1 Introduction

Global food insecurity is rapidly rising, impacting millions of people worldwide due to escalating food prices and unstable political conditions [1, 2]. As a significant portion of the global population continues to migrate to urban areas, the extent of food insecurity has increasingly shifted toward cities. As Djan [3] noted that ensuring food security in urban regions is influenced by factors such as rising food prices, limited access to land and water for agriculture, inadequate infrastructure, poor food supply systems, and economic inequality among nations. According to FAO [4], the number of people affected by hunger in 2023 reached 298.4 million in Africa, 384.5 million in Asia, 41.0 million in Latin America and the Caribbean, and 3.3 million in Oceania. While Asia is home to the majority of the world’s undernourished population, Africa holds the highest prevalence of food insecurity. The FAO's report in 2024 indicated that approximately 28.9% of the global population—equivalent to 2.33 billion people were moderately or severely food insecure, lacking regular access to adequate food. Despite food being a fundamental human right, achieving food security remains a significant challenge both now and for the future [5].

In Africa, over 20% of the total population experiences hunger [1, 6]. This issue is primarily driven by challenges in the global economic environment, conflicts and wars, and adverse climatic conditions [7]. According to Sileshi et al. [8], food insecurity in urban areas, particularly in East Africa where Ethiopia is located, is exacerbated by rapid urbanization and the highest rates of under nutrition compared to other regions in Africa. By 2050, it is projected that two-thirds of the world’s population will reside in cities, with 90% of this growth occurring in Africa and Asia [9]. The same authors highlight that urbanization in Africa, especially in Sub-Saharan Africa, is happening at an unprecedented pace, resulting in elevated levels of urban food insecurity across the continent. According to Berlie [10], in Sub-Saharan Africa, urban food insecurity is emerging as one of the key development challenges. A study by Debela and Abebe [11] revealed that many large areas of Africa specifically Ethiopia, have largely considered female-headed households to be particularly vulnerable to food insecurity.

Ethiopia faces persistent and serious challenges related to chronic food insecurity [12] and ranks first among Africa's most food insecure and famine-affected nations [13]. In Ethiopia, more focus has been given to rural areas that perceive that the urban population is better than the rural poor in terms of food insecurity. This is because much of the research on food security in Ethiopia has forced on the rural poor, paying very little attention to the urban poor [10]. However, the urban poor, marginalized from employment opportunities and other services are chronically food insecure in many measurements of food security indicators. Similarly, Gezimu [14] demonstrated that similar to many developing nations, research on food security in Ethiopia has historically concentrated on rural areas, giving the urban poor little attention.

In many urban areas of Ethiopia, female-headed households are highly food insecure compared to their male. In addition, Akadiri et al. [15] noted that various studies and policymakers have focused on the question of whether households headed by females experience higher levels of food insecurity than households headed by men do. In other words, female-headed households are defined as those in which the female head makes all major decisions for the family and oversees the household’s finances [16]. The same authors also noted that in many parts of Africa, particularly in SSA, where Ethiopia is located, the number of female-headed households has increased significantly mainly because of the presence of war/conflict and other catastrophes for which males are more vulnerable than female-headed households are. However, one can agree that the rates of poverty and food insecurity are high among female-headed households because of a lack of income opportunities, dual responsibilities at home and outside the home, and scarce labor power [10, 17, 18] noted that female-headed households are more disadvantageous than male headed households in many development indicators.

Although female-headed households face two burdens inside and outside the home, they employ different coping strategies to cope from food insecurity and other shocks such as price volatility, income instability, job loss, gender based violence, and health crises. According to Biadgilign [19] and Woldemichel [20], the most popular coping mechanisms employed by female-headed households include engaging in daily labor work, migrating to large urban areas, relying on remittances from relatives, limiting family size, reducing food consumption, begging, borrowing food, depending on less favored meals, borrowing money to buy food, and taking credit to purchase food.

Extremely high food costs have become a significant worldwide issue, especially for poor urban female-headed households in low-income countries such as Ethiopia. However, there is little empirical evidence on urban food insecurity and coping strategies with special emphasis on female-headed households [21]. These findings indicate food insecurity studies have focused on rural areas, paying very little attention poor and vulnerable urban households such as female-headed households [10, 22, 23]. The present study attempted to fill this gap. The problems faced by female-headed households are multidimensional and include poverty, food insecurity, and low-level coping strategies [11, 24,25,26,27]. Nonetheless, no study has been conducted on the severity of female-headed households’ food insecurity using multidimensional food security index which could be addressed in this study.

