Dzudzor et al. BMC Public Health (2024) 24:888 https://doi.org/10.1186/s12889-024-18297-0 RESEARCH Open Access © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecom- mons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. BMC Public Health Food safety and dietary diversity in African urban cities: evidence from Ghana Makafui I. Dzudzor1*, Nicolas Gerber2* and Felix A. Asante3 Abstract Background Food safety is integral to food security and is increasingly becoming a significant concern in the urban areas of Africa, which are rapidly growing in population. In the case of Ghana, many urban households depend on tra- ditional open-air markets for most of their food needs. However, these urban food markets also depend on domestic food supply chains, which are prone to risks, including poor hygiene and sanitation and weather seasonality. Food safety compliance has associated costs which increase the unit cost of food products. Thus, higher food price is a risk factor to food availability and accessibility—fundamental pillars of food security. Method We use food microbial data and food retail data from food market surveys in major cities in Ghana to assess the safety of selected fresh food commodities and how retailers handle the food products they sell. Additionally, based on a two-wave balanced panel household data, we used fixed effects Poisson and Correlated Random Effects (CRE) Probit models to estimate the effect of weather seasonality on the incidence of diarrhoea and urban house- hold dietary diversity score (HDDS). A final sample of 609 households and 565 market respondents participated in the study. Results Our findings show that selected food samples tested positive for Staphylococcus aureus and E.coli and had aflatoxin B1 levels above 5.0 ppb. Additionally, the household incidence of diarrhoea/vomiting, a proxy for food safety status, is higher in the dry season. In the dry season, the household incidence of diarrhoea/vomiting increases on average by a probability of 38% points compared to the rainy season. Regarding HDDS, the average HDDS is 7.3; however, we did not find the effect of seasonality on HDDS to be significant. Conclusions Although urban food availability and household dietary diversity are not challenges for many urban households, food safety is a challenge in the major food markets in Ghanaian cities and is associated with weather seasonality. Foods available in traditional open-air markets are not always safe for consumption, undermining house- holds’ food security. Weak enforcement of food safety regulations contributes to the food safety challenges in Ghana- ian urban food markets. Keywords Food safety, Dietary diversity, Seasonality, Food markets, Urban *Correspondence: Makafui I. Dzudzor mdzudzor@uni-bonn.de Nicolas Gerber ngerber@uni-bonn.de Full list of author information is available at the end of the article http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/publicdomain/zero/1.0/ http://creativecommons.org/publicdomain/zero/1.0/ http://crossmark.crossref.org/dialog/?doi=10.1186/s12889-024-18297-0&domain=pdf Page 2 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Background The current global food system cannot provide healthy and safe diets inclusively and sustainably [1]. Poor diets account for a significant number of deaths, estimated, for instance, at 1-in-5 deaths in 2017 [2]. Poor diets, includ- ing the low intake of whole grains and fruits and the high consumption of sodium, accounted for more than 50% of the deaths related to diet [2]. Additionally, every year, 600  million and 420,000 people fall ill and die, respec- tively, from eating contaminated food [3]. Food safety has become a public health priority globally [4–6]. Food safety is the assurance that there are no adverse health effects from food prepared and consumed by an individ- ual for an intended purpose [7]. Food safety is intricately and inextricably linked to food security [8]. Food security “exists when all people, at all times, have physical, social and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life” [9]. Therefore, food safety underpins food security because there can be no food and nutrition security without food safety [10]. Food security has four foundational pillars—food avail- ability, accessibility, utilisation and stability [9, 11]. The recent addition of agency and sustainability to food secu- rity pillars highlights the importance of peoples’ rights to determine what they eat and the sustainability of the food system that produces the food people consume [11]. Food safety contributes to food security through food utilisation and stability pillars [11]. However, safe food has associated costs to produce it. The costs of invest- ing in equipment, monitoring and implementing food safety protocols increase the unit cost of production [12, 13], especially in developing countries where firms can be small, and the cost of food safety compliance can be expensive for the firm [14]. Ensuring food safety compli- ance adds to the food’s unit cost and selling price, thus impacting households’ food availability and accessibility. Non-compliance with food safety requirements can also lead to food availability and accessibility challenges through increased food waste and disposal [15, 16]. Dis- posing of unsafe foods contributes to food waste, and unsafe food disposal comes at a cost and thus affects food prices, leading to food availability and accessibility chal- lenges, and it is not environmentally sustainable [16]. For example, to comply with food safety regulations, firms must dispose of unsafe foods, and regulatory agencies discard entire shipments of food products because tested samples do not meet phytosanitary and other regulatory standards [16]. Therefore, food safety compliance not only benefits the consumer in terms of access to safe food consumption and promoting good health, but it also ben- efits the producer, who can benefit from the growing base of consumers willing to pay premiums for safer foods [17]. Additionally, the producer saves on cost by avoid- ing punitive fines from regulators, the cost of recalling unwholesome foods, reputational damage, and liabilities [15, 17, 18]. In addition to the food safety challenges facing food systems at various levels, the growing sophistication of consumer demands is increasing the complexity of food systems [1, 4, 5, 19]. The food system’s complexity is com- pounded by environmental factors like seasonal variations in rainfall and temperature [20]. Seasonality can affect household food consumption decisions and food safety. Seasonality influences household food consumption deci- sions via the availability of food varieties and diet com- position [21–23], food accessibility and price [23, 24] and loss of employment [25]. Seasonality shapes household dietary diversity through demand- and supply-side factors [26]. Seasonality and agricultural production are linked, especially in developing countries and rural areas where most agricultural production is rain-fed [27, 28]. Some foodborne diseases are linked to specific weather and cli- matic conditions and are prevalent at specific times and seasons of the year [29–31]. For example, the Salmonella transmission risk increases with high rainfall [30]. So tra- ditional open-air markets, which are the primary source of food markets for urban households in developing coun- tries [32, 33], can be environments where pathogens can easily find their way into food and water if not hygienically maintained [6, 34]. Depending on the level of market integration with the global food system, seasonality affects the availability and prices of some foods in the market [21, 24]. Households that live in big cities that are well connected to global food systems and have higher income levels have higher dietary diversity [35]. However, the pricing of food prod- ucts in some of these markets can also be too expensive for the urban poor, thus curtailing their dietary diversity [36, 37]. Developing countries are the most exposed to unsafe foods and food insecurity and the least equipped to manage them. Africa has the highest burden of food- borne diseases per population [3, 6]. Supermarkets are fast becoming regular features of food systems in devel- oping countries [38–40]. Although their impact on diet quality and diversity is mixed [38, 41–43], they are pro- moted for their improved quality and safety standards along the food supply chain [44]. Therefore, food safety is critical to food security by reducing the risk of foodborne diseases and enhancing food utilisation. However, improved food safety has the potential to increase the cost of food, thus affecting food availability and accessibility, leading to a low household dietary diversity score (HDDS) [45]. HDDS is house- holds’ access to and consumption of different varieties of food groups within a specified period. Therefore, HDDS Page 3 of 21Dzudzor et al. BMC Public Health (2024) 24:888 shows households’ ability to afford a variety of food groups and the availability of the food [46]. Additionally, weather seasonality affects food availability in the market, household dietary diversity, and food safety—manifested in food-related health outcomes like diarrhoea and vom- iting. Reviewing much of the existing literature on food security focuses on rural and agricultural households. It concludes that households are more food secure during the harvest season or soon after harvesting [47, 48]. Given the growing share of people living in urban areas and engaged in non-agriculture activities, the study high- lights the centrality of food safety in food security consid- erations and the seasonality of household food safety and HDDS in urban areas in developing countries. Empirical research combining food safety and availability within an urban food security context is limited. Therefore, the paper explores weather seasonality, urban households’ food safety status, and dietary diversity scores. We test the hypothesis that seasonality does not affect HDDS in urban areas with major food markets. We answer the question: What is the effect of seasonality on urban households’ dietary diversity score and the incidence of diarrhoea/vomiting due to food consumed? Methods Study area The study area is selected cities in Ghana (Fig.  1). The study was conducted in Accra, Kumasi and Tamale Metropolises in Ghana. According to Ghana’s 2010 Popu- lation and Housing Census, these cities are the biggest in Ghana’s southern, middle and northern parts. The three cities are cosmopolitan and have major food markets. Fig. 1 A map of Ghana showing the study sites Page 4 of 21Dzudzor et al. BMC Public Health (2024) 24:888 The different study sites provide different perspectives on the urban food system in Ghana. We presented a detailed description of the three cities in Dzudzor and Gerber [32]. The data used is mainly based on household and market surveys from Ghana. Ghana is a lower-middle-income country in West Africa, with a population of about 31 million [49]. In 2019, the country had a nominal GDP of about US$69 billion [50], with the service sector con- tributing the largest share of 47.2%, followed by industry with 34.2% and the agriculture sector with a contribution of 18.5% [50]. The agriculture sector remains critical to economic development and serves as the fulcrum for the government’s food and nutrition security agenda. Agri- culture remains highly dependent on rainfall. Ghana has different agroecological zones that influence economic and agricultural activities that dominate these regions [51]. As a developing country, urbanisation is on the ascend- ency. The proportion of the Ghanaian population living in urban areas as of 2021 was 57% [49]. Over the past century (1921–2021), Ghana’s urban population has rapidly increased from about 8% in 1921 [52] to almost 57% of the total population in 2021 [49]. Natural popu- lation growth and internal migration increase mainly drive Ghana’s urban