Multi-risks attributed to climate change and urbanization in East Africa: a bibliometric analysis of a science gap

ABSTRACT

This study analyzed research on East Africa, multiple risks and Climate Change using bibliometric analysis. The main findings are that for many countries in East Africa, studies are absent, even on single risk assessments. Overall, multi-risk assessments that analyze hazard and impact chains are missing. Only a few cities have received scientific attention at the city level. The findings can help scientists as well as policymakers identify research blind spots as well as research-rich samples for further studies. This will be important for comparing regions, countries, or cities in East Africa in global assessments or science policy reports.

KEYWORDS:

• Climate change
• multi-risk assessment
• multiple risks
• disaster risk
• urban development
• literature review

1. Introduction

The interconnectedness of disaster and other risks is relevant to human life and health (O'Connor et al., 2021). Climate change is a driver of research and public awareness about extreme events overlaying and interacting with individual risks (IPCC, 2023). Scientific research about multiple risks is ongoing in health, sociology, natural hazards, and other areas. Different terminologies and framings are introduced to better understand the triggering effects of a single event culminating into a disaster (Greiving, 2006; Kasperson et al., 1988; Pescaroli & Alexander, 2015; Zscheischler et al., 2018). Risk is a combination of hazards or threats for an exposed human subject or system that, under conditions of vulnerability, can result in a disaster as a negative outcome, measured by loss of life, damages, or other tangible or intangible outcomes (UNDRR, 2022b). Despite an increasing interest in multi-risk assessments, including in urban and East African contexts, its applications are still wanting (Gallina et al., 2016; Garcia-Aristizabal et al., 2015). In global assessment reports, regions of the United Nations, such as East Africa, are compared (UNDRR, 2022a). Other global assessments and databases reporting climate change or disaster risk rely on scientific articles or country/regional profiles (BEH RUB, 2022; CRED and UCLouvain, 2023; IASC EC, 2022; IPCC, 2023). Reports such as Global Assessment Report 2022 (UNDRR, 2022a) also stress the importance of analyzing multiple risks, systemic risks, and their interrelations and of conducting regional comparisons. Therefore, it is important to analyze the scientific basis for such reports and how topics related to multiple risks and their connections are captured.

The interconnectedness of disasters and other risks significantly affect rapidly urbanizing economies. By 2050, more than two-thirds of the world’s population will live in cities (WEF, 2015). Thus, how well cities are governed will determine our ability to address global risks (Dilley, 2005). Countries in Asia and Africa are in the midst of a transition from predominantly rural to urban living, resulting in diverse urban growth patterns (Wolff et al., 2020). Challenges of rapid urban growth include health risks in slums (Oppong et al., 2015), environmental hazards, urban management, and agricultural land-use change (Marrengane & Croese, 2021). Increasing population density, unplanned development, rural-urban migration, and inadequate infrastructure bring profound risks to land degradation, potential water crises, the spread of diseases, and social instability. Inadequate infrastructure, climate change, poor health, and social instability are challenging cities (WEF, 2015). These risks compound, triggering multiple events and amplifying existing risks. Given these daunting challenges, how researchers identify multiple risks in cities is vital.

In this paper, we analyze the multiple risk aspects already encompassed in scientistific work to reveal recent directions and gaps in the research. Discussing it will help identify topics requiring more holistic approaches including stakeholders such as policymakers or affected people. To do so, we analyze scientific publications, which are purposefully crafted documents reflecting important research fields, often at the instance of funding agencies. Search platforms enable easy access and, therefore, comparability, though their use has several constraints as documented and discussed below.

We selected East Africa to get an overview and to enable comparison between different geographical regions, nations, and cultures. We used United Nations definitions of countries listed for ‘East Africa.’ This is an artificial setting of boundaries but enables comparison between and with similar statistical regions for follow-up studies.

International research in Eastern Africa is noted for multiple risks related to climate change, especially drought, poverty, and migration (Abebe, 2014; Apollo & Mbah, 2021; Haile et al., 2019; Wisner et al., 2012). Many daily risks of urban and rural populations are related to generating sources of income and fragile livelihoods related to environmental, economic, political, and cultural conditions (Doss et al., 2008; Omumbo et al., 2005; Shikuku et al., 2017; Wisner et al., 2004).