The general objective of the study was to identify the determinants of food security among female-headed households in Bahir Dar city, Amhara Region. The specific objectives are as follows:

  1. (i)

    To assess the responses of female-headed households to food insecurity

  2. (ii)

    To examine the severity of food insecurity experienced by female-headed households for policy implications

  3. (iii)

    To identify the factors influencing the coping strategies of female-headed households and their food insecurity access scale.

1.1 Feministic theory in the study of food security

At least as far back as the eighteenth century, feminist thought may be traced to the writings of pioneering liberal feminists such as Mary and John (Sarikkis, Rush & Lane, 2008) cited in Berlie [28]. Defining what feminism means in a particular setting is crucial before discussing feminist frameworks [29].

Feminists generally agree that there are various forms of feminism. It is frequently challenging to make the case for a feminist paradigm in practice because of this diversity. Regardless of these limitations, Phillips [30] defined feminism as “where women and men are free to live creative lives, in security and with bodily health and integrity, where they are free to choose whom they love, and with whom they set up house and whether they want to have or not have children”. In this context, feminism is a word that has tremendous historical baggage [31]. Part of the focus of feminist political ecology studies is how gender and the environment’s meanings of reality are coproduced, changed, and mutually formed across time [32]. The dynamic and context specific power dynamics between men and women [33].

Liberal feminist strategies have been closely examined by critical feminists and development patterns of entirely specified conceptualizations of women’s identities and gendered power relations since they are implemented in conflicting ways both literally and discursively [33]. Interventions aimed at empowering women offer a significant chance to enhance nutrition-related results. However, there is a lack of cross-contextual research on the variables that lead to women’s and girls’ worse nutrition outcomes as well as how women’s empowerment can enhance those outcomes [34]. This could mean that gender is still a highly debated issue, mostly because it is defined by those who study it [35]. While feminism as a movement succeeded much in underlining systemic socioeconomic disparities, the movement also resulted in marginalizing groups of women who did not fit into the more neo-liberal Western understanding of feminism. In addition to the obstacles women encounter in obtaining resources because of their gender, there are societal standards that impede women’s full involvement in the food production process [35].

According to FAO [36], two primary pathways influence women and girls. Their ability to pursue higher education is restricted, and they have fewer job options, which reduce their financial independence and lessen their family’s ability to negotiate. Their weaker negotiating position results in reduced health and nutrition outcomes, unequal feeding and caregiving practices that favor boys and men, little say in household decisions, and food and nutrition instability. In general, feministic theory searches into the interplay between gender and power shaping cultural, societal, and political dynamics and the right to obtain adequate food. It underscores the importance of comprehending and questioning oppressive systems.

1.2 Conceptual framework of the study

Considerable numbers of variables affect female-headed households' food security. As shown in Fig. 1, demographic and socioeconomic characteristics, institutional factors, coping capacity, kilocalorie availability, and the types of food they consume can affect female-headed households’ food security. Asset ownership, poor market access, shortage of labor, and a lack of job opportunities are socioeconomic variables affecting female-headed households’ food security. In relation to this, Berlie [10] suggested that one of the main causes of shock for low–income households living in large cities could be their ability to obtain asset ownership. Social capital such as iddir, equip, mahber, and other associations is a vital factor that directly or indirectly affects female-headed households’ food security. Safety nets, credit availability, and employment opportunities are institutional variables that positively and/or negatively affect female-headed households.

Fig. 1
figure 1

Source: Modified from Berlie [10]

Schematic diagram illustrating the relationship between female-headed households' food security and predictor variables.

Full size image

Some scholars such as Akadiri et al. [15] suggested that educational attainment, higher levels of income, market access, and asset ownership of households have a significant influence on household food security. Likewise, Teferi [37] and Hardiani et al. [38] also noted that household size, household income, age of respondents, educational status, savings, and food insecurity in households were found to be influenced by the source of income from gifts and remittances.

2 Research methodologies

2.1 Description of the study area

The Amhara National Regional State (ANRS), which is situated in northwest Ethiopia, has Bahir Dar as its capital. Situated close to Lake Tana, the source of the Blue Nile, it is a popular tourist attraction. The flat plateau earth construction of Bahir Dar is situated 11° 36″ North and 37° 23″ East (Fig. 2). Six subcities, three satellite cities, make up the city government. There are 40 kebele administrations throughout the 26 urban kebeles and 14 rural kebeles [39].

Fig. 2
figure 2

The study area’s location map

Full size image

The name of the city of Bahir Dar was named because of its closeness to two bodies of water, the River Abay (Nile) and Lake Tana. Thus, Bahir Dar refers to a city that is on or very near the Blue Nile and Lake Tana shores. It is currently one of the largest and fastest-growing cities in the nation [40].

