UN-HABITAT defines slums as urban areas where households lack one or more of the following conditions: access to an improved drinking water source, access to improved sanitation facilities, sufficient living area, durable housing in a non-hazardous location, and security of tenure [2]. As one might expect, these conditions can have severe consequences for human health and are of particular concern when considering their potential impact on the spread and burden of infectious diseases [3],[4]. For example, tuberculosis, influenza, meningitis, typhus, plague, typhoid and cholera are among many infectious diseases historically associated with conditions now common in urban informal settlements. In spite of this, few modern data are available assessing the current association of infectious diseases with ongoing global urbanization, especially at the neighborhood and household level in informal settlements [5],[6]. To begin to address this, we therefore analyzed urban environmental data and the burden of cholera in Dar es Salaam, Tanzania, a rapidly expanding African city.

Dar es Salaam, Tanzania's largest city, covers an area of approximately 1800 km² situated alongside the Indian Ocean. The city is comprised of three municipalities – Kinondoni, Ilala, and Temeke – which are divided into 73 wards (Figure 1). The city grew from a population of 76,000 in 1950 to a population of 3.31 million in 2008 [7]. With a current estimated growth rate of 4.3% per year [8], Dar es Salaam contains 29% of the country's urban population, though this percentage is expected to nearly double by 2010 [7]. Most of the city's growth has occurred along the coastline and along four main arterial roads [9] (Figure 2). Despite increasing efforts on the part of municipal and city governments to address informal settlement expansion, the speed of the city's growth and the inability to invest adequately in housing and infrastructure have led to the growth of existing settlements and to the development of new ones. As a result, Dar es Salaam now has one of the highest proportions of informal-settlement households in East Africa, with 65% of households living in informal areas [7]. Investigating the relationship between cholera incidence and the size of and conditions in informal settlements may help to shed light on some of the health consequences of settlement expansion.

Vibrio cholerae is a water- and food-borne Gram-negative bacterium and the cause of cholera, a severe, dehydrating diarrhea in humans. V. cholerae is unique among the diarrheal pathogens because of its ability to cause global pandemics. The disease has a short incubation period of 18 hours to five days [10], and can thus spread rapidly through a population. Cholera may cause explosive outbreaks in crowded conditions, such as occurred in Goma in 1994 [11] and in Zimbabwe in 2009 [12]. In addition to causing epidemic disease, cholera is also endemic in many areas of the world, especially sub-Saharan Africa and South Asia. If untreated, mortality rates due to cholera can be as high as 50% [13], though nearly all deaths can be avoided if replacement fluids are promptly administered. Antibiotics may also be used to shorten the recovery period for severe cases [14]. Strains of V. cholerae can be differentiated serologically by the O side chain of the lipopolysaccharide (LPS) component of the outer membrane, and the strains that produce epidemic cholera belong to serogroup O1 or O139. V. cholerae O1 itself is classified into two biotypes, classical and El Tor, which differ clinically and biochemically. V. cholerae O1 biotype El Tor is responsible for the current seventh pandemic of cholera, and is the current cause of cholera in Dar es Salaam [15]. Although reporting may be incomplete, cholera is one of the few diseases that require reporting to the WHO under the International Health Regulations.

Though a cholera epidemic was first reported in East Africa in 1836, no cases were reported across Africa between 1870 and 1970 [16], when cholera returned to the continent as part of the seventh cholera pandemic, which started in Asia in 1961 [10]. Cholera cases associated with this most recent pandemic were reported for the first time in Tanzania in 1974, and have been reported each year since 1977. The first major outbreak in Tanzania occurred in 1992, although the largest country-wide epidemic occurred in 1997, with more than 40,000 cases reported. This epidemic was reported to have started in Dar es Salaam [15]. Of all regions in Tanzania, Dar es Salaam has had the most cholera cases since 2002. In an outbreak in 2006, Dar es Salaam was the most affected region, accounting for 63% of the total cases (14,297) and 40% of the total deaths (254) [15].

Our observed association of cholera incidence with population density and the extent of informal settlement may also reflect in part the increased risk of acquiring V. cholerae recently passed by another human in a densely populated urban slum area, compared to the risk of acquiring V. cholerae from ecological reservoirs not recently contaminated by humans. V. cholerae can exist in environmental water sources independent from humans, especially in association with zooplankton and phytoplankton, and these organisms may serve as an important reservoir for infection [10],[31]. However, the passage of V. cholerae through a human intestine leads to a transient hyperinfectious bacterial state that can persist for up to 24 hours in environmental reservoirs, and such hyperinfectiousness may contribute significantly to human-to-human transmission [32]–[36].

Individuals in densely populated urban slums characterized by poor sanitary conditions may be at particular risk of ingesting such hyperinfectious V. cholerae, of therefore becoming involved in explosive outbreaks and epidemics, and of contributing to ongoing human-to-human transmission. Interestingly, recent data suggest that V. cholerae loses a significant degree of its hyperinfectiousness within hours of passage from a human intestine, suggesting that even in densely populated urban slums, relatively minor modifications in water usage patterns (such as securely storing water for perhaps as short as a day prior to ingestion to allow V. cholerae to revert to its non-hyperinfectious state) could possibly have significant impact on the burden of cholera in a slum area [37].