Research on multiple aspects of risk includes those related to disaster, risk reduction, climate change, and health. Within climate change research, there is a trend of compounding events indicating additional or parallel occurring hazard cascades (Zscheischler et al., 2018). ‘Pathways’ and ‘narratives’ are other terms used, especially in climate and global change studies, to express how causal chains can unfold. They are determined by human decision-making with regard to climate change and mitigation actions (IPCC, 2023). ‘Concurrent hazards’ is another term used to express what is sometimes referred to as secondary hazards (Pappadà et al., 2018; Quigley et al., 2020). In critical infrastructure research, interdependencies and cascading effects describe how initial triggers result in a chain of related connections in systems (Rinaldi et al., 2001). This is also related to research on feedback loops and system theory, or complex adaptive systems (Chorley & Kennedy, 1971; Gell-Mann, 1994). The interrelation of multiple risk factors, triggering social and environmental conditions, and long-term drivers have long been acknowledged in research related to East Africa (Baird et al., 1975). In social science research, the amplification of risk is a concept describing how risk factors on the hazard side and the societal side aggravate disaster processes (Kasperson et al., 1988). For example, the pressure and release model underlines the connection between root causes and drivers/processes that finally render people susceptible and exposed (Wisner et al., 2004). ‘Interconnected disaster risks’ is a recent flagship report by United Nations University, indicating the relevance of interrelated or multiple risk factors in a pathway sequence (EHSO'Connor et al., 2021).

The main objective of this paper is to analyze scientific literature to determine which topics addressed in eastern Africa are related to multiple risks. This can help identify existing research directions and indicate samples of information and studies sufficient to generate an overview of a topic. At the same time, it can help identify research gaps regarding countries, types of risks, or underrepresented topics.

We first analyzed several search terms from scientific literature about multiple risks. Research terms were analyzed in their occurrence within the Web of Science (WoS). We conducted cluster analysis of co-occurrence using the VOSviewer tool. We analyzed classes of topics at the global level and their prevalence for eastern African countries. We then selected three cities in different countries to understand local studies related to multiple risks.

2. Materials and methods

Bibliometrics, or quantitative statistical analyses of published work, has been used since at least 1917 (Lawani, 1981) for ‘the illumination of the processes of science and technology by means of counting documents’ (Pritchard, 1969). It uses information transfer networks and graph theory (Pritchard & Wittig, 1981) to analyze subjects of research or authorship (Lawani, 1981). Currently, bibliometric analysis is used to identify the evolution of fields in science or emerging trends (Donthu et al., 2021). Scientific databases such as WoS or Scopus and bibliometric software such as Gephi, Leximancer, or VOSviewer help in analyzing large volumes of data (Donthu et al., 2021).

Current developments in bibliometric analysis include a variety of sophisticated methods to analyze statistical or network relations through, for instance, synthetic knowledge synthesis (Kokol et al., 2022). Many of these methods apply machine learning and other techniques that go beyond the scope of this study, which used databases and tools easily repeatable and accessible for fellow researchers to compare and transfer the findings to other topics and regions.

We conducted a structured literature review using VOSviewer as a bibliometric tool (from Leiden University, Netherlands) to assist in keyword clustering. The objective was to identify the prevalence of research on East Africa on topics of multi-risk. A standard search platform, WoS, was selected since it is widely used (Birkle et al., 2020; Singh et al., 2021) and can therefore be used for comparison and follow-up studies.

We first selected search terms as they are typically used in risk studies, informed by snowball searching and reading articles about East Africa. The derived search terms were entered into WoS, and a literature search was performed for each term or for combinations of terms. (The search terms used in the WoS search are documented in the results section for every individual search term combination.) The search terms used were ‘natural hazard,’ ‘multi-hazard risk,’ ‘multi-risk,’ ‘multiple risk,’ ‘multiple risk’ AND Africa, ‘compounding risk,’ ‘climate risk,’ ‘climate risk’ AND Africa, ‘systemic risk,’ ‘societal risk,’ ‘societal resilience,’ ‘social resilience,’ ‘social risk,’ ‘social vulnerability,’ ‘social vulnerability’ AND Africa, disaster AND Africa, urbani* AND hazard, urbani* AND risk AND Africa, urbani* AND ‘disaster risk,’ disaster AND city AND Africa, and ‘critical infrastructure.’

‘Topic search’ was selected, limiting the search to titles, abstracts, keywords, and keywords plus (words from the article titles appearing more than once in the bibliography). The WoS search returned only articles that are peer-reviewed, in different languages but mostly in English. Several context features such as language, scientific discipline or field, citations over time, years, and country contexts were extracted in WoS. The country contexts were derived from author affiliations. The search was conducted on February 6 and 7, 2023, to extract a consistent data set since new articles are published worldwide daily. Articles from 1971 to 2023 were retrieved during this search. The export from WoS was a tab-delimited file, and the full record was selected. Only the first 1000 hits could be exported, but this was a sufficient sample size and represented the most relevant hits. VOSviewer was selected to cluster and visualize the most common keywords in the articles. The settings of the VOSviewer were kept at the default values so the findings could be compared with those of future studies. A minimum of five occurrences of a keyword was selected.