2.1.1 Socioeconomic and demographic features of the city

The population of Bahir Dar city is estimated to be 389,177, of which 183,984 are males and the rest are females [39]. Owing to its natural and cultural characteristics, Bahir Dar is currently one of Ethiopia's top tourist destination cities. The State of Amhara National Region (ANRS) has many tourist attraction centers. Among these, the Simien Mountains National Park, the Castle of Gondar, and the Rock-Hewn Churches of Lalibela are some to mention. This is a fantastic chance for the city to attract both domestic and foreign visitors. In addition, Bahir Dar is rich in natural and cultural heritage assets such as Lake Tana monasteries, the Blue Nile Falls, and Bezawit Palace. More importantly, because there are many tourist attractions in Bahir Dar city, the city’s tourism industry is one among the fastest-growing in the nation. The city offers excellent possibilities for economic activities such as industry, investment, agriculture, and fish development including an expanding textile sector as well as quickly growing agriculture industries [39].

2.2 Research methods

2.2.1 Research design

The study used a mixed methods research design, which combines quantitative and qualitative research approaches. In food security studies, the use of qualitative and quantitative approaches is gaining recognition among researchers for its need to offer a more thorough analysis, as both methods are inadequate on their own to fully convey the patterns and specificities of a situation [41]. The respondents’ socioeconomic and demographic characteristics were analyzed via quantitative research methods whereas the participants’ opinions and suggestions related to the causes of food insecurity and their coping capacity were analyzed via qualitative research techniques.

2.2.2 Techniques for sampling and determining sample size

The sample households for the study were selected via a four-stage sampling procedure. First, the capital city of the Amhara National Regional State, Bahir Dar, was chosen purposively. This is because the principal investigator worked for a long time which helped him study the multifaceted problems of female-headed households that demand an immediate solution. Second, the three sample sub-cities, namely the Fasilo, Gish-Abay, and Tanna sub-cities were selected randomly among the six sub-cities of Bahir Dar city (see Fig. 1). The three sub-cities selected have urban kebele administrations, which could help to identify urban female-headed households' food security problems and their coping strategies in-depth. Third, the 02 and 03 kebele administrations from the Fasilo sub-city, Abinet and Hidasse kebeles from Gish Abay, and the Shimbet and Ras Agez kebele administrations from the Tana sub-city were selected via a simple random sampling technique. Fourth, with the objective of determining the sample size for completing the questionnaire, the Israel [42] sample size formula was employed as follows:

$$\text{N }=\frac{N}{{1+N(e}^{2)}}$$

where:

n = sample size; N = total number of households in the selected kebeles; e2 = maximum variability or margin of error of 5%.

$$\text{n }=\frac{1205}{\boldsymbol{ }{1+1205(0.05}^{2)}}$$
$$\text{n}=\frac{1205}{\boldsymbol{ }1+1205(0.0025)}$$
$$\text{n}=\frac{1205}{1+3.0125}= 300$$

Using this formula, 300 female-headed households were selected to complete the questionnaire. However, owing to concerns over missing data, 330 was found to be the right sample size (a 10% increase) to account for the possibility of missing data and the nonresponse rate. In relation to this, Naing et al. [43] indicated that if there is a non-response rate, it is advisable to oversample by 10% to 20%. Finally, 330 female-headed households were selected via systematic sampling techniques and sampling frames obtained from each kebele administration office.

2.2.3 Qualitative sampling

Through purposive sampling techniques, participants for the qualitative sampling were selected for the study. Accordingly, from the six sample kebeles six female-headed households, from the City Council, Female Affairs Department, and Disaster Risk Management and Food Security Coordinators Office, three participants were selected as key informants for the study. Among the nine key informants, three of them are working in government offices.

Likewise, sample participants for FGDs were selected by using a purposive sampling technique. Accordingly, Kebele Women's Affairs Head from each sample kebele; one kebele level administrator from each kebele; a female-headed household from each sample Kebele; and one vice speaker from Bahir Dar City Council. Nineteen participants were selected purposely. Hence, there were three FGD groups in the sample kebeles comprising 6 members, and one member from the City Council was selected purposively for the study. Finally, every data point was collected with informed consent obtained from all the participants involved in the study. Ethical approval was also obtained from Bahir Dar University, Faculty of Social Science, in the Department of Geography and Environmental Studies. Thus, the protocol was approved by the commettee established at the department level in accordance with the rules and regulations of Bahir Dar University.

2.2.4 Data collection techniques

Survey questionnaire The questionnaire survey comprised questions that were both closed- and open-ended, covering a range of topics: Respondents’ socioeconomic and demographic traits, asset ownership, safety nests, coping strategies, household food insecurity scale, dietary energy supply, and household dietary diversity score. Twelve randomly selected female-headed households completed the questionnaire for pretesting. Ten enumerators and one supervisor were selected to interview the respondents. The enumerators were first trained by the principal investigator on ways to frame and clarify each question to the respondents and advise them on the purpose of the study and ethical issues. The researcher, one supervisor, and ten enumerators conducted the survey all speaking the local language.