population growth [52]. The growth in urbanisation has been geographically dispropor- tional. Almost half (47.8%) of the growth in urbanisa- tion between 2010 and 2021 occurred in just the Greater Accra and Ashanti regions [49]. Survey data and questionnaire Survey data This paper uses three data sets: household, market (retail) and fresh food microbial analysis data. The house- hold and market surveys were conducted in all three cities to explore the issues of food safety and nutrition, dietary diversity, and food consumption behaviour of households. Additionally, a food safety analysis was con- ducted to detect the presence of some foodborne patho- gens in food commodities when they arrive and are sold in the Agbogbloshie market in Accra. A total of 66 sam- ples were tested for different foodborne pathogens. A combination of the different data sources provides dif- ferent perspectives on the urban food system in Ghana. After data cleaning and management, 609 households and 502 retailers’ (market women/men) complete data were analysed (Table  1) across the study sites. A panel was developed with two rounds of household data to account for seasonality. The period between the two rounds of data collection was about six months (Novem- ber/December 2019 - June/July 2020). Between the two rounds, there was the Covid-19 outbreak that resulted in the partial lockdown of parts of Accra and Kumasi. As a result, households in some Enumeration Areas (EAs) in Accra and Kumasi, with a high concentration of migrants from other regions in Ghana, relocated to their hometowns. These reasons contributed to the fewer respondents in Accra and Kumasi in round 2 of the sur- vey. Tamale had no lockdown, so the same number of respondents were enumerated. The data collection process was done in line with the farming calendar in Ghana. There are two rainy seasons in the southern and middle parts (Accra and Kumasi) of Ghana: major season (April-June) and minor season (September-October) and in-between the dry season. In northern Ghana, there is a single season. The rainy season is June-August, and the dry season is Septem- ber-May. The data collection was modelled after these distinct seasons (rainy and dry seasons). The first round of data collection was done in November-December 2019 (dry season). The dry season is the harvest period of most staple foods in Ghana. The second data collection round was conducted from June to July 2020 (rainy season). This is the main cropping season for most crops. Survey questionnaire As part of the data collection process, the NOURIC- ITY project team trained enumerators to administer the questionnaires in the local language. During the train- ing, technical terms and concepts were discussed and explained in the local language to ensure all enumerators used the same nomenclature during interviews. After training, we pre-tested the questionnaire. We corrected challenges identified during the pre-testing to develop the final questionnaire we administered. Table 1 Data used in the study Study sites Round 1 Round 2 Household survey [number of respondents] Accra 216 175 Kumasi 240 218 Tamale 216 216 Total 672 609 Market survey [number of respondents] Accra 205 Kumasi 200 Tamale 160 Total 565 Food microbial analysis (Accra only) [number of food samples] Microbial analysis 43 Aflatoxin B1 analysis 23 Total 66 Page 5 of 21Dzudzor et al. BMC Public Health (2024) 24:888 The household questionnaire administered was struc- tured and had several sections. The sections related to food, safety and nutrition included household demo- graphics (household head, age, sex, marital status, edu- cation, household size, employment status, relationship with household head), food security (household food consumption in the last 24  hours and experience any anxiety of not having access to food in any month in the last 12 months), housing characteristics (type of house, number of rooms, type of floor, wall and roofing mate- rial, type of toilet facility, type of water source, lighting, and cooking fuel), household assets and income, well- being and food expenditure, food safety knowledge, atti- tude and practices (KAP) (based on the WHO’s five keys to safer foods) and self-reported health status, access to infrastructure, and household members’ health status related to diet. During the second round, the survey also included questions on the potential effect of Covid-19 on the availability and access to food, changes in the price of food commodities in the market and the employment status of household members. We utilised the same approach as the household survey to create the final questionnaire for the market survey. The market survey questionnaire consists of various sec- tions. It covered the respondent’s profile (including age, sex, education level, business experience and the type of product sold, such as vegetables, fruits, roots and tubers, grains, pulses, nuts, livestock, meat, dairy products, eggs and miscellaneous items), whether the respondent has migrated from another part of the country, type of busi- ness operation, reasons for choosing the business loca- tion, working hours and days, level of processed products sold; constraints faced by the business, the respondent’s food safety knowledge, and access to lighting and sani- tary facilities. For the microbial analysis, the team collected informa- tion on the lorry tracks when they arrive in the market, the wholesaler whose product it is, the location details of the wholesaler, and the retailers who come to buy from the wholesalers. The enumerators followed the retailer to their selling locations for follow-up data collection. We label all the products by date, name and type of seller before putting them in an ice chest and transporting them to the laboratory for analysis. Sampling design The data used in this study is based on the survey and sampling design employed under the Partnership for Healthy Diets and Nutrition in Urban African Food Sys- tems-Evidence and Strategies (NOURICITY) project. The NOURICITY project was designed and conducted in Ghana, South Africa, and Uganda to investigate the individual and systemic drivers and dynamics of urban food systems in Africa [53]. Regarding research activities in Ghana, we obtained ethical clearance from the Eth- ics Committee for Humanities, University of Ghana and the Ethics Committee of the Centre for Development Research (ZEF), University of Bonn. In all household and market interviews, we obtained informed consent from respondents. Market survey design We used a two-stage sampling approach. We used pur- posive and random sampling approaches in the first and second stages. We used purposive sampling to select three markets in Ghana’s south, middle and northern parts. We selected the markets from Accra, Kumasi and Tamale. The markets selected were the Makola and Agbogbloshie Markets in Accra, Kumasi Central Market in Kumasi and Tamale Central Market in Tamale. The criteria for selecting these cities are as follows: they have major food markets that are hubs for aggregating and redistributing a wide variety of food products to other cities and regions in Ghana and neighbouring countries. Also, these markets play essential roles in the national and regional food systems. Urban households depend directly and indirectly on these markets for their food needs. In addition, retailers in smaller and satellite mar- kets in these cities source many of their products from these major markets for onward sale in communities far from the major markets. The selected markets also pro- vide a reliable outlet for agricultural products from pro- duction (rural communities and towns) areas to be sold. In Accra, we identified the boundaries of the Makola and Agbogbloshie markets and conducted a mapping survey (retailer listing) of the types of food retailers and structures in the markets. Although the selected markets have thousands of actors, there is homogeneity (groups/ clusters) in the types of products sold and the struc- ture of the selling outlets. Therefore, we sampled based on products sold and structures in the markets. In the mapping survey, we randomly selected retailers within a particular food commodity cluster for fair geographical distribution—about 1000 retailers were sampled. During sampling, we also sampled retailers among a particular cluster other than the one in which we expected to find them. For example, when a vegetable seller is among cereal (maize) sellers, the vegetable seller is enumerated. After the mapping survey, we randomly sampled about 205 respondents for the market survey, which involved administering a more detailed questionnaire. Based on the experience from Accra, in Kumasi and Tamale mar- kets, we did a recognisance visit to the market to identify the main food clusters based on types of food sold and types of structures (e.g. wholesalers, retailers and immo- bile hawkers) and their location. After identifying the Page 6 of 21Dzudzor et al. BMC Public Health (2024) 24:888 clusters, we did random walks to enumerate respondents. We sampled 200 and 160 respondents in the Kumasi and Tamale Central markets. Fresh food sample collection technique We selected four food commodities, picked samples from the Agbogbloshie market in Accra, and performed microbial analysis to determine the presence of some selected foodborne pathogens. We selected tomatoes, cabbage, maize and groundnuts. The microbial analyses performed are total Coliform, E. coli, Staphylococcus, and Salmonella counts, as well as detection tests for Sal- monella spp. and Listeria monocytogenes. The selected foodborne pathogens are some of the most common pathogens that cause food contamination [8]. Further, given the high incidence of aflatoxins in cereals [54], the maize and groundnut samples were tested for Aflatoxin B1 (AFB1) concentrations. For the fresh food microbial analysis, we focused on traditional open-air markets. We traced the food prod- ucts from the point when they arrive in the market to the point where it is sold by the final retailers. We traced and collected samples over several days. The food samples were collected at various stages (sampling points) when the food commodities arrived in the market. The first samples were collected when the food trucks arrived at the market (when the trucks were offloading). We assume that samples collected at this stage will capture the con- ditions of the food commodities from the source of pro- duction through the transportation phase to the market. Therefore, we documented the wholesalers who received these goods on the first day. On the second day after the delivery day, we collected the second sample from whole- salers who received the commodities on the first day. The 2-day time span between the first and second sam- ples captures some of the market conditions that affect the food commodities (e.g. environment, sanitation and storage conditions). After another two days, we collected a final food sample from the retailers selling in smaller quantities (most customers buy from these sellers). These final samples are not necessarily from the initial trucks sampled, but they were samples bought by the retailers from similar trucks that delivered the food products on the same day the first samples were taken from the sam- pled trucks. Table  2 shows the number of samples collected from the Agbogbloshie market. The testing of the food samples was conducted by and at the Noguchi Memorial Institute for Medical Research (NMIMR), University of Ghana. The NMIMR tested the samples using their “Standard Operating Procedure (SOP) for enumeration and detec- tion of pathogens from food and animals (Bac-047-1.0)” [55]. The SOP outlines the procedures used to analyse the food samples collected. Similarly, they tested the afla- toxin levels in line with Aflatest® High-Performance Liq- uid Chromatography (HPLC) for corn, raw peanuts and peanut butter [56]. In total, 43 samples were collected and tested for selected food microbes. We