This methodology was selected to identify and analyze scientific literature in a standardized way, enabling follow-up studies in other regions or at later stages for comparisons. The analysis started at a broad regional level for East Africa, then differentiated findings for its countries, and finally narrowed down to three cities from three countries with the most findings in the search. For those cities, articles were manually reviewed to analyze content and whether or how they presented multi-risk aspects.

3. Results

3.1. Publication availability and gaps

The analysis of keywords from WoS related to ‘hazards’ and ‘risks’ in countries in East Africa revealed great heterogeneity (Annex, Table A1). Certain countries predominated with relatively high numbers of findings. We also found that certain terms, such as ‘climate risk’ and ‘multiple risks,’ were covered by more publications and countries than others. East African countries such as Uganda, Kenya, and Tanzania dominate the publications list. Only for those two search terms we also narrowed it down by adding the term ‘Africa’ to the search in WoS.

Overall, ‘multi-hazard’ and ‘systemic risks’ were hardly covered. For the search terms ‘multi-risk’ or ‘compounding risk,’ there were no findings in East Africa and, therefore, they are not listed in Table 1.

Table 1. Keywords related to citation topics meso (scientific field), entry year, languages, and sample size.

Keyword	Citation topic #1	Citation topic #2	Citation topic #3	Earliest entry year	Number of languages	Sample size
Hazards and risks
natural hazard	Climate Change	Geotechnical Engineering	Oceanography etc.	1976	9	2354
multi-hazard risk	Climate Change	Geotechnical Engineering	Concrete Science	2008	2	146
multi-risk	Climate Change	Psychiatry & Psychology	Economics	1971	4	197
multiple risk	Psychiatry & Psychology	Urology & Nephrology	Diabetes	1968	13	5664
multiple risk Africa	HIV	Urology & Nephrology	Psychiatry & Psychology	1996	2	92
compounding risk	Economic Theory	Economics	Climate Change	1992	2	109
climate risk	Climate Change	Oceanography etc.	Sustainability Science	1998	6	1461
climate risk Africa	Climate Change	Oceanography etc.	Economic Theory	2005	2	128
systemic risk	Economics	Ophtalmology	Knowledge Engineering	1992	8	3472
Societal aspects
societal risk	Safety & Maintenance	Climate Change	Combustion	1980	4	350
societal resilience	Climate Change	Political Science	Oceanography etc.	2000	2	134
social resilience	Climate Change	Marine Biology	Psychiatry & Psychology	1994	6	560
social risk	Psychiatry & Psychology	Obstetrics & Gyneology	Healthcare Policy	1971	12	2333
social vulnerability	Climate Change	Healthcare Policy	Nutrition & Dietetics	1964	11	3552
social vulnerability Africa	Climate Change	HIV	Nutrition & Dietetics	1999	3	95
disaster Africa	Climate Change	Oceanography etc.	Virology	1985	7	1204
Urban aspects and infrastructure
urbani* hazard	Climate Change	Geotechnical Engineering	Geochemistry etc.	1962	6	2277
urbani* risk Africa	Nutrition & Dietetics	Parasitology	Urology & Nephrology	1991	3	778
urbani* disaster risk	Climate Change	Geotechnical Engineering	Forestry	2003	2	199
disaster city Africa	Climate Change	Human Geography	Oceanography etc.	1994	4	111
critical infrastructure	Climate Change	Knowledge Engineering	Safety & Maintenance	1996	11	2618

Analyzing keywords related to social aspects of ‘risk,’ ‘resilience,’ and ‘vulnerability’ revealed similar patterns for countries that dominate research findings (Annex, Table B1). Most findings exist for ‘disaster Africa’ and ‘social vulnerability.’ The fewest occurrences exist for ‘societal risk’ or ‘societal resilience.’ ‘Social resilience’ has more findings than ‘societal resilience,’ which indicates research directions.

Analyzing keywords related to urban aspects (Annex, Table C1), we also included the topic of critical infrastructure since it adds a current trend on the topic tightly coupled to multiple risks. The findings again revealed a consistent pattern of certain countries having more findings than others. High numbers of publications also exist for ‘urban risk Africa,’ followed by ‘urban hazard’

3.2. Relation to fields of research

In analyzing the keywords related to hazards and risks (Table 1), we compared them among different academic fields of highest occurrence in the findings. In the area of hazards and risk, climate change is among the top three research fields. Another field that dominates is health and psychology in ‘multiple risk’ and ‘multiple risk Africa.’ Other fields are more heterogeneous, with oceanography, engineering, and psychology occurring repeatedly.

Analyzing the keywords related to social aspects (Table 1), climate change again predominates as a field of research. Other fields are quite heterogeneous. Analyzing the keywords related to urban aspects and infrastructure (Table 1) again shows climate change as a predominating field of research. Engineering occurs repeatedly. Urban risk in Africa is dominated by health.