Key informant Interviews The key informant interviews, which were semi-structured focused on the problems associated with dietary diversity, asset ownership, food prices, and coping strategies. With the consent of the participants, smartphones were used to obtain time to listen and make eye contact. As an ethical issue, those who were not volunteers to participate in the interviews were excluded from the data collection.

Focus group discussions As discussed above, three focus group discussions were conducted for this study to obtain detailed information about household asset ownership, female-headed households’ coping strategies, causes of food insecurity, and the distribution of safety nets. The principal investigator moderated the discussion. Likewise, with the consent of the participants, a smartphone was utilized to obtain time to listen and make eye contact.

2.2.5 Methods for analyzing the data

Qualitative and quantitative data analysis methods were used for the study. A qualitative analysis was conducted on the data obtained from key informant interviews and focus group discussions. The collected data were analyzed via transcriptions, textual presentations, and thematic analysis. In this context, three themes were created socioeconomic strategies, coping strategies, and causes of female-headed household food insecurity.

The quantitative data that were gathered were coded and added to SPSS version 20 for further analysis. Inferential and descriptive statistics were used to present and analyze the quantitative data for this study. Inferential statistics such as one-way ANOVA, Pearson correlation, chi-square tests, and linear regression modeling were employed to show relationships and differences between variables.

To identify determinant variables affecting a household's reduced coping strategy index (rCSI) and household food insecurity access scale (HFIAS), ordered logistic regression model employed. This is because the dependent variables; household food insecurity access scale and reduced coping strategy index were measured in ordered categorical variables. The independent variables as shown in the CFW were demographic and socio-economic. Model assumptions such as multicollinearity, independence of observation and outliers were checked. The goodness of fit of the models were also checked using the log likelihood, model summary and the likelihood chi-square test.

2.2.6 Measuring the indicators of food security

  1. (i)

    Reduced coping strategy index

Measuring food security is an expensive and difficult task. In countries with severe food insecurity such as Ethiopia, experts need regular measurements for interventions. However, time is limited, and field conditions do not permit lengthy and intensive data collection. Hence, easy to administer, straightforward, and rapid enough to provide real-time information tools are needed. The coping strategies index (CSI) is one such technique. A context-specific or full coping strategies index (CSI) and a reduced coping strategies index (rCSI) are two versions of the coping strategies index. However, experiences indicate that a subset of five questions from the full CSI adequately represents regular behaviors across different contexts. Hence, in this study, the rCSI was employed because of its advantages. The reduced coping strategy (rCSI) is based on a list of five food-related coping strategies that the household used in the seven days before the survey [44] (Table 1).

Table 1 The five reduced coping strategies based on their severity and frequency
Full size table

The rCSI is between 0 and 56. That is, the maximum rSCI is 56 if all 5 strategies (shown in Table 1) are applied every day (for 7 days). rCSI combines the frequency of each strategy (how many times each strategy was practiced for the last 7 days?) and their (severity) (how serious is each strategy?) for households reporting food consumption problems. This is related to the weights given for each coping strategy [44]. There are three groups according to the severity of standard weight: low coping (0–3), medium coping (4–9), and high coping (≥ 10). In general, the higher the rCSI is, the more severe the coping is applied by a household and the more food insecure it is.

  1. (ii)

    Household dietary diversity score (HDDS)

The household’s access to a variety of foods is reflected in the HDDS. As a result, this indicator may be used as a substitute for a person's diet’s nutrient adequacy. The HDDS comprises a simple count of the number of food groups a household consumed in the reporting period FAO [45] adopts a recall period of 24 h to limit the recall error as much as possible, which is the standard recall period. For the collection of information from each food group bivariate (yes = 1 and no = 0) items were employed. Research has indicated a positive relationship between household food security and increased diversity at the household level [46]. To compute the HDDS, twelve dietary groups were used. These products include (a) cereals, (b) roots/tubers (c) vegetables (d) fruits, (e) meat/poultry (f) eggs (g) fish/shellfish (h) pulses/legumes/nuts (i) dairy products (j) oil/fat (k) sugar/honey and (l) miscellaneous (coffee/tea/condiments) [47]. The HDDS is usually measured by summing the number of food groups consumed over a reference period as shown below.

$$\text{HDDS}={{\Sigma}}(\text{a}+\text{b}+\text{c}+\text{d}+\text{e}+\text{f}+\text{g}+\text{h}+\text{i}+\text{j}+\text{k}+\text{l})$$

FAO [48] asserts that those households that consume ≤ 3 food groups are classified into nutritional food inadequacy; 4 food groups are classified as medium/average; and ≥ 5 food groups are classified as nutritionally food adequacy (food secure).