also tested 23 samples for Aflatoxin B1. We focused on the common foodborne pathogens linked to sanitation, hygiene and storage. Unfortunately, due to budget constraints, this study could not collect samples from all three major mar- kets surveyed in the study. Therefore, although the total number of food samples collected from the Agbogbloshie market is not nationally representative, it indicates the levels of foodborne pathogens present in food commodi- ties sold in major food markets in Ghana. Household survey design We adopted a three-stage sampling approach. In the first stage, we used purposive sampling to select the three big- gest cities in Ghana’s south, middle and northern parts based on Ghana’s 2010 Population and Housing Census (PHC). In the second stage, we randomly selected a spec- ified number of Enumeration Areas (EA)—the lowest geographical units demarcated by the Ghana Statistical Service (GSS) for national population census purposes— within each of the three cities. Finally, in the third stage, we randomly selected households from each EA. A detailed description of the design of the NOURCITY household survey is presented in Dzudzor and Gerber [32]. Key dependent and independent variables Household diarrhoea/vomiting incidence The primary outcome variable in the study is house- hold incidence of diarrhoea/vomiting and illness from food consumption—the variable is used as a proxy for food safety. The variable is computed as a dummy (1/0) and a count variable. The dummy variable is 1 for households with reported diarrhoea/vomiting and ill- ness from food consumption over the last month and 0 otherwise. The incidence of diarrhoea/vomiting as a count variable is the number of cases of diarrhoea/ vomiting and illness from food consumption suf- fered by household members over the last month. The household incidence of diarrhoea/vomiting and illness Table 2 Total food samples tested for selected food pathogens Source: NOURICITY, 2020 Samples tested Tomatoes Cabbage Maize Groundnuts Total Microbial analysis 10 11 14 8 43 Aflatoxin B1 13 10 23 Page 7 of 21Dzudzor et al. BMC Public Health (2024) 24:888 from food consumption is based on the self-reported cases of households. The authors acknowledge multi- ple causes and sources of diarrhoea/vomiting [3, 57]. However, contaminated food and water are the most common sources of diarrhoea [8, 58]. Additionally, there is a positive correlation between food contami- nation and safety and the incidence of diarrhoea and vomiting [59, 60]. Household dietary diversity score (HDDS) HDDS is another key dependent variable. It is the num- ber of unique food groups the household consumed over a given period. The HDDS is based on a 24-hour recall period to improve the accuracy of the informa- tion collected. The HDDS consists of 12 food groups, which are their nutritional values-cereals; roots and tubers; vegetables; fruits; meat, poultry and offal; eggs; fish and seafood; pulses, legumes and nuts; milk and milk products; oil and fats; sugar and honey; and mis- cellaneous (e.g. condiments, coffee, tea). The HDDS ranges from 0 to 12 for each household, and the aver- age HDDS for the sampled group will be the proportion of the sum of all HDDS to the total number of house- holds sampled. In addition, the HDDS serves as a proxy to measure the socio-economic level of the household, given that a higher HDDS correlates positively with high-quality protein and household income [46]. Explanatory and control variables An explanatory variable in the study is weather sea- sonality. Weather seasonality is used as a dummy (1/0). Seasonality is assigned a 1 in the dry season (round 1), and 0 is assigned the rainy season (round 2). The next explanatory variable is the average monthly prices of staple crops (maize and tomatoes) in Ghana from 2013 to 2020. The prices are from the weekly food prices collected by ESOKO-Ghana from markets across the country, including the Agbogbloshie, Makola, Kumasi Central and Tamale Central markets. Another explana- tory variable is household food safety knowledge. We compute household food safety knowledge as a set of 11 true/false statements on household food safety knowl- edge. Each household’s total score indicates the level of household food safety knowledge. The questions are from the WHO’s “5 keys to safer foods” [61]. Other control variables include household characteristics like sex, age, education, marital status and employment sta- tus of household head; household size; proportion of household members employed; and household wealth status (a proxy for income) computed based on house- hold assets and housing characteristics and amenities. Empirical strategy We are interested in knowing the effect of seasonality on urban households’ dietary diversity and incidence of diarrhoea/vomiting. We use the regression models of the form: where yit is the respective outcome variables—HDDS and the incidence of diarrhoea/vomiting. The incidence of diarrhoea/vomiting is both a dummy and a count var- iable, and HDDS is a count variable. Subscripts i and t denote household observation and time (survey round), respectively. Season ( S ) is a dummy variable. P is a vector of prices of staples in Ghana. X is a vector of household characteristics. The coefficientα1 , measures the effect of seasonality on the outcome variables. We used fixed effect models to control for unobserved time-invariant variables that may influence the outcome variables and other covariates. We use the Poisson fixed effects model to estimate the count outcome variables (HDDS and the number of cases of diarrhoea/vomiting). For the effect of seasonality on HDDS [62–65], the Poisson model can be expressed as: where yit is the HDDS that varies across households ( i) and over time (t) . We assume the Poisson distribution to have a conditional mean (�it) , which depends on a vector of exogenous variables ( Xit) . According to Cameron & Trivedi [65], the conditional mean ( �it ) can be expressed as a log-linear model of the form: where Xit and Zi are vectors of time-variant and time- invariant exogenous variables, with β and γ as the respective vectors of parameters to be estimated, ǫi rep- resent unobserved household effects, and µt represents time-specific effects. From Eq.  (3), if the unobserved household effects ( ǫi ) are not correlated with any other covariate ( Xit andZi ), then we can use random effects panel estimators to achieve unbiased estimates [63, 65]. However, although we assume that weather seasonality is not correlated to other unobserved household charac- teristics, the unobserved household characteristics may correlate with other covariates in our model. For exam- ple, households’ skills, culture and attitudes towards food and health may correlate with their dietary diversity decisions (HDDS) and other covariates like household food safety knowledge, employment and income. For example, higher income correlates with higher HDDS [46], and other household characteristics like education and skills affect employment type and income earnings. (1)yit = α0 + α1Sit + α2X it + α3Pit + ∈it (2)Prob Yit = yit |Xit = e−�it� yit it /yit ! (3)In�it = βXit + γZi + ǫi + µt Page 8 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Under these conditions, the HDDS will partly depend on the unobserved variables leading to measurement error and endogeneity issues, and the estimated coef- ficients of HDDS suffer from selection bias [62, 63]. Therefore, we use household fixed effects to control for selection bias and eliminate time-invariant unobserved factors [62, 63]. Additionally, we use household wealth status instead of household income, which is less prone to endogeneity issues in the model [66]. To estimate the effect of seasonality on the incidence of diarrhoea/vomit- ing (dummy variable), we use Correlated Random Effects (CRE) Probit model. The CRE Probit addresses the inci- dental parameter problem associated with using Probit fixed effects [67, 68]. The incidental parameter problem arises in panel data analysis when running a non-linear regression (e.g. Logit, Probit) and the time (T) dimension is small (e.g. survey period = 2), and the number of obser- vations (cross-sectional units) is large (N→ ∞). Under such circumstances, only a fixed number of time periods are available to estimate the unobserved heterogeneity parameters for each cross-sectional unit and thus result in inconsistent estimates [67, 69]. The CRE approach accommodates time-constant variables and fixed effects estimates on the time-varying covariates [70]. The CRE estimation can be expressed as follows: where yit is the incidence of diarrhoea/vomiting status for household i at time t , xit is the time varying explana- tory variables of households, − xi is time averages of the time varying explanatory variables, β is the fixed effects estimate, ( ri + uit ) is a composite error term, ri is the time-constant unobservable variables, and uit is the idi- osyncratic error term. Adding the time averages ( − xi ) in the model controls for the correlation between the unob- served effects ( αi ) and the sequence{xit : t = 1,2} [70, 71]. For the robustness check for the effect of seasonality on the number of cases of incidence of diarrhoea/vomiting (count variable), we run CRE Poisson and the Poisson pseudo-maximum likelihood estimator with multiple lev- els of fixed effects (PPMLHDFE) for our estimation. We use the PPMLHDFE model because of the likelihood of a high number of households that did not experience diar- rhoea/vomiting (high number of zeros) in our sample and the non-convergence of the Poisson fixed effects model [72]. All regressions were done using STATA 15 [73]. Results Descriptive statistics Household demographics Households’ summary statistics are presented in a previ- ous study [32]. In summary, the majority of households are male-headed, the average age of a household head (4)yit = α + βxit + γ − xi + ri + uit is 47 years, and the average household size is 3.9. Some household characteristics are different across cities. A detailed description of household characteristics is pre- sented in Dzudzor & Gerber [32]. Market respondents’ summary statistics About 89% of the respondents in the market survey are women (Table 3). The average age of respondents is about 44 years. About 23% of the retailers do not have formal education. This percentage is exceptionally high in Tamale, where about 52% of respondents do not have formal education. Retailers in Accra have the highest proportion of migrants from other parts of the coun- try. On average, respondents have been engaged in their retail business for 15 years. The average expenditure per customer among small retailers is GHS14.35. Small retailers form the majority of respondents in the survey. Small retailers are immobile sellers selling food products on the floor, other materials (mats, rigid cardboard and polythene), and table tops. Most small retailers source their fruits and vegetables, cereals, meat and starchy sta- ples locally. They source most of their commodities from within the market from distributors/transporters who bring the products directly to the market. Respondents can access waste disposal bins, toilet facilities and run- ning water in the various markets. However, only 24% of respondents in the Tamale market have access to running water. In addition, market supervision by health and sani- tation officers of the local assembly is low. About 42% of respondents have never received any form of visitation from any sanitation officer since they started operating their business at their current location. Food prices and retailers’ food safety knowledge Food retailers’ food safety knowledge More than 55% of retailers know at least one safety meas- ure about their products (Table  4). However, there are variations in the proportion of sellers aware of safety measures in Accra, Kumasi and Tamale. Higher propor- tions of sellers of meat products, pulses and vegetables are aware of