3.3. Analyzing multiple risks

The research about multi-risk is manifold and cannot be easily summarized under a single umbrella term. One problem is the exact phrasing. We analyzed literature accessible in WoS (see the methods section) and identified 197 studies about ‘multi-risk’ as a search term and 5664 about ‘multiple risk’ (Figure 1).

Figure 1. Literature findings in the Web of Science for “multi-risk” (N=197, 1971–2023), top graph and “multiple risk” (N=5664, 1968–2023), bottom graph.

‘Multiple risk’ not only has many more findings but also sees a more continuous trend than ‘multi-risk.’ Both terms range back to the early 1970s, so it is not a new phenomenon in science (but has received growing interest). We analyzed the trends for all other search terms used in this study; all are steadily increasing. Thus, the production of papers overall is growing, and there is a growing trend of one term over the other. However, regarding the topics’ content, we found differences using VOSviewer clustering (Figure 2; see also Annex, Table E1).

Figure 2. “Multi-risk” clusters of co-occurrence of keywords (38 items, threshold 5), top, and. ‘multiple risk’ clusters of co-occurrence of keywords (40 items, threshold 15), bottom.

‘Multi-risk’ studies were related to the hazard domains of climate change, natural hazards, risk management, and insurance, as well as to children, behavior, and prevention. By contrast, ‘multiple risk’ studies were related to children, adults, and risk factors to health, such as alcohol; other health aspects predominate, and climate change or other hazards in this direction do not appear. The term ‘multiple risk’ was applied largely among medical sciences, compared to the ‘Multi-risk’ used in physical sciences.

3.4. Analyzing climate risk

The keyword clusters related to climate risk per continent or large state union (Table 2) reveal that all the regions analyzed shared many keywords in the highest occurrences. There were slight differences between the regions, and especially in Africa, topics about farmers, food, and security; related aspects of water and drought prevailed. The Americas, which are studies about Latin America, also had one cluster about drought. Double counting of ‘United States of America’ had to be removed. In the United States, agricultural aspects seemed to prevail, but, as was to be expected, no findings about poverty as in Africa. ‘Variability’ and ‘vulnerability’ were recurring topics in the remaining continents (Asia, Australia, and Europe) but were found in all.

Table 2. Keyword clusters fromKeywords related to citation topics meso VOSviewer related to ‘climate risk’ and region name.

Interestingly, there were many more studies in Africa than in other continents or regions, which we checked by another search (Annex, Table D1). The predominant research fields were climate change, oceanography, and multiple fields related to vegetation and agriculture, such as forestry, crop science, or soil science.

3.5. Analyzing disaster in Africa

The highest occurrences of keywords in the data set of ‘disaster Africa’ (N = 1204) were analyzed in VOSviewer (Annex, Figure A1). From these, the top hazards/threats and related disasters in Africa were climate change, drought, rainfall, conflict (poverty), floods, and war, with more than 20 occurrences each.

Selecting only East African countries (N = 152), the clusters (Figure 3) retained climate change and drought as the most prominent hazards. In addition, health impacts, governance and management, gender, and food security were topics within social impact aspects.

Figure 3. Clusters of keywords for East African countries (N=152).

3.6. Analyzing cities

Analysis of cities in East Africa revealed that certain topics, such as urbanization, risk, and disaster, returned many more findings than compared to climate risk (Annex, Table F1). Capital cities, compared to findings on second-largest cities, had much higher numbers of findings. This is different for cities such as Dar es Salaam or Mombasa, which are large cities. Regarding urbanization risks, the highest numbers of studies were found for Nairobi, followed by Dar es Salaam, Kampala, and Addis Ababa. For the term ‘disaster,’ the most findings involved Nairobi, Dar es Salaam, Kigali, and Harare. Originally, Kampala had six findings, but we removed one after a more in-depth assessment (see the section below). The names of the following cities were problematic since duplications of similar cities worldwide had to be excluded manually: Port Luis, Saint-Denis, Saint-Paul, and Victoria.

Three cities with the highest number of hits in both categories, disaster and urbani* risk, were selected for further analysis: Nairobi in Kenya, Kampala in Uganda, and Dar es Salaam in Tanzania. In VOSviewer, the keywords were adjusted differently per city and manually checked in different iterations to obtain the best representation of clusters emerging from the co-occurrence of keywords.

A review of the urbanization risks for the selected three cities (Dar es Salaam, Kampala, and Nairobi) revealed certain similarities and differences (Table 3). The hazards differed and were related to single disasters, such as terror attacks, geography, and exposure to floods or sea-level rise. Topics in all cities were related to population and health topics. Transmission of vector-borne diseases and poverty were topics in these urban areas and their management.