  1. (iii)

    Household food insecurity access scale (HFIAS)

To guide, monitor, and assess program activities, several programs and organizations require indicators of the access component of household food insecurity that are easy to understand but methodologically acceptable. In relation to this, food insecurity's access component is measured by the HFIAS. The prevalence of household food insecurity can be determined via the data produced by the HFIAS. The HFIAS covers a recall period of 30 days and consists of two types of questions: nine occurrence questions and nine frequency-of-occurrence questions (Table 2).

Table 2 The nine occurrence questions for the calculation of HFIAS
Full size table

First, the respondent is questioned on an occurrence question in the past 4 weeks (yes or no). A frequency-of-occurrence question was asked the respondents if they indicated that they had occurred in the previous 30 days rarely (one or twice), sometimes (three to ten times), or often (more than ten times). The codes for each frequency of occurrence question are added to determine the HFIAS score variable for each household. The 4 food insecurity categories should be created sequentially. They are food secure (0–1), mildly food insecure (2–8), moderately food insecure (9–10), and severely food insecure (11–27) [2, 49, 50]. Note that the maximum is 27 and the minimum is 0. In general, a household with a higher score has greater food insecurity.

3 Results and discussion

3.1 Analysis of reduced coping strategy index of female-headed households

Age, educational level, marital status, and family size are demographic variables that can influence households' coping strategies. As shown in Table 3, individuals aged between 20 and 40 years had many medium and high level coping strategy indices. However, the highest age group exhibited a low reduced coping strategy index. These relationships were also significant at P < 0.01. Eskezia [51] reported that the age of female-headed households was one factor affecting the rCSI. The marital status of female-headed households revealed that divorced households have a relatively large number of medium and high-level coping strategy indices. The association between marital status and the reduced coping strategy index was statistically significant at P < 0.01. The study also noted that female-headed households who cannot read and write are exposed to medium and high- level coping strategies whereas a low-level coping strategy index was investigated for those who have secondary education or above. In relation to this, Argaw [23] indicated that the educational attainment of female household heads negatively related to the coping strategy index.

Table 3 The associations between the rCSI and demographic variables of female-headed households
Full size table

Table 3 shows that among the three sub-cites, high levels of coping were observed at Gish Abay accounting for 60%. A relatively low percentage of coping strategies was observed at Fasilo which was 52%. This result implies that certain sub-cities need immediate action to reduce food insecurity in female-headed households. Compared with kebeles houses and own houses, female-headed households who rent private houses exposed to high levels of coping. This showed that households that do not have shelter are highly exposed to severe food insecurity since house rent increased sharply in all the subcities of Bahir Dar. The results of this study were consistent with the findings of In general, 86.4% of female-headed households experience high coping. Similarly, Etana [52] found that 70% of female-headed households also reported high coping.

In relation to this, FGDs held in Gish-Abay suggested their opinions in the following ways:

Many of the female-headed households in our kebele have no formal education. This is due to many of them coming from rural and poor families in urban areas. This condition forced them to engage in nonproductive sectors and daily laborers. As a result, many of them live destitute that is dangerous for their health (FGD participant, November 2023).

As shown in Table 4, the total income and total expenditure of households were considered socioeconomic variables. From Table 4, the lowest mean income that female-headed households experienced was high coping which was a total mean income of 1801 ETB. On the other hand, households with low coping ability have a total mean of 3623.81 ETB. The one-way ANOVA results were statistically significant at P < 0.001 (F (2,327 df) = 51.5, P = 0.000). As shown in Table 4, total expenditure is a direct reflection of the total income of the households. Accordingly, low-level coping has a total expenditure of ETB 1923.381 high-level coping has a total expenditure of 1546.77. This variation was also statistically significant at P < 0.001 (F (2,327 df) = 18.306, P = 0.000). This result was consistent with the works of [53].

Table 4 The variations between socioeconomic variables and the rCSI
Full size table

In Bahir Dar city the number of female-headed households has increased over time because of migration from rural or urban areas to search for job opportunities. However, the condition is quite different from what they expected. In relation to this, a key informant from Fassilo sub-city mentioned her opinions in the following ways:

In my kebele administration, the number of female-headed households increased from time to time. This is due mainly to divorce, widowed, or separation. These situations forced me to engage in different activities to collect money to feed and educate my children. The current condition is too problematic to obtain money. Hence, passing the whole day without eating food was a common practice in my family. The problem is compounded because of the high house rent associated with high food prices (KIIs, November 2023).

