and ongoing changes in tactics as an outbreak progresses (Josseran et al. 2009
). Other exposure indicators are also important, sometimes using remote sensing (Johnson et al. 2009
). Other longer-term indicators should be tracked at larger geographic and administrative scales, if possible. In the United States, this might include health indicators that are or soon will be tracked at state and national levels (English et al. 2009
). Particular attention should be paid to vulnerable populations (Balbus and Malina 2009
). Monitoring should also capture system interactions and capacity. For instance, both short- and medium-term electrical power generation capacity are important determinants of EHE adaptation; although utilities monitor capacity, there is little coordination to increase public health preparedness.
Evaluation in adaptive management is explicitly focused both on the efficacy of the intervention (management objectives) and on increasing understanding of the system being managed (learning objectives) (Satterstrom et al. 2007
). This introduces the need for statistical support of pre- to postassessments in an iterative process, often involving Bayesian frameworks (Henriksen and Barlebo 2008
). Such pre- to postassessment is fundamentally probabilistic and requires both managers and stakeholders be educated on this approach, although it is often intuitive even for stakeholders without significant specific training (Webster and Jian 2011
Carrying through the extreme heat example, several issues can complicate evaluation efforts. Often multiple interventions are mounted concurrently, as was the case after the European heat wave of 2003, making it difficult to parse their relative contributions. Moreover, because of constantly shifting baseline conditions, it is difficult to generate baseline estimates of disease burden. However, comparing one extreme event with another can give some indication of efficacy, as with the 2003 and 2006 heat waves in Europe, where the later heat wave resulted in far lower mortality after significant prevention measures were taken (Fouillet et al. 2008
Adjustment is crucial to adaptive management. The adjustment phase is when future decisions regarding management and research are made, linking to the next cycle (). During adjustment, stakeholders are again actively engaged, results of the initial management decisions are conveyed, and stakeholders and system managers convey input regarding the next cycle. Adjustment is thus a process of information synthesis and communication as well as enhanced decision making and the point at which significant learning occurs (Bormann et al. 2007
). Adjustment also has important implications for the social integration of stakeholders, which has been shown to improve resilience to climate change in other sectors (Tompkins and Adger 2004
Adjustment is also where the cycle is at greatest risk. Reviews of adaptive management efforts have shown that inattention to key social learning elements—particularly rapid knowledge acquisition, effective information management, and explicit attention to creating shared understandings among diverse stakeholders—are key culprits (McLain and Lee 1996
). This is a concern in any discipline, but public health, with its emphasis on the social determinants of health and integration within community based organizations, has a set of tools for facilitating such processes (Baker et al. 2005
; Rowitz 2004
). Coupled with appropriate tools for managing information flow within and between organizations and a strong stakeholder commitment to the process, these tools are crucial for the adjustment phase.