Table 3. Urbanization risks in Dar es Salaam, Kampala, and Nairobi.

Analyzing disasters per city (disaster And city name) revealed for Dar es Salaam a difference in classes related to topics of hazards or drivers of risk (Annex, Figure B1). Main hazard topics were climate change, health, and flood. Investigating ten sources in-depth, flood was a predominant topic in at least four papers. The others were more generic and unrelated to a single hazard (except to the context of climate change) but rather to sustainable development and social vulnerability. Two studies were deselected manually since they were about something other than the city. Regarding the impact and social side, general topics of a conceptual nature pointed to adaptation, resilience, vulnerability, management, and disaster risk reduction.

The clusters revealed multiple risks facing each city. This is substantiated by the Stimson Center’s Climate and Ocean Risk Vulnerability Index¹ (CORVI), a holistic assessment of Dar es Salaam’s climate risks (Stuart et al., 2021). Dar es Salaam is a sprawling city, home to an estimated 6.4 million people, and is a fast-growing city in East Africa. It is expected to grow to 13.3 million by 2035. Rapid urbanization is a key issue, coupled with unplanned settlements, with 75% of the population living in informal housing. Low-lying flood-prone areas make these informal settlements susceptible to climate risks. This issue is compounded with inadequate solid waste management. Solid waste regularly blocks drainage points, flooding neighborhoods and conducing the spread of vector-borne diseases (Stuart et al., 2021). Flooding in December 2011 and January 2012 affected around 10,000 people, with 40 fatalities and with the emergence of malaria in malaria-free regions, due to the prevalence of unplanned urbanization in Dar es Salaam (USAID, 2018). Urban areas largely rely on surface water sources. Unfortunately, rivers, lakes, and wetlands are unsafe sources threatened by wash mining and by commercial and domestic pollution. They are also affected in coastal cities by saltwater intrusion. These assessments correspond with research articles highlighting different risks and the interlinkages among them.

Analyzing Kampala, the two clusters revealed fire as the only hazard mentioned (Annex, Figure C1). Most other terms related to gender and children, displacement, and political and cultural aspects. Analyzing the studies by abstract, one paper was about fire; two were about displacement; one was about the Corona pandemic; and one was conceptually about resilience. One study about the USA was excluded manually. This published literature adopted disciplinary perspectives and rarely captured multiplicity.

Kampala has been struggling with urban heat and flooding. Localized rainfall, heavy encroachment on wetlands, uncontrolled run-off, and blockage along stream pathways significantly contribute to the flooding in the city. Flooding affects the main roads disproportionately, and secondary roads, which severely disrupts the economy indirectly and the access to basic facilities (Rentschler et al., 2019). The city is facing increased urban heat risk. Sseviiri et al. (2022) revealed that informal settlements and business corridors are hotter than surrounding areas. Highly dense settlements, poor vegetation cover, degrading ecosystems, air pollution, and the prevalence of critical transport hubs increase the vulnerability to heat risk. Heat risk is associated largely with individuals and workers dependent on the informal sector. It increases headaches, causes excessive sweating, dizziness, and dehydration, and increases health expenditures (Sseviiri et al., 2022). It disproportionately affects children, pregnant women, the elderly, the chronically ill, the homeless, and people working in the open. Heatwaves increase energy costs and air pollution. A multi-stakeholder workshop organized by the Development Network of Indigenous Voluntary Association (DENIVA) along with the Global Network of Disaster Reduction (GNDR) in 2022 highlighted the increase in disaster losses in the last 5–10 years in Uganda; the impact has been considerable among women, and there is lack of a coherent strategy to reduce risks, adapt to climate change, and reduce poverty (DENIVA, 2022).

For Nairobi, the clusters indicated that the guiding topics of disasters were climate change, health, earthquake, and terrorism (Annex, Figure D1). One study was manually deselected as it was unrelated to Nairobi. The majority of studies related to disaster and Nairobi discussed terror attacks. A few more, by comparison, were about health, communicable disease, and general conditions in slums and due to housing conditions.

Apart from hazards, topics in Nairobi involved social aspects related to science, policy, and knowledge of vulnerability and resilience (in the first cluster) and informal settlements. The second, third, and fourth clusters seemed to center around psychological aspects related to health and terrorism. These studies focused on single risks and prioritized strategies accordingly.

Nairobi is rapidly urbanizing, putting vulnerable people and workers at risk. Higher land surface temperatures are associated with areas surrounded by informal settlements combined with natural and anthropogenic factors such as dense housing and inadequate vegetation (Ochola et al., 2020; Scott et al., 2017). In Nairobi, over half residents of settlements live in a 3.5 by 4 m room accommodating at least five persons (Wanjohi, 2018).