As indicated, job opportunities in Bahir Dar city are a challenge because of the prevailing conditions and population pressure in the city. Sources of income are so scarce that many poor households particularly those headed by women are not able to lead their families. In relation to the sources of income and expenditures of the households, one key informant from the Shimbet kebele administration suggested the following:

Currently, I have three children; however, I do have a problem with money to lead my family. I am engaging in daily labor such as washing cloth and backing Injera with low payment. Nonetheless, the income that I received from different activities is not sufficient to feed my family. As a result, circumstances forced me to engage in prostitution. The money I obtained from different activities was entirely spent on food that I did not have any reserves and this is a sign of destitution (KIIs, November 2023).

3.1.1 Determinants of female-headed households’ rCSI

As outlined in the methodology section, ordered logistic regression was utilized because the dependent variable, the Reduced Coping Strategy Index (rCSI), is an ordered categorical variable coded as follows: (0) low coping, (1) medium coping, and (2) high coping. In this model, the reference category is low coping. Checks for multicollinearity were conducted, and no significant outliers were identified. Eleven variables were included in the model, with six demonstrating significance at p < 0.001 and p < 0.05 (Table 5). The likelihood ratio (LR) chi-square statistic (11) indicated that at least one predictor has a regression odds ratio that is not equal to zero, confirming the model's adequacy. The small p-value from the LR test, p < 0.001, suggests that at least one of the regression coefficients is statistically significant. Additionally, the pseudo R2 indicates that the predictor variables explain approximately 62.7% of the total variation in the model.

Table 5 Determinants of rCSI using ordered logistic regression
Full size table

The results from the ordered logistic regression indicated that Kebele Hidassie is significantly more likely to fall into the medium or high coping categories compared to Kebele 02, which serves as the reference category, with an odds ratio of 39.34536 and p < 0.05 (Table 5). This suggests that female-headed households in Hidassie are more inclined to be classified as medium or high coping rather than low coping. Similarly, Kebele Ras-Agez and Kebele Shimbet exhibited odds ratios of 59.11378 and 13.97352, respectively, both with p < 0.05. These findings imply that households in these two kebeles are also more likely to be medium or high coping rather than low coping. These study findings are consistent with those of Mekonen et al. [54].

Importantly, female-headed households in Ras-Agez and Shimbet classified as medium or high coping are indicative of food insecurity. When controlling for other variables, a one-unit increase in total income is associated with odds of being low coping versus medium or high coping that are 0.9981948 times lower. This suggests that higher total income correlates with a decreased likelihood of high coping, indicating that as income increases; female-headed households tend to stabilize and adopt lower coping strategies, thereby enhancing food security. Additionally, the results reveal that households renting either government or private accommodations are more likely to engage in medium or high coping, with high odds ratios of 236.2194 and 126.801, respectively (Table 5). Currently, in Bahir Dar, rising house rents are exacerbating food insecurity among female-headed households, as high coping strategies serve as a proxy indicator of this insecurity.

3.2 Analysis of the household food insecurity access scale (HFIAS)

As shown in Table 6, for the question “In the past 4 weeks, did you or any household member have to eat a limited variety of foods owing to lack of food in the house”. For this question, 75% of the respondents selected sometimes and often indicated the severity of household food insecurity.

Table 6 Household food insecurity access scale results
Full size table

For the first question, 65% of the respondents practiced sometimes and often. This showed that the majority of the households were worried that they did not have enough food. However, for the second question, the majority (54%) of the respondents indicated rarely for the last 30-day recall period. In this context, a study by Abraham and Abera [9] conducted in Southern Ethiopia, which utilized a multi-index-based assessment; found that 90.6% of female-headed households reported concerns about insufficient food. In all the questions listed in Table 6, female-headed households are exposed to severe food insecurity.

As shown in Fig. 3, HFIAS varies across sub cities in Bahir Dar city. Among the three subcities selected for the study, the severity of food insecurity was the highest (42%) at Gish Abay followed by Tana 34.5%) and the lowest was observed at Fasilo (23.2%). However, the Fasilo subcity has the highest percentage food secure households (75%) followed by Tana which accounts for 25% of all food secure households. From the results, it can be concluded that the Gish Abay subcity was vulnerable to food insecurity compared with other sub cities.

Fig. 3
figure 3

Household food insecurity access scale across sub-cities

Full size image

As can be seen in Fig. 3, HFIAS varies across sub-cities in Bahir Dar city. From the three sub-cities selected for the study, the severity of food insecurity was the highest (42%) at Gish Abay followed by Tana which was 34.5%, and the least was observed at Fasilo (23.2%). However, the Fasilo sub-city has the highest food secure households (75%) followed by Tana which was 25%. From the results, it can be concluded that the Gish Abay sub-city was vulnerable to food insecurity as compared to other sub-cities.

