An increasingly popular trend among news magazines (e.g., U.S. News, Forbes, Money Magazine, U.S. News and World Report, News Week) feature reports that aim to provide the public with convenient annual rankings of e.g., the best retirement cities, nation’s top 10 hospitals, greenest companies and top U.S. colleges. Often these rankings lead to competition among counties, states and college boards because of the potential economic gain (e.g., increased admissions, economic development or visibility).

An example of a health assessment tool which facilitates such comparative ranking is the University of Wisconsin’s Population Health Institute and the Robert Wood Johnson Foundation County Health Rankings providing information on a wide array of health outcome metrics for every county in all 50 states [3].

With the objective of informing and encouraging citizen participation in policy and land use decisions, SustainLane [4] and Earth Day’s Network [5], via publication with easily extracted datasets and a user guided web site, conducted assessments by ranking best performing based on metrics related to, for example, affordable housing, transportation, environment, health and long term susceptibility measures. Here, we further these efforts by extracting environment (e), health (h) and sustainability (s) metrics from these two sources, and combining them to construct broader based indices, denoted using upper case letters: “E”,“H” and “S” for the combined environment, health and sustainability indices, respectively. We also created an index combining the three indices, denoted “EHS”. In this paper, we provide an assessment of the how these measures interact with and influence each other as well as their relationships with sociodemographic measures for 50 U.S. Cities.

E Index was derived by averaging 3 separate scores for: air quality (e1), waste and toxics (e2) and water quality (e3). The H Index considered n= 9 (h) metrics. The S Index was derived by averaging 15 individual (s) metrics. Some of the S, H and E indice’s component (s) metrics had missing values. The missing values were not substituted with values so, in effect, this was the same as setting the missing metrics equal to the mean with non-missing metrics for that city (Figure 2).

Scores or rankings extracted from the 2008 Urban Environmental Report (UER) and SustainLane differed. For example, SustainLane ranked cities from 1 to 50 while UER’s scoring convention used scores 1–5. For consistency all variables including those from the 2010 U.S. census were normalized by sorting the variable values from worst (lowest rank) to best (highest rank) and assigning ranks (R) calculating where Φ represents the cumulative normal distribution function; N = number of values that were ranked. Among groups of cities with the same variable value (i.e., ties), we calculated the average Z and assigned it to each city in the group. For each E, H and S Index, the mean of all normalized variables contributing to the Index was then itself normalized to create the final E, H, S Indices. Finally the integrated EHS Index was derived from E and H and S indices using the following formula;

where the notation Z(V) stands for the variable with the normalized scores for variable V. For a city, the mean of all normalized variables contributing to an index was set equal to the mean of the non-missing variable values. All calculations including the standard normalization procedure described above were performed using SAS version 9.2.

Descriptive statistics (before normalizing) for the socioeconomic measures for the 50 cities are shown in Table 2. The total population represented by the various analyses was approximately 46.6 million people. The smallest and largest cities examined were Honolulu (337,256) and New York City (8,175,133), respectively. Cities with the greatest population gain over a ten year period (2000–2010) were Charlotte, North Carolina (+35.2%) and Fort Worth, Texas (+38.6%). In contrast, Detroit, Michigan and New Orleans, Louisiana lost population (−25.0% and −29.1%), respectively (Data not shown).

Figure 3 is an example of how the rank order of the top 10 best performing cities (based on the S Index) change if those same cities are rank ordered based on an (EH) Index.

For example; using the ten best ranked cities (based on City’s S Index); three of the 10 top performing cities drop in ranking to the lowest 10 performing cities when the performance ranking is defined by the (EH) Index, (A higher rank is better).

Figure 4 shows how the construction of the integrated EHS Index makes more difference for some cities, i.e., those cities that are farther (vertically) from the bulk of points than others, (i.e., those that are close to the X = Y line). The E indices (red squares) in general, are most unusual or inconsistent relative to the bulk of S and H indices. The graphic shows how individual E, H and S indices furthest from the X = Y line can be used by decision makers to identify those cities for which improvements in E, S and H indices would have the most relative impact. For example, Cities # 1 and 2 have a low integrated EHS Index (i.e., fall within the lowest 10 cities); yet rank highest with regard to the E Index (city 1) and or S Index (city 2). Thus, decisions aimed at improving H and or S metrics for city 1 would have the greatest impact on improving the overall integrated EHS Index ranking. Similarly, city (3 and 4) with a high performance ranking (based on the integrated EHS Index) would best benefit from policy decisions aimed at improving the E and S Index for city 3 and E and/or Health Index for city 4.

Circled cities have divergent indexes relative to the 50 cities examined. For example, Cities 1 and 2 have high ranks for one index and low ranks for the other two indexes. Cities 3 and 4 have high ranks for two Indexes but low ranks for one Index.

There are several limitations to our study. Although we observed statistically significant associations between sociodemographic features of the cities and the EHS Index, causality is limited because it was based solely on bivariate analyses. Further, other variables of importance were not considered e.g., city crime rate, annual energy consumption, unemployment rate, and availability of social and community services. Miranda et al. [19] argue that all policy decisions should include an analysis of social impact assessments. Secondly, the associations observed between S, H and E indices and their component metrics are likely more complex than the simple representation of the interconnections shown in Figure 2. Carlson [20] for example, suggests that when considering relationships e.g., between built environment and health, that the individual measurements that go into the analyses may not be directly correlated but instead may be correlated through a series of feedback loops that may regulate risk in different ways and different contexts. To understand how a change in one metric may result in changes of other metrics in other categories requires a more thorough evaluation of the processes that link these components.

Some of the data sources limitations are detailed in a paper by Lobdell et al. [21] who are developing an Environmental Quality Index for all 3141 U.S. counties. Briefly, these difficulties relate to finding data sources that track and present data at the same level of spatial and temporal aggregation. Earth Day Network’s methodology [5] indicates that, in some cases, complete data were not available and the scores used in the assessment analyses were not always presented at the city level, but rather at county, state or other relevant subunit and that defining clear city boundaries is not always possible. Our analyses do not take into consideration the multivariate and time-varying interdependencies and interactions between environment, sustainability and health metrics, nor did we evaluate important associations between component health and environmental metrics such as asthma rates and air quality. Such analyses might help define linkages between exposure and health outcomes measures. Further, we did not critically evaluate the approaches used by SustainLane and Earthday Network to develop their individual (e) (h) and (s) components and rankings. While this is an important analysis to conduct, it was beyond the scope of the current paper.

There currently is no single agreed upon, benchmarking approach, nor consistent set of metrics or indicators, that would allow communities to either track improvements or compare them with others with regards to short or long term sustainability goals. For example, Tanguay [22] in his analysis among 17 urban sustainability studies, reports the frequency of use of over 185 sustainability indicators. Niemeijer and Degroot [23] argue that indicators should be used only when it is clear that this can help inform decision-making. Here, we propose a benchmarking approach to allow cities to compare and rate their performance based on a broad array of environmental and sustainability and health metrics, many recommended by the WHO European Healthy Cities Network (WHO-EHCN) [24]. Our analyses demonstrate how socioeconomic, environmental, health and sustainability data can, when taken together with the expertise and local knowledge of community members, can be used to defend through scientific evidence, the impact of various policy changes. Bhatia and Corburn [25] argue that such assessments can be used to increase public awareness, encourage routine monitoring of more broad based determinants for health advocacy and accountability. These approaches may be used to track improvements made over time and/or to compare broad based indicators with those of other communities. Such comparisons could incentivize communication and sharing of successfully implemented policy decisions involving changes in specific environmental, health and sustainability measures.