Further, urbanization has grown into hazardous areas (Fekete, 2022). Fekete (2022) estimated that urban growth rates are 10- and 26-fold. However, growth into flood-exposed areas ranges from 2- to 100-fold. Tom et al. (2022) called for a comprehensive understanding of the flood dynamics at various scales for integrated city flood risk management.

4. Discussion

4.1. Methodology

Review metric analysis using VOSviewer is trending (Leal Filho et al., 2022; Matandirotya et al., 2022; Nyathi et al., 2022). This bibliometric tool enables a comparison between topics of research and global regents. The benefit of the methodology lies in the clustering analysis, and it is flexible in analyzing co-occurrences of keywords, co-authorships, or countries. It is helpful to visualize data, as doing so enables non-specialists to quickly capture main research areas. Limitations of bibliometric analysis and VOSviewer have been documented as well (Belter, 2015; Hajar & Karakus, 2022).

While using this tool, we observed several constraints, which may help in understanding our study’s limitations (as well as help other researchers). The WoS platform has obvious limitations that are known (Archambault et al., 2009; Yang & Meho, 2006). It is restricted to a certain number of large publishers and is not openly accessible to all researchers. However, it has a reputation for standardization and is often applied, which is why we selected it. We are also in a group of researchers analyzing East Africa (on other topics) that agreed to have a comparable data source. WoS collects multiple languages, but our assessment shows that English is predominant; other languages are more often represented in other databases. However, we quickly realized that we needed a large sample size for our study. This would not have been possible with other languages since they need to report a sufficient number of studies on our field of interest. Another constraint we encountered is the limit of exporting 1000 literature sources from WoS. However, 1000 was a sample large enough for cluster analysis, and it was also the case that the first 1000 findings were more relevant than the remaining sources, as we confirmed by reading. This limitation can be circumvented by dividing publications into more corpora, for example according to years, first exporting publications from i.e., 2018–2023, then 2010–2017, etc. In addition, WoS has a different database than, for example, Scopus. We used some of the same search queries (‘natural hazard’ and ‘multi risk’) in Scopus to identify differences. More literature samples were found in Scopus, which would lead to another corpus and hence, other interpretations. We therefore recommend that researchers consider this when conducting comparison studies. Again, WoS was used since we are part of a larger research group that had decided to use WoS for internal consistency of different studies.

A more general constraint in using assessment methods, such as bibliometric analysis, is that the information base consists of articles written subjectively by authors worldwide. That means that keywords, but also title words, and how they write abstracts differ widely. While some journals recommend abstract styles repeating certain keywords, most authors, in our experience, write papers quite differently. The selection of keywords to enter for an article closely relates to the framing of the authors, where they wish to find an audience. Therefore, a higher occurrence of a term, such as ‘climate change,’ may only sometimes direct researchers to a paper mainly about climate change. However, this may reflect the intention of researchers that their paper is important in relation to climate change.

The constraints of the VOSviewer tool are mainly related to cluster analysis. The clusters identified by co-occurrences are related to high repetitions of certain terms. We found that they cannot supplement reading and understanding individual papers. In our results about the city-level studies, we documented this and found that certain terms in the cluster analysis may be overrepresented and not match the real focus and number of articles on a certain topic, such as fire, earthquake, or similar disasters. The WoS findings also need to be manually checked.

4.2. Multi-risks

The study and its methodology were useful in generating a structured and comparable analysis of prevailing research. At the global level, ‘multi-risk’ is a current trend in climate change and natural hazard research (Gallina et al., 2016; Schmidt et al., 2011). But slightly shifting the phrasing to ‘multiple risks’ revealed a different picture, a much larger sample size, and a richer tradition of health and social studies. This is an important finding for researchers who want to direct their studies to appropriate audiences. It also helps in considering whether multi-risk research in the disaster and climate change community needs to better align with the topics of livelihoods, health, and social conditions. Despite a certain stream of research pushing in this direction within sustainability development or social vulnerability research, there still appears to be an imbalance between hazard and societal impact research.

At the level of eastern Africa, our study revealed that not only East African countries as such, but also that within the region (eastern Africa) specific countries are underrepresented. At the city level, representation related to cities’ size and international visibility. Our sample size was not large, but capital cities were overrepresented compared to the next largest city in the country in the analyzed topics. Climate change is a recurring topic within eastern African countries, while other types of natural (and non-natural) hazards are more heterogeneous. It is interesting to compare the occurrence of disasters as counted by loss of life, people affected, or economic loss in a specific database, such as EMDAT (CRED UCLouvain, 2023)(Table 4).

Table 4. EMDAT, top 20; region: East Africa, all disaster types (accessed 8 February 2023).