the food safety issues related to the prod- ucts they sell. Further inquiry into these food safety issues shows that they are mostly related to food preservation and how to maintain a longer shelf life of the products. This indicates that retailers are mainly driven by profit motives and not necessarily safety concerns. For exam- ple, most tomato and yam sellers indicated that “heat” (high temperatures) is unsuitable for their products, so they ensure they store them in cool and ventilated places. On the other hand, sellers of cereals and dry pulses were concerned about moisture. They mentioned that a moist environment makes moulds develop on their products Page 9 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Table 3 Market respondents’ summary statistics Note: Standard deviation in parenthesis for averages Variable Accra Kumasi Tamale Total % female respondents 92.68 89.00 82.50 88.50 Age(mean) 44.59 (10.72) 44.52 (12.27) 41.31 (13.38) 43.64 (12.14) % respondents with no education 12.68 11.50 51.88 23.36 % respondents who migrated to current location to do business 34.63 4.00 6.88 15.93 Average length of doing business (years) 13.61 (9.64) 15.60 (11.19) 14.46 (10.31) 14.55 (10.42) Sample size of all retailers (N) 205 200 160 565 Average purchase per customer (small retailers) [GHS] 15.28 (16.58) 13.51 (14.72) 14.28 (14.22) 14.35 (15.26) Sample size of small retailers (n) 121 127 90 338 Do you have access to: (%) Waste disposal 78.54 66.50 63.75 70.09 Toilet facilities 90.73 80.50 88.75 86.55 Running water 80.00 61.50 23.75 57.52 Visit from sanitation officers/inspectors (%) Never 29.76 47.00 51.25 41.95 Annually 32.20 13.00 16.25 20.88 Monthly 18.05 16.00 6.25 13.98 Weekly 6.83 6.00 12.50 8.14 Quarterly 7.32 8.00 5.63 7.08 Daily 1.95 9.00 1.88 4.42 Bi-annually 3.41 0.50 3.13 2.30 Fortnightly 0.49 0.50 3.13 1.24 N 205 200 160 565 Source of primary product sold (%) Within the market 42.93 65.50 29.38 47.08 Other sellers within the region 21.46 12.00 22.50 13.81 Other sellers outside the region 8.29 7.00 12.50 13.63 Other sellers within this community 14.63 7.50 16.25 12.57 Own production 10.24 7.50 14.38 10.44 Outside the country 2.44 0.50 3.13 1.95 Others 0.00 0.00 1.88 0.53 N 205 200 160 565 Most important cost constrain (%) Transportation 44.39 48.00 32.50 42.30 Staffing (wages) 1.95 0.00 0.00 0.71 Storage 6.34 10.50 10.00 8.85 Spoilage 12.68 17.00 18.13 15.75 Debts 9.27 6.50 14.37 9.73 Rent of trading spot 11.71 13.50 11.25 12.21 Rent of living space 0.98 0.00 1.25 0.71 Electricity 11.71 3.00 3.75 6.37 Other 0.98 1.50 8.75 3.36 N 205 200 160 565 Page 10 of 21Dzudzor et al. BMC Public Health (2024) 24:888 and decreases their shelf life, so they have to wrap their products and keep them on shelves to avoid contact with moisture and dust. Seasonality and price of food commodities Food prices show some seasonal trends (Fig.  2). Maize, cassava and rice prices show lower seasonal fluctuations than yam, plantain and tomatoes. Food price volatility is a significant issue in Ghana [74]. Critical causes include high dependence on rain-fed agriculture, poor stor- age facilities and lack of value addition and processing. Therefore, producers quickly flood the market with their products when they harvest to minimise spoilage (e.g. perishables like tomatoes), resulting in oversupply and a sharp price fall. Also, the food supply is low in the lean season, resulting in high prices. Additionally, there is a growing demand for staples like maize for alternative uses besides food. The poultry industry’s use of maize as feed [75] puts pressure on the price of maize in the mar- ket. The effect of these price changes on urban house- holds who buy most of their food commodities from the market is likely higher than their rural counterparts. Table 4 Awareness (self-reported) of food safety issues linked to main food product sold ANOVA conducted across cities. *** p<0.01, ** p<0.05, * p<0.1 Accra Kumasi Tamale Total p-value % N % N % N % N % of …retailers Vegetables 45.8 48 56.4 55 78.2 37 58.57 140 0.0090*** Fruits 33.3 3 47.1 17 71.4 7 51.85 27 0.4708 Roots/tubers/plantain 28.6 14 51.9 27 88.2 17 56.90 58 0.0021*** Dry grains 62.5 16 47.4 19 57.9 19 55.56 54 0.6608 Pulses 50.0 6 28.6 14 100.0 12 59.38 32 0.0003*** Starchy staples 55.6 18 20.0 10 81.8 11 53.85 39 0.0152** Meat (fresh meat) 37.5 8 70.0 10 77.8 9 62.96 27 0.2110 Total 46.90 113 49.34 152 78.57 112 57.29 377 0.0000*** Fig. 2 National average monthly prices of major food staples in Ghana, 2013–2020. Source: Authors’ construction, 2021 based on data from ESOKO-Ghana [77] and World Bank [78] Page 11 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Therefore, weather seasonality affects food prices, lead- ing to potential household food accessibility and dietary diversity challenges [76]. Food safety and foodborne microbial analysis From the microbial analysis (Table  5), no Salmonella spp. was enumerated from cabbage, tomatoes, maize or groundnut samples. Similarly, no E. coli was enumerated from cabbage, tomatoes and groundnut samples except for maize. One maize sample from a wholesaler had an E. coli level of 1.71 log cfu/25g. Staphylococcus aureus was seen in the sampled cabbage from one of the trucks and a groundnut wholesaler. Listeria spp. and Salmonella spp. were not detected in any of the 43 samples tested. However, Enterococcus faecalis was observed in one maize sample from a wholesaler, a tomato wholesaler and retailer, and groundnuts from a truck. The aflatoxin analysis found that both maize and groundnut samples contained high levels of aflatoxins (AFB1). The European Union (EU) has set a limit of 5.0 ppb for AFB1 in maize meant for human consumption. In comparison, groundnuts have a limit of 2.0 ppb for direct consumption and 8.0 ppb for those undergo- ing sorting and other physical treatment [79]. Accord- ing to Ghana’s National Policy for Aflatoxin Control in Food and Feed (NPACFF), Ghana has not set aflatoxin- acceptable limits for all food items [54]. However, it has set the limit for groundnuts to be 5.0 ppb. The results show that only one out of the thirteen maize samples had AFB1 levels below 5.0 ppb—maize sample from a truck had AFB1 value of 4.9 ppb—while all groundnut samples exceeded the limits of 2.0 ppb and 8.0 ppb, except for one sample from a groundnut retailer with an AFB1 value of 6.8 ppb. Table 5 Presence of selected foodborne pathogens in selected purchased food commodities Source: Summary based on results of samples tested at NMIMR a EU Commission Regulation (EU) 2023/915 of 25 April 2023 on maximum levels for certain contaminants in food and repealing Regulation (EC) No 1881/2006 has set the permissible limit at less than 5.0 ppb for maize, less than 2.0 ppb for groundnut for direct consumption and less than 8.0 ppb for groundnut subjected to sorting and other physical treatment b US Food and Drugs Administration set the total aflatoxin action limit to food and feed at 20.0 ppb A Microbial levels Commodity Sampling Point No. of samples tested E. coli (log cfu/25 g) Staphylococcus spp. (log cfu/25 g) Salmo- nella spp. (log cfu/25 g) Cabbage Trucks 3 0.00 3.99 0.00 Wholesalers 4 0.00 0.00 0.00 Retailers 4 0.00 0.00 0.00 Tomatoes Trucks 3 0.00 0.00 0.00 Wholesalers 3 0.00 0.00 0.00 Retailers 4 0.00 0.00 0.00 Maize Trucks 3 0.00 0.00 0.00 Wholesalers 7 1.71 0.00 0.00 Retailers 3 0.00 0.00 0.00 Groundnuts Trucks 3 0.00 0.00 0.00 Wholesalers 1 0.00 3.00 0.00 Retailers 4 0.00 0.00 0.00 B Aflatoxin B1 contamination Commodity Sampling Point No. of samples tested No. of samples with AFB1 permissible limits (> 5.0 ppb)a (> 8.0 ppb)a (> 20.0 ppb)b Maize Trucks 4 3 0 Wholesalers 6 6 1 Retailers 3 3 2 Total 13 12 3 Groundnuts Trucks 3 3 3 1 Wholesalers 2 2 2 0 Retailers 5 5 4 0 Total 10 10 9 1 Page 12 of 21Dzudzor et al. BMC Public Health (2024) 24:888 If these food products are used for food and ani- mal feed, the US Food and Drugs Administration (FDA-US) has set a permissible limit of 20.0 ppb for total aflatoxin (AFB1, AFB2, G1, G2) [80, 81]. How- ever, some samples had only AFB1 levels exceeding 20.0 ppb. Three of the thirteen maize samples (23.1% of total samples) had AFB1 concentrations above the permissible limit of 20.0 ppb. One of the wholesalers who processed maize into corn dough1 had an AFB1 concentration level of 23.1 ppb, while two of the three maize retail samples had AFB1 concentrations of 25.3 ppb (corn dough) and 33.4 ppb (maize grains). Simi- larly, one out of ten groundnut samples had AFB1 above 20.0 ppb—a sample from one of the groundnut trucks had an AFB1 level of 27.3 ppb. The other raw, roasted, and paste groundnut samples had AFB1 levels lower than 20.0 ppb. Household dietary diversity and food related health status Household dietary diversity score (HDDS) HDDS is presented in Fig. 3; Table 6. Figure 3a presents the percentage of households consuming different num- bers of food groups in the dry and rainy seasons. Most households consumed between 6 and 9 food groups dur- ing the two periods. However, HDDS was higher in the rainy season (7.5) than in the dry season (7.0) (Table 6). Accra and Kumasi have similar and higher HDDS over the two periods compared to households in Tamale. For the specific food groups consumed (Fig. 3b), cereal and cereal products, oils and fats, and sugar and honey products were consumed by more than 50% of house- holds in both seasons. Most households also consumed white tubers and roots. The share of households that consumed oil and fats, and sugar and honey products increased by more than 10% in the rainy season. Simi- larly, the share of households that consumed vegetables increased by about 9% in the rainy season. The share of households that consumed fruits, meat, offal and poul- try, and dried beans, nuts and seeds were broadly similar between seasons. Household health and diet Table  7 presents households’ illnesses related to food contamination. More households suffered diarrhoea/ vomiting in the dry season than in the rainy season. About 9 and 8% of households suffered from diarrhoea or vomiting in the dry and rainy seasons, although this difference is not significant. More households also suf- fer illnesses from consuming food away from home than home-cooked food. On average, 10% of households suf- fered from illnesses related to food consumed away from home compared to about 3% of households who suffered from illnesses related to food consumed at home. The dif- ference in illness attributed to food consumed away from home and food consumed at home is statistically signifi- cant, indicating a lesser food safety status of food away from home. Effect of seasonality on household incidence of diarrhoea/ vomiting Table 8 presents the results of Correlated Random Effects (CRE) Probit estimations. From the analysis, weather seasonality affects the incidence of diarrhoea/vomiting among urban households when controlling for household fixed effects. Columns 1 and 2 present the regression results with and without households’ self-reported effect of Covid-19 on food prices, respectively. The results show that the incidence of diarrhoea/vomiting in sampled urban households is higher in the dry season compared to the rainy season. The results show that all things equal, in the dry season, the incidence of diarrhoea/vomiting increases by a probability of 0.38 compared to the rainy season. The literature on diarrhoea infections in Ghana shows that diarrhoea is seasonal and that children are the most vulnerable. The results also show that the incidence of diarrhoea/ vomiting varies across household heads’ educational levels. In households where the household head has a primary education, diarrhoea/vomiting decreases by a probability of 0.158 compared to households with no formal education. Furthermore, the results also show that the price of maize is positively associated with the incidence of diarrhoea/vomiting. All things equal, a one- unit change in the price of maize increases the prob- ability of the incidence of