By suggesting that variations in the economic and physical conditions between and within informal settlements may translate into variations in disease patterns, this study validates previous findings that statistical analysis at the city level may mask important intra-urban differences. For example, infant and child mortality rates among Nairobi's slum populations were found to be three to four times higher than the city's average, and higher even than the average for rural areas in Kenya [38]. In Accra, Ghana, the relative risk of infectious and parasitic diseases was also found to be twice as high in areas of the city with the worst social and physical environmental conditions compared to the burden in the best areas [39]. Furthermore, this study adds to the growing body of research demonstrating that even within and between informal settlement areas, relatively small variations in social and physical conditions can result in markedly different rates of infection [18],[40],[41].

This pattern resembles the conditions resulting from rapid growth of European cities in the eighteenth and nineteenth centuries, which often led to a poorer health environment in urban areas [42]. However, the rate at which urban growth has been and is occurring in developing countries is unprecedented, and the resources available to address the myriad issues coupled with this growth are often extremely limited [4],[43]. Although urban-associated infectious disease burden will predominantly be borne by the poorest in many of the world's fastest growing cities, the increased movement of people within and between population centers and countries will also facilitate the spread of these diseases between poor urban centers and other areas and populations [3]. Again, using cholera as an example, between 2006 and 2007, more than 82,000 cases of cholera and 3,000 deaths were reported in Angola, the worst outbreak ever reported in that country [44]. The epidemic was reported to have started in one of the poorest and most overcrowded informal settlements in the capital city of Luanda and, over the course of the epidemic, spread to 16 of the country's 18 provinces [45].

Our study has a number of limitations. We analyzed cholera data collected during an epidemic, and patterns of endemic disease and factors involved in its transmission during endemic periods might be different than those we assessed. However, our use of data during the selected period allowed us to analyze a sufficient cholera case burden in the context of our measured environmental parameters. We used extant data sets reporting cholera burden that themselves were largely based on a syndromic classification system using WHO criteria. It is possible that cases classified as cholera where not caused by V. cholerae infection, and that other actual cases of cholera were not captured in the municipal reporting system; however, all cases included in this analysis met WHO criteria, microbiologic data were highly supportive when performed, and reporting criteria did not change during the period of data collection. In addition, though the negative binomial model used in this analysis accounted for high variability in the number of cholera cases, it did not account for any spatial structure that might have been present in the data. Given that our model tended to underestimate the number of cholera cases in the center of the city while overestimating the number of cases farther from the center, a spatial model might have provided additional insight, but was not able to be used given the small number of wards for which data were available. By using an extant data set, we were also limited by incomplete survey information; however, we felt that the ability to analyze an urban environmental survey undertaken immediately prior to an urban cholera epidemic provided a unique opportunity for analysis and, at least in part, mitigated this limitation.

Similarly, while we may have obtained a more robust understanding of the risk of cholera had we analyzed data at the household or individual level (as opposed to the ward level), such data were not available. Such an analysis may have shown that access to specific water sources or certain types of sanitary conditions or behaviors may affect the risk for cholera within a household. For instance, in a randomized control trial in low-income squatter settlements with poor water and sanitation in urban Pakistan, children in households encouraged to wash their hands with soap had a 53% lower incidence of diarrhea than in control households [46].

It should also be mentioned that we made the assumption in this analysis that people who became ill with cholera did so due to conditions in their area of residence, while it may also be the case that where people work and the types of activities they engage in away from home may have an effect on their risk of diarrheal illness. For example, consumption of food prepared away from the home is increasing in many developing countries, and was found in Nairobi to be associated with the distance one worked from the home. Preparation of these foods often involves questionable hygienic practices, which may increase the risk of diarrheal disease for those who consume them [47]. However, during an outbreak such as the one in 2006 in Dar es Salaam, the home environment may well play an important role in the spread of the disease.

In order to determine whether the relationship between environmental conditions and cholera in Dar es Salaam is unique or indicative of a more general relationship between environmental conditions and the risk for infectious disease, it would also be useful to compare the pattern of cholera incidence with patterns of other diseases across the city to see whether the nature and spatial pattern of risks are similar. For instance, one might predict that the risk for other enteric diseases such as shigellosis might similarly correlate with population density and sanitation, while tuberculosis and influenza might correlate with crowding, and vector-borne diseases such as arboviral infections and malaria might correlate more with water sources, vector biting habits, and human behavior.

Equally important, as the municipalities have taken important steps in recent years to address conditions in informal areas, evaluating the impact of these efforts on the incidence of disease should be a crucial part of determining their effectiveness. For instance, in at least one example of informal settlement upgrading, improvements in health have already been noted. Specifically, the upgrading of an informal settlement in Hanna Nassif ward in Dar es Salaam from 1994 to 1998, which focused on the installation of water vending kiosks, the management of solid waste collectors, and construction of improved access roads and storm water drainage channels, led to a 39% reported decline in waterborne diseases from the start to the finish of the project [48].