The household food insecurity scale also revealed relationships between the marital status of female-headed households. Figure 4 indicates that the severity of food insecurity was highest in divorced households (54%) followed by widowed households (24%). A total of 75% of the food -secure female-headed households were in the windowed category followed by 25% in the divorced category. In general, single female-headed households presented more severe food insecurity than other households did. In general, 87.9% of female-headed households are food insecure. Similarly, a study by Etana [52] reported that 93.2% of female-headed households experienced food insecurity, whether mild, moderate, or severe. The chi-square test was run to show the associations between marital status and household food insecurity access scale The results revealed statistically significant associations at P < 0.05 (χ2 (6)) = 14, P = 0.027.

Fig. 4
figure 4

Household food insecurity access scale about the marital status of respondents

Full size image

The status of female-headed households’ food security is indicated in Fig. 5. The results revealed that approximately 67% of the female-headed households were severely food insecure. The results also indicated that approximately 32% of the participants were mildly or moderately food insecure. These results revealed that the majority of female-headed households (87.9%) were food insecure. However, a study by Abraham and Abera [9] in Southern Ethiopia indicated that a higher percentage of food security was found among male-headed households (51.3%) compared to female-headed households (49.7%). Similarly, Neguse et al. [55] reported that the pooled estimate of food insecurity among female-headed households in their study area was 66.11%.

Fig. 5
figure 5

The status of household food insecurity using the HFIAS indicator

Full size image

3.2.1 Determinants of households’ food insecurity access scale (HFIAS)

As outlined in the methodology section, ordered logistic regression was utilized because the dependent variable, the Household Food Insecurity Access Scale (HFIAS), is an ordered categorical variable coded as follows: (0) food secure, (1) mildly food insecure, (2) moderately food insecure, and (3) severely food insecure. In this model, the reference category is food secure. Multicollinearity checks were performed, and no significant outliers were detected. As shown in Table 7, a total of eleven variables were included in the model, with seven variables showing significance at P < 0.001 and p < 0.05. The likelihood ratio (LR) chi-square statistic (11) indicated that at least one of the predictors has a regression odds ratio that is not equal to zero, confirming that the model is well-fitted. The small p-value from the LR test, p < 0.001, suggests that at least one of the regression coefficients is statistically significant. The pseudo R2 indicates that the predictor variables account for approximately 15.8% of the total variation in the model.

Table 7 Determinants of HFIAS using ordered logistic regression
Full size table

The ordered logistic regression results showed that Kebele Abinet is significantly more likely to fall into the food secure category compared to Kebele 02, the reference category, with an odds ratio of 5.887 and P < 0.001. This indicates that female-headed households in Abinet are more likely to be mildly or moderately food insecure but not severely food insecure. Similarly, Kebele Hidassie shows an odds ratio of 2.873 at p < 0.05. Additionally, for each one-unit increase in household size, the odds of being severely food insecure compared to being mildly or moderately food insecure increase by 1.221, assuming all other variables are held constant. This finding suggests that larger household sizes are associated with a higher likelihood of food insecurity.

Holding other variables constant, a one-unit increase in total income results in odds of being severely food insecure versus mildly or moderately food insecure that are 0.99899 times lower. Abraham and Abera [9] also reached similar results. This indicates that higher total income enhances household food security, as increased income is associated with a reduced likelihood of severe food insecurity. As shown in the Table 7, access to water significantly decreases food insecurity, with an odds ratio of 0.4717935 when controlling for other variables. Conversely, households in private rented accommodations are more likely to experience severe or moderate food insecurity, with an odds ratio of 4.402219 at p < 0.001. This suggests that renting privately, in comparison to owning a home, is linked to a greater risk of food insecurity. A unit increase in the number of households experiencing moderate or severe food insecurity is associated with an odds ratio of 1.221, which is statistically significant at P < 0.05. This indicates that larger family sizes require more resources for feeding, leading to greater food insecurity. This finding aligns with the research by Abo et al. [26], which reported that the odds of achieving food security decrease as household size increases, with an odds ratio of 0.302.

3.3 Analysis of the household dietary diversity score (HDDS)

As shown in Table 8, the dominant food groups consumed within the 24-h recall period were grains, tubers, and roots (93%) followed by pulses (56%) and vegetables such as cabbage (57%). Since cabbage is relatively cheap it is easily accessible to poor households. Similar results were also reported by Berlie [10]. For female-headed households, the consumption of fish, root crops, sugar/honey, meat, and butter was low.

Table 8 Food types consumed by the households for the last 24 h before data collection
Full size table

As a key informant from female-headed households mentioned, these severe constraints of dietary diversity emanated from their asset poverty and lack of job opportunities.