diarrhoea/vomiting by 1.6. The incidence of diarrhoea/vomiting is higher in households that said Covid-19 affected the price of staple foods than those who did not. The contrary was observed for house- holds who said Covid-19 affected the prices of vegetables. We checked the robustness of our results on the asso- ciation between weather seasonality and the incidence of diarrhoea/vomiting (number of cases of diarrhoea/vom- iting) (Supplementary Table 1, Additional file 1). We run CRE Poission and, due to the high number of households that did not experience diarrhoea/vomiting (high number of zeros), we applied the Poisson pseudo-maximum like- lihood regressions (PPML) with multi-way fixed effects [72], which can control for the high number of zeros in the estimation. The CRE Poisson and PPMLHDFE results show a positive and statistically significant difference in 1 Corn dough is maize that is soaked in water for about 2 days, drained and milled into fine flour. Page 13 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Fig. 3 A Household Dietary Diversity Score; B Food groups consumed by households over the past 24 h by season Page 14 of 21Dzudzor et al. BMC Public Health (2024) 24:888 the incidence of diarrhoea/vomiting between seasons: the number of cases of diarrhoea/vomiting is higher in the dry season than in the rainy season. This is consistent with the results of the CRE Probit estimation (Table  8). Compar- ing the magnitude of the coefficients of CRE Poisson and PPMLHDFE show that PPMLHDFE has a higher magni- tude for the association between weather seasonality and the incidence of diarrhoea/vomiting. The difference in the magnitude of the coefficients can be attributed to the abil- ity of the PPMLHDFE to handle the high number of zeros in the model. Effect of seasonality on HDDS Table 9 shows the effect of weather seasonality on HDDS. We present the results of a Poisson fixed effects estima- tion for HDDS. We did not compute marginal effects for the Poisson estimation because the coefficients can be interpreted as semi-elasticities [65]. The results indicate that weather seasonality does not have a statistically significant effect on urban households’ HDDS when household fixed effects are controlled. Other regression results show that large household sizes positively affect HDDS. A unit increase in household size is associated with a 3.4% increase in HDDS. Further- more, the household wealth index has a positive effect on HDDS. Wealthier households eat more diversified foods. A unit increase in wealth index would result in 3.6% increase in HDDS. It implies that a marginal increase in households’ wealth status increases the number of food groups households consume by 3.6%. Household knowl- edge of food safety is positively associated with HDDS. A unit increase in household food safety knowledge will translate into a 0.4% increase in HDDS. This result may be attributable to the general positive effect of increased knowledge of food safety. However, there was no statisti- cally significant difference in HDDS of households who attributed price changes in fruit and vegetables and sta- ple foods to Covid-19 and those who did not. We assessed the robustness of our results using differ- ent estimation approaches. We ran pooled Poisson and CRE Poisson estimations for HDDS (Supplementary Table  2, Additional file  1). The pooled Poisson results show that seasonality affects HDDS. HDDS is lower in the dry season compared to the rainy season. All other things equal, the coefficient indicates that HDDS decreases by 5.4% in the dry season compared to the rainy season. When using CRE Poisson, HDDS is lower in the dry season; however, the difference is not statisti- cally significant. Therefore, the Poisson fixed effects and CRE Poisson results are consistent. Discussion Food safety in cities in Ghana Our market and microbial food analysis results show that food safety is a challenge in major food markets in cities in Ghana. Foodborne pathogens like Staphylococ- cus aureus, Enterococcus faecalis, E.  coli and aflatoxins (AFB1) were present in selected food commodities at dif- ferent sampling points in the market. Some of the tested foodborne pathogens are present before the food com- modities get to the market, and the market environment also potentially introduces or spreads other pathogens. The presence of Enterococcus faecalis in some samples is indicative of faecal contamination. The evidence of fae- cal matter contamination in some food samples raises Table 6 HDDS by season and cities Significance level: *** p<0.01, ** p<0.05, * p<0.1 Superscript numbers indicate how the difference between dry and rainy seasons are computed (Diff. across rounds (2-1)) Cities Mean scores Diff. across rounds ( 2–1) Dry season1 Rainy season2 Total HDDS Accra metropolis 7.091 7.577 7.334 0.486** Kumasi metropolis 7.151 7.638 7.394 0.487** Tamale metropolis 6.907 7.356 7.132 0.449** Overall (N = 609) 7.048 7.521 7.284 0.473*** Table 7 Household health and diet related illness *** p < 0.01, ** p < 0.05, * p < 0.1; t-test of diff. between illness resulting from food consumed away from home and food cooked at home are statistically significant The superscripts indicate how the difference (2-1) or (a-b) are computed Dry season1 Rainy season2 Total Difference (2 − 1) Suffered diarrhoea or vomiting (%) 9.195 7.882 8.539 -1.313 Illness related to food consumed away from home (%)a 10.345 10.181 10.263 -0.164 Illness related to food consumed at home (%)b 2.299 2.956 2.627 0.657 Difference between (a-b) 8.046*** 7.225*** 7.635*** Total number of households (N) 609 609 1218 Page 15 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Table 8 Effect of seasonality on households’ incidence of diarrhoea/vomiting Variables 1 2 Coef. AME Coef. AME Season (Dry) 2.499** 0.378*** 2.453** 0.383*** (1.061) (0.134) (1.034) (0.134) Characteristics of household Age of household head 0.005 0.001 0.019 0.003 (0.053) (0.007) (0.054) (0.007) Sex of household head (male) -0.838 -0.126 -1.271 -0.209 (0.823) (0.141) (0.852) (0.164) Household size 0.036 0.005 0.016 0.002 (0.164) (0.022) (0.166) (0.023) Education of household head* Primary -5.608*** -0.158*** -4.991** -0.158*** (2.021) (0.008) (1.957) (0.008) Secondary -1.872 -0.330 -0.966 -0.155 (1.728) (0.329) (1.600) (0.303) Tertiary 0.199 0.029 0.991 0.205 (2.526) (0.406) (2.443) (0.671) Household wealth status Lower-middle 0.017 0.002 -0.078 -0.010 (0.302) (0.041) (0.298) (0.039) Middle -0.167 -0.021 -0.204 -0.026 (0.349) (0.042) (0.346) (0.041) Upper-middle -0.425 -0.049 -0.445 -0.051 (0.382) (0.037) (0.386) (0.037) Upper 0.125 0.018 0.152 0.022 (0.466) (0.068) (0.466) (0.071) Household food safety knowledge 0.007 0.001 0.007 0.001 (0.008) (0.001) (0.008) (0.001) Marital status of household head Single -0.321 -0.038 0.235 0.035 (1.438) (0.148) (1.353) (0.225) Monogamous 0.516 0.068 0.591 0.080 (0.885) (0.118) (0.849) (0.116) Polygamous 3.570 0.838*** 1.825 0.486 (3.146) (0.250) (3.068) (0.999) Price of maizea 11.977** 1.612** 12.638** 1.726** (5.464) (0.722) (5.437) (0.733) Price of tomatoesa 0.173 0.023 0.137 0.019 (0.192) (0.026) (0.184) (0.025) Employment status Employment status of household head 0.122 0.016 0.193 0.024 (0.344) (0.042) (0.343) (0.040) Percent of household members employed 0.009 0.001 0.008 0.001 (0.006) (0.001) (0.006) (0.001) Self-reported covid-19 effect Affected price of staple foods 0.613** 0.103** (0.262) (0.052) Affected price of vegetables -0.772** -0.072*** (0.356) (0.021) Page 16 of 21Dzudzor et al. BMC Public Health (2024) 24:888 concerns about the safety and wholesomeness of water used for cultivating vegetables, the personal hygiene of the transporters and the vehicles in which these products are transported and the hygiene and sanitation condi- tions that pertain in the market. E. coli and Enterococcus faecalis are linked to cholera and diarrhoea [57]. All the sacks of groundnut had high levels of aflatoxin contami- nation before arrival in the market; this indicates that aflatoxin contamination occurs during the production and transportation stages of the supply chain. Also, the high aflatoxin levels in maize at different sampling points show that aflatoxin contamination can occur at any stage of the supply chain if the products are not correctly han- dled. This is because samples taken from all three maize trucks had AFB1 concentrations less than 20.0ppb, but as the products were stored with other bags of maize and processed into corn dough, higher aflatoxin contamina- tion occurred. Mixing different sacks of maize from dif- ferent sources during storage can lead to contaminated maize, affecting good maize that may have arrived from the farm. From the self-reported responses of retailers, food retail- ers have limited food safety knowledge. Although about 57% of sampled food retailers have some food safety knowledge of the commodities sold, this knowledge is mainly linked to food preservation and how to extend the shelf life of their commodities. The implication is that reducing associated costs is important to food retailers relative to other food safety considerations. Most food retailers sell the same types of commodities throughout the year. Thus, they know how to handle their products under different environmental conditions to optimise their profits. Selling the same types of food commodities throughout the year also indicates all- year food availability. However, because of seasonal price changes, urban households may suffer from food accessibil- ity (affordability), adversely affecting the food security status of urban households, especially the vulnerable. Households whose food security status is threatened by higher prices are likely to shift to the consumption of lower quality foods, including the consumption of foods low in dietary diversity and unsafe foods, which increases the risk of food poison- ing, disease and malnutrition [82, 83]. Within the household, weather seasonality affects households’ food-related diarrhoea/vomiting infections (food safety). Households in the dry season recorded more cases of diarrhoea/vomiting than in the rainy sea- son. Previous studies in Ghana [84–86] are in tandem with our results that the incidence of diarrhoea is higher in the dry season and also influenced by the wealth sta- tus of households and source of food purchases [59]. The seasonal incidence of diarrhoea is also consistent with the literature that shows that the quantity of water sup- ply and water availability to households strongly affects their WASH behaviour and health outcomes [87, 88]. So, households suffer from food-related diarrhoea/vom- iting in the dry season, potentially because inadequate or inconsistent access to improved water heightens the risks of noncompliance with WASH behaviour [89]. So, a constant safe water supply improves households’ WASH behaviour and health outcomes. Our findings are incon- sistent with studies by Anyorikeya et  al. [90] and Asa- moah et al. [91], who found the incidence of diarrhoea to be higher during the rainy season in their study areas. The incidence of diarrhoea/vomiting is lower among household heads with a higher formal education than those with no formal education. Kumi-Kyereme and Amo-Adjei [92] found that in Ghana, children in house- holds with higher wealth status and mothers with higher formal education had lesser odds of suffering from diar- rhoea. Generally, socioeconomic disparities significantly influence households’ access to resources, including healthcare. Therefore, as households’ socioeconomic status (including wealth status and education) improves, incidences of diarrhoea are likely to decline [93, 94]. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 Coef.