Year	Disaster Group	Disaster Subgroup	Disaster Type	Disaster Subtype	Event Name	Country	Total Deaths
1983	Natural	Climatological	Drought	Drought		Ethiopia	300 000
1901	Natural	Biological	Epidemic			Uganda	200 000
1973	Natural	Climatological	Drought	Drought		Ethiopia	100 000
1981	Natural	Climatological	Drought	Drought		Mozambique	100 000
2010	Natural	Climatological	Drought	Drought		Somalia	20 000
1973	Natural	Climatological	Drought	Drought		Somalia	19 000
1988	Natural	Biological	Epidemic	Bacterial disease		Ethiopia	7385
2008	Natural	Biological	Epidemic	Bacterial disease	Cholera	Zimbabwe	4276
2022	Natural	Climatological	Drought	Drought		Uganda	2465
1997	Natural	Hydrological	Flood	Riverine flood		Somalia	2311
1992	Natural	Biological	Epidemic	Bacterial disease	Cholera	Tanzania, United Republic of	2231
1997	Natural	Biological	Epidemic	Bacterial disease	Cholera	Tanzania, United Republic of	2025
1965	Natural	Climatological	Drought	Drought		Ethiopia	2000
1935	Natural	Biological	Epidemic	Bacterial disease	Bubonic	Uganda	2000
1998	Natural	Biological	Epidemic	Bacterial disease	Cholera	Tanzania, United Republic of	1871
1996	Natural	Biological	Epidemic	Parasitic disease		Zimbabwe	1311
1986	Natural	Biological	Epidemic	Bacterial disease	Cholera	Somalia	1307
1985	Natural	Biological	Epidemic	Bacterial disease	Cholera	Somalia	1262
1999	Natural	Biological	Epidemic		Cholera	Kenya	1237
2007	Natural	Biological	Epidemic	Bacterial disease	Cholera	Somalia	1133

Epidemics and droughts were less visible in the literature analysis, even while the loss and damage numbers in the EMDAT data document them. This might indicate a need for further research on whether studies on city- or country-level disasters are still underrepresented in eastern Africa.

However, we were interested in urbanization dynamics, and the study on cities showed the limits of bibliometric analysis in this area at present. The main problem was that there were too few studies, except for a few selected cities. Further, WoS does not cover studies from gray literature, leaving a gap in the literature search. And while it is quite amazing to find more than a dozen studies on certain disasters in just one city, this still limits the usage of bibliometric analysis. It is not an unexpected finding that an in-depth investigation of individual studies and on-site research needs to be conducted, as well as other forms and methods.

We identified three cities with relatively high availability of studies. On the one hand, this suggests that in-depth investigations into these three cities and countries. It is relevant to identify through on-site and field studies in these cities, which aspects of multiple risks in these cities are missing from the literature. On the other hand, the findings of this study can also prompt research in cities and countries that have few or no studies yet.

5. Conclusion

We conducted a literature review based on a bibliometric analysis of search terms and studies related to risk in East Africa. The main finding is that for many countries in East Africa, studies are lacking, even on single risk assessments. Overall, multi-risk assessments that analyze hazard and impact chains are missing. Only certain large cities have received scientific attention at the city level. Topics related to disaster risk, climate change, and urbanization are quite heterogeneous for both countries and cities. In some cities, daily health risks predominate, while other cities have multiple studies about particular natural hazards. The highest casualties in the region are recorded for droughts and epidemics.

Studies have taken a disciplinary perspective in understanding single risks. However, in many cities, multiple risks co-occur, which underlines the need for more interconnected research of a) daily and low probability risks and b) hazards and impact chains, across regions and sectors. Since topics also seem closely related to particular scientific fields or disciplines, the findings indicate that more research is needed in other fields. It could also be the case that certain research and funding trends, such as on climate change, influence the number of studies published or how the context and wording of articles are framed, but confirming this would need further research. The findings of our study can help scientists identify research blind spots and research-rich samples. It will also be important to know the prevalence of scientific studies when making policy because it could distort the representation of certain regions, countries, or cities for East Africa when comparing them with other regions in global assessments or science policy reports (IPCC, 2023; UNDRR, 2022a).

Acknowledgement

This paper emerged from a DKN (Deutsches Komitee für Nachhaltigkeitsforschung in Future Earth) working group on „Multiple Risks and Societal Resilience to Extreme Events“ supported by DKN and funded by German Research Foundation (DFG).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Deutsche Forschungsgemeinschaft.

Notes

1. It is an analytical tool designed to help governments, businesses, and financial institutions assess climate risks in coastal cities and pinpoint areas of action to adapt to current and future risks.

Annex

Table A1. Keywords related to hazards and risks.