As indicated in the methodology section, HDDs are classified into three categories. They are nutritional food inadequacy (consuming fewer or equal to 3 food groups), medium food adequacy (consuming four food groups), and nutritional food adequacy (consuming five or more food groups). Regarding the Household Dietary Diversity Score (HDDS), the study conducted by Mengesha [56] revealed that seven food groups were reported to be consumed, with mean HDDS values of 3.42 for the total sample, 3.84 for participants, and 3.21 for non-participants.

As shown in Fig. 6, the majority of the households (54%) reported nutritional food inadequacy. However, only 28% were nutritionally food adequate. As HDDS is a proxy indicator of food security, the majority of the households led by females were food insecure.

Fig. 6
figure 6

The classification of household dietary diversity scores

Full size image

Household dietary diversity scores were also investigated across kebele administrations in the subcities. Accordingly, Shimbet (62%), Abinet (60.7%), and Kebele 02 (59%) were found to be nutritional food inadequacies (Table 9). On the other hand, Hidassie (39.7%) and Kebele 03 (31%) were nutritionally adequate. In general, approximately 54% were nutritionally inadequate where as 28% were nutritionally adequate. This finding revealed that there was great variation between kebeles with respect to HDDS. The findings of this study are consistent with those of [10, 52]. Similar studies conducted in the USA and Canada by Castell et al. [2] also found that women who experienced food insecurity had lower intakes of several nutrients, thereby increasing the risk of nutrient deficiencies.

Table 9 Household dietary diversity scores across the Kebeles administration in sub cities
Full size table

3.4 Spearman correlations between food security indicators

As shown in Table 10, the correlations between indices were investigated. This is vital for observing the strength, direction, and significant relationships among the indices. The results indicated that HFIAS and HDDS, as well as HFIAS and Kcal, have a negative relationship and these differences are also significant. The implications of these results indicate that the higher the HDDS is the lower the HFIAS. This means that food-insecure households consume little nutritional adequacy food. Similarly, the higher the number of kilocalories consumed by household is the lower the HFIAS of that household. The findings of this study are consistent with those of Faber et al. [57], which state that food-insecure households consumed a diet with less variety compared to food-secure households, as evidenced by the negative association between the household food insecurity access scale (HFIAS) and the dietary diversity score (DDS).

Table 10 The Spearman correlation between food security indicators
Full size table

As shown in Table 10, RCSI and HFIAS have positive relationships. In relation to this [52, 58] and [57] indicate that HFIAS and the rCSI were positively correlated. This means that the higher the rCSI is the higher the HFIAS. From the results, we can conclude that the indicators selected for this study were reliable and good measures of female-headed households’ food insecurity. Different studies such as Maxwell et al. [59] and Argaw [23] have reached similar conclusions.

3.5 Limitation of the study

The ongoing conflict in the Amhara Region and the nearly year-long internet blackout posed significant limitations that were beyond the authors' control. Nevertheless, by employing techniques such as using a modem and ensuring safety from conflict zones, we managed to conduct our research as effectively as possible. While this study focuses on female-headed households in Bahir Dar city, future research should also include male-headed households for comparative analysis in larger urban areas of the Amhara Region.

4 Conclusions and the way forward

At present, the number of female-headed households has increased from time to time. This resulted in problems for female-headed households in that they were able to feed and afford school materials to their children. If these problems continue sometime in the future, they will be a great burden on the country and could be an obstacle to development. Taking these predicaments into consideration, this study investigated the intertwined problems faced by female-headed households to secure food at the household level. Three food security indices HDDS, HFIAS, and rCSI were used for the study. These indices selected for this study are measures that can be used to identify female-headed households’ severity of food insecurity. That is, measurements of food security indices are imperative for targeting and implementing interventions for vulnerable female-headed households to safety net programs in the study area and elsewhere in the country. The findings of the study revealed that there were variations in the indices among sub cities, kebele administrations, marital status, and level of education. The results indicated that divorced and illiterate female-headed households were exposed to severe food insecurity. Similarly, the Shimbet and Abinet kebele administrations were classified as nutritionally food inadequate compared with the other kebeles in the selected sub cities. With respect to female-headed household HFIAS, the majority (66.7%) of the respondents were severely food insecure. The results of the ordered logistic regression indicated that total income, household size, kebele administrations, and housing ownership were significant predictors of the households' reduced coping strategy index (rCSI). These findings are similarly applicable to the household food insecurity access scale (HFIAS) for female-headed households. The food security indices selected are simple to use and are appropriate for monitoring and evaluating vulnerable female-headed households. That is, it helps to plan monitoring and interventions following shocks and to evaluate food security policies and programs.