-CRE Probit coefficients; AME-Average Marginal Effects *Reference base for educational level is “No formal education” a Real price of maize and tomatoes are computed based on ESOKO-Ghana December and June price averages from 2013 to 2020 Table 8 (continued) Variables 1 2 Coef. AME Coef. AME Constant -1.862* -1.720* (1.005) (1.000) Time varying averaged regressors Yes Yes Yes Yes Self-reported covid-19 effect Yes Yes No No Number of observations 1212 1212 1212 1212 Number of unique households 606 606 Page 17 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Similarly, increases in the price of staple foods like maize imply that food-insecure households adopt coping strate- gies that may compromise their food quality. They shift to consuming food of lesser quality leading to increased risk of disease and essential nutrition deficiency [79]. Households attributed suffering more food contamina- tion from consuming food outside the home than from eating home-cooked food. Also, households with more employed people outside the house tend to eat food cooked outside the home. There are hygiene and food safety concerns in Ghana regarding food vendors and eat- eries. Major food contamination incidences have occurred at both high-end restaurants and street food vendors, with some resulting in deaths [95, 96]. In addition, quality food outside the home is expensive [97]. So, consumers may rely on the trustworthiness and reputation of the food out- let [98] to continue patronising their food so long as they do not fall sick. Thus, consumers face a higher risk of food contamination in Ghana when patronising food outside the home [59]. Food availability and HDDS in cities in Ghana Our household results also show that the proportion of households that consume different food groups varies between the dry and rainy seasons. The proportion of households that consumed animal-based protein foods (egg and fish) declined in the rainy season, while meat, offal and poultry increased marginally. The greater avail- ability and affordability of green leafy vegetables during the rainy season increased household consumption, mak- ing them a healthy option to add to food dishes. How- ever, the increase in the share of households consuming oil and fats, and sugar and honey products are not nec- essarily healthy diet options. Consequently, avoiding and eliminating industrially-produced trans-fatty acids in the food system should be promoted [99]. In addition, balanced consumption of sugar and healthier oils, like olive, canola, coconut, and avocado, should be promoted and made affordable because oil and fats, and sugar and honey are common ingredients in most dishes in devel- oping countries. When controlling for household fixed effects, we did not observe a statistically significant effect of weather seasonality on HDDS in urban Ghana. This may be because urban households, especially in major cities, pur- chase most of their food from traditional open-air mar- kets [32, 33]. In addition, major urban markets have more diversified food supply sources (Table  3), so food avail- ability is less of a challenge in urban markets. The differ- ent food markets and the diversified food supply sources curtail the effect of weather seasonality on HDDS. How- ever, weather seasonality can affect household food consumption through the prices of food commodities. Table 9 Effect of seasonality on HDDS using Poisson fixed effects Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 * Reference base for educational level is “No formal education” +Real price of maize and tomatoes are computed based on ESOKO-Ghana December and June price averages from 2013 to 2020 Variables HDDS Season (Dry) -0.025 (0.120) Characteristics of household Age of household head 0.005 (0.009) Sex of household head (Male) 0.237 (0.146) Household size 0.033** (0.017) Education of household head* Primary 0.163 (0.154) Secondary 0.088 (0.178) Tertiary -0.318 (0.363) Household wealth index 0.035*** (0.009) Household food safety knowledge 0.004** (0.001) Marital status of household head Single -0.929*** (0.277) Monogamous -0.459*** (0.098) Polygamous -0.676* (0.372) Price of maize+ 0.136 (0.641) Price of tomatoesa 0.016 (0.023) Employment status Employment status of household head -0.010 (0.043) Percent of household members employed 0.001* (0.001) Self-reported covid-19 effect Affected price of staple foods -0.031 (0.030) Affected price of vegetables 0.027 (0.034) Constant Observations 1,212 R-squared Number of unique respondents 606 Page 18 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Kaminski et  al. [76] showed that seasonal food price changes are inversely correlated with household food consumption. From our analysis, the price of maize and tomatoes has a minimal or no effect on HDDS when con- trolling for household fixed effects. This may be because HDDS counts the number of food groups a household consumes and does not measure the quantity and quality of food consumed. Thus, the quantity of food consumed by households from different food groups may change between seasons, but the dietary diversity score recorded may remain the same. However, wealthier households eat more diversified foods and spend more on food per capita, a potential indication of healthier food choices. Wealthier households can spend more on food per capita and consume more diversified foods within food groups and more food groups [46, 100]. Also, household size is positively associated with HDDS. The finding is corroborated by Thorne-Lyman et al. [45], who found that household size positively cor- relates to dietary diversity and negatively correlates to per capita monthly total food expenditure. Household size can positively affect dietary diversity because, with many household members and their varied ages, the household is likely to have higher income and consume food with high dietary diversity to meet members’ nutri- tional needs [101, 102]. Conversely, household size can negatively affect household dietary diversity. Due to large family sizes, poor households may not be able to spend more on adequate nutritious food and thus reduce their diet quality and diversity to meet the hunger needs of all household members [103, 104]. Limitations of study Even with our findings, the study may be constrained by the definition of some variables. An index to measure food safety based on more indicators is more desirable than the use of the incidence of diarrhoea as a proxy for food safety. Similarly, measuring seasonality based on environmental and socioeconomic indicators may be more desirable than a dummy variable. Additionally, a longer panel would provide more robust evidence of the effect of seasonality in urban areas. Also, the effect of Covid-19, which occurred before the second round of data collection, may influence our outcome variables like dietary diversity and households’ attitudes towards food safety and health. Further studies are required to decouple the effects of seasonality and Covid-19 on the outcome variables. Conclusions A sustainable urban food system is one of the most cru- cial priorities of a growing Ghanaian urban population because of its impact on food security and public health. This study set out to address the issue of food safety and HDDS in urban areas because they are important deter- minants of food security and public health. We answered whether household food safety and dietary diversity changed across seasons. The study also provided evidence of the presence of selected foodborne pathogens in selected purchased food commodities from an urban market. Our results emphasise that in developing countries like Ghana, food safety is a challenge in major city food mar- kets and is associated with weather seasonality. How- ever, weather seasonality in cities does not significantly affect urban households’ HDDS. Households’ average HDDS is 7.3 and, therefore, cannot be considered food insecure. However, they purchase foods from markets where selected samples tested positive for high levels of aflatoxin B1 and other harmful foodborne microorgan- isms. Therefore, although there is food availability, the safety of food from traditional open-air markets is not assured. Therefore, urban food security can be improved by enhancing, in addition to food availability and acces- sibility, food safety along the food supply chain. Ghana’s current food system will require a trans- formation to become sustainable. The domestic mar- ket food supply chain lacks adequate monitoring and enforcement of food safety standards. The National Food Safety Policy (NFSP) identified the high infor- mality of the sector in Ghana as accounting for many activities not sufficiently regulated and the difficulty with product traceability [105]. Therefore, in the short term, Ghanaian local authorities such as sanitation and public health officers at the various markets should enforce laws (e.g. Public Health Act, 2012 (Act 851)) and regulations on food safety. The food safety surveil- lance and early warning system in the NFSP should be able to prevent the outbreak of diseases and detect and seize unsafe foods before they get to the markets. In the long run, technology, innovation, and logistics investment are required to upgrade traditional open- air markets. Although food availability is not a chal- lenge for urban households, the focus should be on food safety and accessibility (affordability), especially for the urban poor. Abbreviations CRE Correlated Random Effect EA Enumeration Area FAO Food and agricultural Organisation FDA Food and Drugs Authority GSS Ghana Statistical Service HDDS Household Dietary Diversity Score MoH Ministry of Health NFSP National Food Safety Policy NMIMR Noguchi Memorial Institute for Medical Research PPMLHDFE Poisson pseudo-maximum likelihood estimator with multiple levels of fixed effects WASH Water, Sanitation and Hygiene WHO World Health organisation Page 19 of 21Dzudzor et al. BMC Public Health (2024) 24:888 Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12889- 024- 18297-0. Additional file 1: Supplementary Table 1. Robustness of the results of the effect of seasonality on households’ incidence of diarrhoea/vomiting using alternative estimation methods. Supplementary Table 2. Robust- ness of the results of the effect of seasonality on HDDS using alternative estimation methods.Additional tables of the research. Acknowledgements We want to thank our partners at the Institute of Statistical, Social and Economic Research (ISSER) for their immense support during the implementa- tion of the NOURICITY project in Ghana. We also thank all the data collectors for assisting in the data-gathering process. We would also like to thank the participants for agreeing to be part of this study. Authors’ contributions N.G and F.A were the principal investigators of the NOURICITY project. N.G, F.A and M.D planned and conducted the study. M.D supervised the data col- lection and analyzed the data. M.D wrote the initial manuscript. N.G, F.A and M.D reviewed the final manuscript. All authors read and approved the final manuscript. Funding Open Access funding enabled and organized by Projekt DEAL. The European Union’s Horizon 2020 research and innovation program under the grant agreement No. 727715 provided funding for the NOURICITY project. Open Access funding enabled and organized by Projekt DEAL. Availability of data and materials The data that support the findings of this study are available from the NOU- RICITY project, but restrictions apply to the availability of these data, so they are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the NOURICITY project. Declarations Ethics approval and consent to participate The NOURICITY project obtained ethical clearance from the Ethics Committee for Humanities, University of Ghana and the Ethics Committee of the Centre for Development Research (ZEF), University of Bonn. Informed consent was obtained from all individual participants included in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Author details 1 Centre for Development Research (ZEF), University of Bonn, Genscherallee 3, Bonn D-53113, Germany. 2 Institute for Food and Resource Economics (ILR), University of Bonn, Nussallee 19, Bonn D-53115, Germany. 