Country in East Africa	Natural Hazard	Multi-Hazard Risk	Multiple Risk	Multiple Risk Africa	Climate Risk	Climate Risk Africa	Systemic Risk
Burundi
Comoros
Djibouti
Eritrea
Ethiopia	7		9	3	11	5
Kenya	4	1	9	4	31	15	1
Madagascar
Malawi	1	1	1	1	6	4
Mauritius
Mayotte
Mozambique			2	1	2	1
Réunion
Rwanda	1		2	1	5	3
Seychelles
Somalia
South Sudan
Uganda	2		4	2	10	4
United Republic of Tanzania			3	2	8	5
Zambia	2				3	1	1
Zimbabwe					8	7

Table B1. Keywords related to social aspects.

Country in East Africa	Societal Risk	Societal Resilience	Social Resilience	Social Risk	Social Vulnerability	Social Vulnerability Africa	Disaster Africa
Burundi							1
Comoros
Djibouti
Eritrea							2
Ethiopia	1			1	5		24
Kenya	1	2	4	3	9	6	54
Madagascar					1		1
Malawi				2	1	1	22
Mauritius			1
Mayotte
Mozambique					2		9
Réunion
Rwanda				1	2	1	11
Seychelles
Somalia
South Sudan
Uganda			1	3	3	1	21
United Republic of Tanzania			4	1	7	6	19
Zambia			1		1		9
Zimbabwe			2	1	9	2	20

Table C1. Keywords related to urban aspects and critical infrastructure.

Country in East Africa	Urbani* Hazard	Urbani* Risk Africa	Urbani* Disaster Risk	Disaster City Africa	Critical Infrastructure
Burundi
Comoros
Djibouti
Eritrea
Ethiopia	6	18	2		1
Kenya	9	57			3
Madagascar		4
Malawi	2	14	2	5
Mauritius
Mayotte				6
Mozambique		9
Réunion
Rwanda		1		2	1
Seychelles
Somalia
South Sudan
Uganda	3	24			1
United Republic of Tanzania	9	31	2		1
Zambia		10		1
Zimbabwe	2	5		1	1

Table D1. Keyword analysis of climate risk in different continents related to citation topics meso.

Keyword	Climate Risk Africa	Climate Risk America	Climate Risk Asia	Climate Risk Australia	Climate Risk Europe	Climate Risk United States
Citation topic #1	Climate Change	Oceanography etc.	Climate Change	Climate Change	Climate Change	Climate Change
Citation topic #2	Oceanography etc.	Climate Change	Oceanography etc.	Oceanography etc.	Oceanography etc.	Oceanography etc.
Citation topic #3	Economic Theory	Forestry	Soil Science	Crop Science	Forestry	Sustainability Science
Earliest entry year	2005	2005	2006	2004	2006	2006
Number of languages	2	1	1	1	1	1
Sample size	128	37	46	71	44	62

Table E1: “Multi-risk” clusters of co-occurrence of keywords (38 items, threshold 5) and“multiple risk” clusters of co-occurrence of keywords (40 items, threshold 15).

Table F1. Selected search term hits per most populous cities in East Africa.

Countries in East Africa	Capital City	Climate Risk	Disaster	Urbani* Risk	Largest or second largest city	Climate Risk	Disaster	Urbani* Risk
Burundi	Gitega				Bujumbura
Comoros	Moroni				Mutsamudu
Djibouti	Djibouti City			1	Ali Sabieh
Eritrea	Asmara			1	Keren
Ethiopia	Addis Ababa			17	Dire Dawa		2	1
Kenya	Nairobi	1	27	42	Mombasa		4	2
Madagascar	Antananarivo		2	3	Mahajanga
Malawi	Lilongwe		1	4	Blantyre			3
Mauritius	Port Louis			2	Vacoas-Phoenix
Mayotte	Mamoudzou
Mozambique	Maputo		1	3	Nacala
Réunion	Saint-Denis				Saint-Paul
Rwanda	Kigali		6	1	Muhanga
Seychelles	Victoria
Somalia	Mogadishu		2		Hobyo
South Sudan	Juba		1		Malakal
Uganda	Kampala	1	5	21	Mbarara			4
United Republic of Tanzania	Dodoma				Dar es Salaam		10	37
Zambia	Lusaka	1	2	8	Kitwe			1
Zimbabwe	Harare		6	7	Bulawayo		1

Figure A1. Occurrences of keywords in the search for ‘disaster Africa.

Figure B1. Keyword co-occurrence clusters for ‘disaster Dar es Salaam’ (N = 10).

Figure C1. Keyword co-occurrence clusters for ‘disaster Kampala’ (N = 5).

Figure D1. Keyword co-occurrence clusters for ‘disaster Nairobi’ (N = 27).