3 Institute of Statisti- cal, Social and Economic Research (ISSER), University of Ghana, P.O. Box LG 74, Legon, Accra, Ghana. Received: 11 January 2024 Accepted: 6 March 2024 References 1. von Braun J, Afsana K, Fresco LO, Hassan MH. Food systems: seven priorities to end hunger and protect the planet. In: Science and innovations for food systems transformation. Cham: Springer Interna- tional Publishing; 2023. p. 3–9. 2. Afshin A, Sur PJ, Fay KA, Cornaby L, Ferrara G, Salama JS, Mullany EC, Abate KH, Abbafati C, Abebe Z, Afarideh M. Health effects of dietary risks in 195 countries, 1990–2017: a systematic analysis for the global burden of disease study 2017. Lancet. 2019;393(10184):1958–72. 3. World Health Organization. WHO estimates of the global burden of foodborne diseases: foodborne disease burden epidemiology reference group 2007-2015. Geneva: World Health Organization; 2015. Available from: https:// apps. who. int/ iris/ handle/ 10665/ 199350. 4. Gizaw Z. Public health risks related to food safety issues in the food market: a systematic literature review. Environ Health Prev Med. 2019;24:1–21. 5. HLPE. Nutrition and food systems. A report by the High Level Panel of Experts on Food Security and Nutrition of the Committee on World Food Security. Rome: FAO; 2017. Available from: https:// www. fao. org/3/ i7846e/ i7846e. pdf. 6. World Health Organization (WHO). WHO global strategy for food safety 2022-2030: towards stronger food safety systems and global coopera- tion. Geneva: World Health Organization; 2022. Available from: https:// www. who. int/ publi catio ns/i/ item/ 97892 40057 685. 7. Codex Alimentarius. General Principles of Food Hygiene, CXC 1-1969. Rome: Food and Agriculture Organization of the United Nations (FAO)/ World Health Organization (WHO) Codex Alimentarius. 2020, https:// www. fao. org/ fao- who- codex alime ntari us/ sh- proxy/ en/? lnk= 1& url= https% 253A% 252F% 252Fw orksp ace. fao. org% 252Fs ites% 252Fc odex% 252FS tanda rds% 252FC XC% 2B1- 1969% 252FC XC_ 001e. pdf 8. World Health Organisation (WHO). Food safety. Fact sheet No. 399. Geneva: WHO media centre; 2022. https:// www. who. int/ news- room/ fact- sheets/ detail/ food- safety. 9. Food and Agriculture Organisation (FAO). The State of Food Insecurity in the World 2001. Rome. 2001, ISBN 92-5-104628-X, https:// www. fao. org/3/ y1500e/ y1500e. pdf. 10. Food and Agriculture Organisation (FAO). The future of food safety. There is no food security without food safety. Rome, FAO. 2019 CA4289EN/1/04.19. https:// www. fao. org/3/ ca428 9en/ CA428 9EN. pdf. 11. HLPE. Food security and nutrition: building a global narrative towards 2030. Report by the High Level Panel of Experts on Food Security and Nutrition of the Committee on World Food Security. Rome: FAO; 2020. Available from: https:// www. fao. org/3/ ca973 1en/ ca973 1en. pdf. 12. Charalambous M, Fryer PJ, Panayides S, Smith M. Implementation of food safety management systems in small food businesses in Cyprus. Food Control. 2015;57:70–5. 13. Henson S, Heasman M. Food safety regulation and the firm: under- standing the compliance process. Food Policy. 1998;23(1):9–23. 14. Hoffmann V, Moser C, Saak A. Food safety in low and middle-income countries: the evidence through an economic lens. World Dev. 2019;123:104611. 15. Priefer C, Jörissen J, Bräutigam KR. Food waste prevention in Europe–A cause-driven approach to identify the most relevant leverage points for action. Resour Conserv Recycl. 2016;109:155–65. 16. HLPE. Food losses and waste in the context of sustainable food sys- tems. A report by the High Level Panel of Experts on Food Security and Nutrition of the Committee on World Food Security. Rome: FAO; 2014. Available from: https:// www. fao. org/3/ i3901e/ i3901e. pdf. 17. Yu H, Neal JA, Sirsat SA. Consumers’ food safety risk perceptions and willingness to pay for fresh-cut produce with lower risk of foodborne illness. Food Control. 2018;86:83–9. 18. Mensah LD, Julien D. Implementation of food safety management systems in the UK. Food Control. 2011;22(8):1216–25. 19. Uyttendaele M, Franz E, Schlüter O. Food safety, a global challenge. Int J Environ Res Public Health. 2016;13(1):67. 20. Sibhatu KT, Qaim M. Rural food security, subsistence agriculture, and seasonality. PLoS One. 2017;12(10):e0186406. 21. Hirvonen K, Taffesse AS, Hassen IW. Seasonality and household diets in Ethiopia. Public Health Nutr. 2016;19(10):1723–30. 22. Matz JA, Kalkuhl M, Abegaz GA. The short-term impact of price shocks on food security-evidence from urban and rural Ethiopia. Food Secur. 2015;7:657–79. 23. Becquey E, Delpeuch F, Konaté AM, Delsol H, Lange M, Zoungrana M, Martin-Prevel Y. Seasonality of the dietary dimension of household food security in urban Burkina Faso. Br J Nutr. 2012;107(12):1860–70. 24. Gilbert CL, Christiaensen L, Kaminski J. Food price seasonality in Africa: measurement and extent. Food Policy. 2017;67:119–32. https://doi.org/10.1186/s12889-024-18297-0 https://doi.org/10.1186/s12889-024-18297-0 https://apps.who.int/iris/handle/10665/199350 https://www.fao.org/3/i7846e/i7846e.pdf https://www.fao.org/3/i7846e/i7846e.pdf https://www.who.int/publications/i/item/9789240057685 https://www.who.int/publications/i/item/9789240057685 https://www.fao.org/fao-who-codexalimentarius/sh-proxy/en/?lnk=1&url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXC%2B1-1969%252FCXC_001e.pdf https://www.fao.org/fao-who-codexalimentarius/sh-proxy/en/?lnk=1&url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXC%2B1-1969%252FCXC_001e.pdf https://www.fao.org/fao-who-codexalimentarius/sh-proxy/en/?lnk=1&url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXC%2B1-1969%252FCXC_001e.pdf https://www.fao.org/fao-who-codexalimentarius/sh-proxy/en/?lnk=1&url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXC%2B1-1969%252FCXC_001e.pdf https://www.who.int/news-room/fact-sheets/detail/food-safety https://www.who.int/news-room/fact-sheets/detail/food-safety https://www.fao.org/3/y1500e/y1500e.pdf https://www.fao.org/3/y1500e/y1500e.pdf https://www.fao.org/3/ca4289en/CA4289EN.pdf https://www.fao.org/3/ca9731en/ca9731en.pdf https://www.fao.org/3/i3901e/i3901e.pdf Page 20 of 21Dzudzor et al. BMC Public Health (2024) 24:888 25. Hillbruner C, Egan R. Seasonality, household food security, and nutri- tional status in Dinajpur. Bangladesh Food Nutr Bull. 2008;29(3):221–31. 26. FAO, WFP, & IFAD. The state of food insecurity in the world: how does international price volatility affect domestic economies and food secu- rity? Rome: Food and Agriculture Organisation of the United Nations; 2011 p. 9. https:// www. fao. org/3/ i2330e/ i2330e. pdf. 27. Rosegrant MW, Cai X, Cline SA. World water and food to 2025: dealing with scarcity. Washington, DC: International Food Policy Research Insti- tute; 2002. p. 1–322. Available from: https:// www. ifpri. org/ publi cation/ world- water- and- food- 2025. 28. Cooper PJ, Dimes J, Rao KP, Shapiro B, Shiferaw B, Twomlow S. Coping better with current climatic variability in the rain-fed farming systems of sub-saharan Africa: an essential first step in adapting to future climate change? Agric Ecosyst Environ. 2008;126(1–2):24–35. 29. Simpson RB, Zhou B, Naumova EN. Seasonal synchronization of foodborne outbreaks in the United States, 1996–2017. Sci Rep. 2020;10(1):17500. 30. Lee D, Chang HH, Sarnat SE, Levy K. Precipitation and salmonellosis incidence in Georgia, USA: interactions between extreme rainfall events and antecedent rainfall conditions. Environ Health Perspect. 2019;127(9):097005. 31. Smith BA, Meadows S, Meyers R, Parmley EJ, Fazil A. Seasonality and zoonotic foodborne pathogens in Canada: relationships between climate and Campylobacter, E. Coli and Salmonella in meat products. Epidemiol Infect. 2019;147:147. 32. Dzudzor MI, Gerber N. Urban households’ food safety knowledge and behaviour: choice of food markets and cooking practices. J Agric Food Res. 2023;14:100728. 33. International Food Policy Research Institute (IFPRI). 2017 Global Food Policy Report. International Food Policy Research Institute. 2017. https:// doi. org/ 10. 2499/ 97808 96292 529. 34. World Health Organisation (WHO). A guide to healthy food markets. Geneva: World Health Organization. 2006. Retrieved 2023 from http:// apps. who. int/ iris/ bitst ream/ handle/ 10665/ 43393/ 92415 93938_ eng. pdf. 35. Kc KB, Legwegoh AF, Therien A, Fraser ED, Antwi-Agyei P. Food price, food security and dietary diversity: a comparative study of urban Cam- eroon and Ghana. J Int Dev. 2018;30(1):42–60. 36. Battersby J, Peyton S. The geography of supermarkets in Cape Town: Supermarket expansion and food access. Urban Forum. 2014;25:153–64 Springer Netherlands. 37. Birhane T, Shiferaw S, Hagos S, Mohindra KS. Urban food insecurity in the context of high food prices: a community based cross sectional study in Addis Ababa, Ethiopia. BMC Public Health. 2014;14(1):1–8. 38. Reardon T, Tschirley D, Liverpool-Tasie LS, Awokuse T, Fanzo J, Minten B, Vos R, Dolislager M, Sauer C, Dhar R, Vargas C. The processed food revo- lution in African food systems and the double burden of malnutrition. Global food Secur. 2021;28:100466. 39. Béné C, Prager SD, Achicanoy HA, Toro PA, Lamotte L, Cedrez CB, Mapes BR. Understanding food systems drivers: a critical review of the litera- ture. Global Food Secur. 2019;23:149–59. 40. Qaim M. Globalisation of agrifood systems and sustainable nutrition. Proc Nutr Soc. 2017;76(1):12–21. 41. Otterbach S, Oskorouchi HR, Rogan M, Qaim M. Using Google data to measure the role of big food and fast food in South Africa’s obesity epidemic. World Dev. 2021;140:105368. 42. Debela BL, Demmler KM, Klasen S, Qaim M. Supermarket food purchases and child nutrition in Kenya. Global Food Secur. 2020;25:100341. 43. Rupa JA, Umberger WJ, Zeng D. Does food market modernisation lead to improved dietary diversity and diet quality for urban Vietnamese house- holds? Australian J Agricultural Resource Econ. 2019;63(3):499–520. 44. Reardon T, Henson S, Gulati A. Links between supermarkets and food prices, diet diversity and food safety in developing countries. In: Hawkes C, Blouin C, Henson S, Drager N, Dubé L, editors. Trade, Food, Diet and Health: Perspectives and Policy Options. Hoboken: Wiley- Blackwell; 2010. p. 111–30. 45. Thorne-Lyman AL, Valpiani N, Sun K, Semba RD, Klotz CL, Kraemer K, Akhter N, De Pee S, Moench-Pfanner R, Sari M, Bloem MW. Household dietary diversity and food expenditures are closely linked in rural Bang- ladesh, increasing the risk of malnutrition due to the financial crisis. J Nutr. 2010;140(1):S182–188. 46. Swindale A, Bilinsky P. Household dietary diversity score (HDDS) for measurement of household food access: indicator guide. Washington, DC: Food and Nutrition Technical Assistance Project, Academy for Educational Development; 2006. 47. Brander M, Bernauer T, Huss M. Improved on-farm storage reduces sea- sonal food insecurity of smallholder farmer households–evidence from a randomized control trial in Tanzania. Food Policy. 2021;98:101891. 48. Gelli A, Aberman NL, Margolies A, Santacroce M, Baulch B, Chirwa E. Lean-season food transfers affect children’s diets and household food security: evidence from a quasi-experiment in Malawi. J Nutr. 2017;147(5):869–78. 49. Ghana Statistical Service (GSS). Ghana. Population and housing census. General report, highlights. Accra: Ghana Statistical Service; 2021. p. 2022. 50. Ghana Statistical Service (GSS). Rebased 2013-2019AnnualGross domestic product. April 2020 edition. Accra: Ghana Statistical Service; 2020. 51. Ghana Statistical Service (GSS). 2017/18 Ghana Census of Agriculture (GCA) National report. Accra: Ghana Statistical Service; 2020. 52. Ghana Statistical Service (GSS). 2010 Population and housing census report. Urbanisation. Accra: Ghana Statistical Service; 2014. 53. Gerber N. NOURICITY- Partnership for Healthy Diets and Nutrition in Urban African Food Systems-Evidence and Strategies. CIHEAM, Bari I. 2022, October 8–10. Retrieved Jan 1, 2024 from https:// nouri city. org/ output/ poster/. 54. Government of Ghana [GoG]. National Policy for Aflatoxin Control in Food and Feed. 1st ed. Accra; 2022. p. 1–137. ISBN: 978-9988-9168-4-8. Available from: https:// libra ry. faraa frica. org/ 2022/ 10/ 12/ natio nal- policy- for- aflat oxin- contr ol- in- food- and- feed/. 55. Noguchi Memorial Institute for Medical Research (NMIMR-UG). Bac-047- 1.0 Standard Operating Procedure (SOP) for enumeration and detection of pathogens from food and animal products. 2023. 56. Vicam LP. Afla Test Instruction Manual., Watertown MA. 1999;2472. http:// www. biotic. com. tw/ index. files/ GN- MC9508- 5% 20(AflaT est). pdf. 57. Kirk MD, Pires SM, Black RE, Caipo M, Crump JA, Devleesschauwer B, Döpfer D, Fazil A, Fischer-Walker CL, Hald T, Hall AJ. World Health Organization estimates of the global and regional disease burden of 22 foodborne bacterial, protozoal, and viral diseases, 2010: a data synthe- sis. PLoS Med. 2015;12(12):e1001921. 58. World Health Organisation (WHO). Safer water, better health. 2019 update. Geneva: World Health Organization; 2019. 59. Larbi RT, Atiglo DY, Peterson MB, Biney AA, Dodoo ND, Dodoo FN. Household food sources and diarrhoea incidence in poor urban com- munities, Accra Ghana. PLoS ONE. 2021;16(1):e0245466. 60. Kirk MD, Angulo FJ, Havelaar AH, Black RE. Diarrhoeal disease in children due to contaminated food. Bull World Health Organ. 2017;95(3):233. 61. World Health Organisation (WHO). Five keys to safer food manual. WHO 2006. https:// www. who. int/ foods afety/ publi catio ns/ consu mer/ manual_ keys. pdf. Accessed 10 Oct 2019. 62. Islam AH, von Braun J, Thorne-Lyman AL, Ahmed AU. Farm diversifica- tion and food and nutrition security in Bangladesh: empirical evidence from nationally representative household panel data. Food Secur. 2018;10:701–20. 63. Kouser S, Qaim M. Impact of Bt cotton on pesticide poisoning in small- holder agriculture: a panel data analysis. Ecol Econ. 2011;70(11):2105–13. 64. Silva JS, Tenreyro S. Further simulation evidence on the performance of the Poisson pseudo-maximum likelihood estimator. Econ Lett. 2011;112(2):220–2. https:// doi. org/ 10. 1016/j. econl et. 2011. 05. 008. 65. Cameron AC, Trivedi PK. Regression analysis of count data. Cambridge University Press; 2013. 66. Muthini D, Nzuma J, Nyikal R. Farm production diversity and its associa- tion with dietary diversity in Kenya. Food Secur. 2020;12:1107–20. 67. Wooldridge JM, Zhu Y. Inference in approximately sparse correlated random effects probit models with panel data. J Bus Economic Stat. 2020;38(1):1–8. 68. Greene WH. Econometric Analysis. Pearson Education. 2003. ISBN: 0-13-066189-9. 69. Cruz-Gonzalez M, Fernández-Val I, Weidner M. Bias corrections for probit and logit models with two-way fixed effects. Stata J. 2017;17(3):517–45. 70. Wooldridge JM. Introductory econometrics: A modern approach. 5th ed. Mason: South-Western, Cengage learning; 2013. https://www.fao.org/3/i2330e/i2330e.pdf https://www.ifpri.org/publication/world-water-and-food-2025 https://www.ifpri.org/publication/world-water-and-food-2025 https://doi.org/10.2499/9780896292529 http://apps.who.int/iris/bitstream/handle/10665/43393/9241593938_eng.pdf http://apps.who.int/iris/bitstream/handle/10665/43393/9241593938_eng.pdf https://nouricity.org/output/poster/ https://nouricity.org/output/poster/ https://library.faraafrica.org/2022/10/12/national-policy-for-aflatoxin-control-in-food-and-feed/ https://library.faraafrica.org/2022/10/12/national-policy-for-aflatoxin-control-in-food-and-feed/ http://www.biotic.com.tw/index.files/GN-MC9508-5%20(AflaTest).pdf https://www.who.int/foodsafety/publications/consumer/manual_keys.pdf https://www.who.int/foodsafety/publications/consumer/manual_keys.pdf https://doi.org/10.1016/j.econlet.2011.05.008 Page 21 of 21Dzudzor et al. BMC Public Health (2024) 24:888 71. Mundlak Y. On the pooling of time series and cross section data. Econo- metrica. 1978;47:69–85. 72. Correia S, Guimarães P, Zylkin T. Fast Poisson estimation with high- dimensional fixed effects. Stata J. 2020;20(1):95–115. 73. StataCorp. Stata 15 base reference Manual. College Station, TX: Stata; 2017. 74. Abokyi E, Folmer H, Asiedu KF. Public buffer stocks as agricultural out- put price stabilization policy in Ghana. Agric Food Secur. 2018;7(1):1–2. 75. Andam KS, Johnson ME, Ragasa C, Kufoalor DS, Das Gupta S. A chicken and maize situation: the poultry feed sector in Ghana. IFPRI Discussion Paper No. 1601. 2017; (1601):1–52. 76. Kaminski J, Christiaensen L, Gilbert CL. The end of seasonality? New insights from sub-Saharan Africa. Policy Research Working Paper. Wash- ington, DC: World Bank; 2014. p. 1–43. http:// hdl. handle. net/ 10986/ 18771. 77. ESOKO-Ghana. Weekly food prices in Ghana (2013-2020). [Unpublished dataset]. Accra Ghana. Received on 6th May, 2021. 78. World Bank. Ghana. Climate change overview. Country summary. Climate change knowledge portal. 2023. https:// clima tekno wledg eport al. world bank. org/ count ry/ ghana. Accessed 10 March 2023. 79. European Union (EU). Commission Regulation (EU) 2023/915 of 25 April 2023 on maximum levels for certain contaminants in food and repeal- ing regulation (EC) 1881/2006. Official J Eur Union. 2023;L 119:112–5. 80. Food US, Drugs Administration (FDA-US). Guidance for Indus- try: Action Levels for Poisonous or Deleterious Substances in Human Food and Animal Feed. AUGUST and. 2000. Docket Num- ber: FDA-2013-S-0610. https:// www. regul ations. gov/ docket? D= FDA- 2013-S- 0610. 81. Cai Y, McLaughlin M, Zhang K. Advancing the FDA/office of regula- tory affairs mycotoxin program: new analytical method approaches to addressing needs and challenges. J AOAC Int. 2020;103(3):705–9. 82. Farzana FD, Rahman AS, Sultana S, Raihan MJ, Haque MA, Waid JL, Choudhury N, Ahmed T. Coping strategies related to food insecurity at the household level in Bangladesh. PLoS One. 2017;12(4):e0171411. 83. Leung CW, Epel ES, Ritchie LD, Crawford PB, Laraia BA. Food insecurity is inversely associated with diet quality of lower-income adults. J Acad Nutr Dietetics. 2014;114(12):1943–53. 84. Avoka JA, Dun-Dery EJ, Seidu I, Abou AN, Twene P, Tandoh IO, Dun- Dery F. Time series analysis of the relationship between diarrhea in children and Rota 2 vaccine in the Fanteakwa District of the eastern region of Ghana. BMC Pediatr. 2021;21:1–9. https:// doi. org/ 10. 1186/ s12887- 021- 02540-3. 85. Tetteh J, Takramah WK, Ayanore MA, Ayanore AA, Bisung E, Alamu J. Trends for diarrhea morbidity in the Jasikan District of Ghana: estimates from district level diarrhea surveillance data, 2012–2016. J Trop Med. 2018;2018:4863607. 86. Enweronu-Laryea CC, Boamah I, Sifah E, Diamenu SK, Armah G. Decline in severe diarrhea hospitalizations after the introduction of rotavirus vaccination in Ghana: a prevalence study. BMC Infect Dis. 2014;14(1):1–6. 87. Usman MA, Gerber N, von Braun J. The impact of drinking water quality and sanitation on child health: evidence from rural Ethiopia. J Dev Stud. 2019;55(10):2193–211. 88. Tucker J, MacDonald A, Coulter L, Calow RC. Household water use, poverty and seasonality: wealth effects, labour constraints, and minimal consumption in Ethiopia. Water Resour Rural Dev. 2014;3:27–47. 89. Prüss-Ustün A, Bartram J, Clasen T, Colford JM Jr, Cumming O, Curtis V, Bonjour S, Dangour AD, De France J, Fewtrell L, Freeman MC. Burden of disease from inadequate water, sanitation and hygiene in low‐and mid- dle‐income settings: a retrospective analysis of data from 145 countries. Tropical Med Int Health. 2014;19(8):894–905. 90. Anyorikeya M, Ameme DK, Nyarko KM, Sackey SO, Afari E. Trends of diar- rhoeal diseases in children under five years in the War Memorial Hospital- Navrongo, Ghana: 2010–2013. Pan Afr Med J. 2016;25(Suppl 1):8. 91. Asamoah A, Ameme DK, Sackey SO, Nyarko KM, Afari EA. Diar- rhoea morbidity patterns in Central Region of Ghana. Pan Afr Med J. 2016;25(Suppl 1):17. 92. Kumi-Kyereme A, Amo-Adjei J. Household wealth, residential status and the incidence of diarrhoea among children under-five years in Ghana. J Epidemiol Global Health. 2016;6(3):131–40. 93. Sumampouw OJ, Nelwan JE, Rumayar AA. Socioeconomic factors associated with diarrhea among under-five children in Manado Coastal Area, Indonesia. J Global Infect Dis. 2019;11(4):140. 94. Alirol E, Getaz L, Stoll B, Chappuis F, Loutan L. Urbanisation and infectious diseases in a globalised world. Lancet Infect Dis. 2011;11(2):131–41. 95. Food and Drugs Authority (FDA). FDA conducts investigations into foodborne disease outbreak in Oyibi, Accra. FDA, Accra Ghana. 2023 http:// www. fdagh ana. gov. gh/ img/ press/ FDA% 20CON DUCTS% 20INV ESTIG ATIONS% 20INTO% 20FOOD% 20BOR NE% 20DIS EASE% 20OUT BREAK% 20IN% 20OYI BI- ACCRA. pdf. Accessed 10 Feb 2023. 96. Food and Drugs Authority (FDA). Public notice. Update: Marwako food poisoning incident. FDA, Accra Ghana. 2022. http:// www. fdagh ana. gov. gh/ img/ press/ MAR_ 11- 01. pdf. Accessed 10/02/2023. 97. FAO, IFAD, UNICEF, WFP and WHO. The state of food security and nutrition in the world 2020: transforming food systems for affordable healthy diets. Food & Agriculture Org; 2020. p. 1. https:// doi. org/ 10. 4060/ ca969 9en. 98. Rheinländer T, Olsen M, Bakang JA, Takyi H, Konradsen F, Samuelsen H. Keeping up appearances: perceptions of street food safety in urban Kumasi, Ghana. J Urb Health. 2008;85:952–64. 99. WHO. Countdown to 2023: WHO report on global trans-fat elimination 2022. Geneva: World Health Organization; 2023. 100. Jones AD, Shrinivas A, Bezner-Kerr R. Farm production diversity is associ- ated with greater household dietary diversity in Malawi: findings from nationally representative data. Food Policy. 2014;46:1–2. 101. Usman MA, Callo-Concha D. Does market access improve dietary diversity and food security? Evidence from Southwestern Ethiopian Smallholder coffee producers. Agricultural Food Econ. 2021;9:1–21. 102. Jateno W, Alemu BA, Shete M. Household dietary diversity across regions in Ethiopia: evidence from Ethiopian socio-economic survey data. PLoS One. 2023;18(4):e0283496. 103. Addai KN, Ng’ombe JN, Temoso O. Food poverty, vulnerability, and food consumption inequality among smallholder households in Ghana: a gender-based perspective. Soc Indic Res. 2022;163(2):661–89. 104. Powell B, Bezner Kerr R, Young SL, Johns T. The determinants of dietary diversity and nutrition: ethnonutrition knowledge of local people in the East Usambara Mountains, Tanzania. J Ethnobiol Ethnomed. 2017;13(1):1–2. 105. Ministry of Health [MoH]. National Food Safety Policy. Accra-Ghana. 2022. Retrieved Sept 27, 2023, from https:// www. moh. gov. gh/ wp- conte nt/ uploa ds/ 2022/ 07/ NFSP- Docum ent- signed- and- launc hed. pdf. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. http://hdl.handle.net/10986/18771 http://hdl.handle.net/10986/18771 https://climateknowledgeportal.worldbank.org/country/ghana https://climateknowledgeportal.worldbank.org/country/ghana https://www.regulations.gov/docket?D=FDA-2013-S-0610 https://www.regulations.gov/docket?D=FDA-2013-S-0610 https://doi.org/10.1186/s12887-021-02540-3 https://doi.org/10.1186/s12887-021-02540-3 http://www.fdaghana.gov.gh/img/press/FDA%20CONDUCTS%20INVESTIGATIONS%20INTO%20FOOD%20BORNE%20DISEASE%20OUTBREAK%20IN%20OYIBI-ACCRA.pdf http://www.fdaghana.gov.gh/img/press/FDA%20CONDUCTS%20INVESTIGATIONS%20INTO%20FOOD%20BORNE%20DISEASE%20OUTBREAK%20IN%20OYIBI-ACCRA.pdf http://www.fdaghana.gov.gh/img/press/FDA%20CONDUCTS%20INVESTIGATIONS%20INTO%20FOOD%20BORNE%20DISEASE%20OUTBREAK%20IN%20OYIBI-ACCRA.pdf http://www.fdaghana.gov.gh/img/press/MAR_11-01.pdf http://www.fdaghana.gov.gh/img/press/MAR_11-01.pdf https://doi.org/10.4060/ca9699en https://doi.org/10.4060/ca9699en https://www.moh.gov.gh/wp-content/uploads/2022/07/NFSP-Document-signed-and-launched.pdf https://www.moh.gov.gh/wp-content/uploads/2022/07/NFSP-Document-signed-and-launched.pdf Food safety and dietary diversity in African urban cities: evidence from Ghana Abstract Background Method Results Conclusions Background Methods Study area Survey data and questionn