The Accra Metropolitan Area (AMA) in Ghana was chosen for this study due to the availability of data (GSS, 2021a), especially the level of information on the current city sanitation situation (Boadi, 2019), the high percentage of the school-aged population (GSS, 2021a; GSS, 2021c), and accessibility to institutions holding data. The AMA, commonly called Accra, is situated in the southern part of the Greater Accra region of Ghana (Boadi, 2019). It comprises an area of 140 km² and has 10 sub-metropolitan districts with a population of 1,281,570, of whom 25% are of school age (GSS, 2021a; GSS, 2021c).

In AMA, schools are differentiated by the age of the students (Figure 1) and their ownership. Private and public senior high schools along with basic schools (a combination of primary and junior high schools) are spread across all 10 sub-metropolitan districts, with an average of two private senior high schools and 123 private and public basic schools per district (GES, 2020). Two of the 10 districts have no private senior high school (Nkonde, 2022). In Ghana, education is managed by the Ministry of Education through the Ghana Education Service (GES), which formulates and implements policies in partnership with other organizations, such as the United Nations Children’s Fund (GSS, 2019; African Union & UNICEF, 2020). The GES is currently implementing free compulsory basic and secondary education (FCUBE) (GSS, 2019).

This case study used a mixed-methods approach, quantitative data was gathered from secondary data sources e.g. Ghana Statistical Service, structured observations and interviews, while qualitative data was gathered from secondary sources e.g. reports and interviews. This approach was used so that the data could be validated. Data collection was undertaken between July and September 2022 after obtaining ethical approval from the IHE, Delft, The Netherlands (IHE-RECO-2022-003).

The schools were selected using stratified convenience sampling, with one school chosen from each category of schools present in each district. Twenty six schools were selected: one public senior high school, one private senior high school, 17 public basic schools, and seven private basic schools. Secondary data were collected from the Ghana Statistical Service (GSS) and GES on the school population, school policies, and sanitation situation in the schools and cities (Table 1).

A total of 80 key informants were interviewed using a semi-structured approach. The headteachers of the selected schools were interviewed, and they also identified students who could be interviewed (Table 2). The School Health Education Program (SHEP) coordinators were interviewed from each district (Table 2). Other key informants were also purposively sampled based on their roles, responsibilities, and knowledge (Table 2), but harder-to-access key informants such as private operators and utility employees were sampled through snowball sampling (Table 2). All interviews were conducted in English, either at the offices of these individuals or at the schools; the interviews lasted between 30 and 45 min, and informed consent was obtained prior to participation. Interviews with the students were conducted in the presence of a parent, guardian, or headteacher, and informed consent was obtained from both the student and the adult present. Notes were taken during the interviews and analyzed using themes related to the objectives of this research. After each interview with the headteachers, structured observations of the school sanitation systems were conducted. The structured observation sheets were developed based on the WHO Sanitation Sanitary Inspection Forms (WHO, 2018) and is included in Supplementary Table S1.

SFDGs were generated using the online SFDG generator following the standard methodology in the SFD manual (SFD PI, 2018). Three SFDGs were developed to explore the impact of school sanitation on the excreta flows of the city. SFDG1 depicts the excreta flow of the current population in AMA; this was used to understand the city sanitation situation in 2022. SFDG2 depicts the current excreta flow from only the schools in AMA; this was generated to visualize the status of school sanitation. SFDG3 depicts the excreta flow of the city by including the inputs from the school sanitation; this SFDG was generated by removing the excreta flows of the students from their home toilets and incorporating the school sanitation excreta flows of the pupils from the school toilets.

To develop SFDG3, the students’ excreta flows had to be split between their households and schools. The proportions of pupils’ excreta disposed of at school were calculated using the number of waking hours per day (h) and the number of days in a year that a pupil is in school (d). This depends on the type of attendance, i.e., residential (R) or day (D). Not all students use toilets at school, so the proportion of students using school sanitation facilities (C) was needed. For each student using the school toilet, some excreta may still be deposited at home, so the proportion of excreta deposited in the school sanitation system (k) was needed. These data were used to calculate the school sanitation usage population equivalent (SUP _e ) given by Equations 1–3

School sanitation usage population equivalent: S U P e = S U P e R + S U P e D . (1)

School sanitation usage population equivalent for day (D) or residential (R) students: SUP e D = S p D × t D × C D × k D or S U P e R = S pR × t R × C R × k R . (2)

The proportion of time spent by the students in school in a year: t D = h s D h W × d D 365 or t R = h s R h W × d R 365 , (3)

The accuracy of the various data sources varied; to ensure that the results were not unduly influenced by inaccurate data, a two-step validation process was used. First, the data used to generate the SFDGs were color-coded based on the level of triangulation to show the level of confidence. Data with a wide range of sources were coded green, while data from a single source without triangulation were coded amber. Sensitivity analysis was performed using different scenarios to assess the impact of variations in the amber-coded data. The range of results was assessed, and if this range was large then the middle value was chosen; however, if there was little difference in the results, the value deemed by the authors to be most valid was chosen for generating the SFDGs.

The SFDGs were compared using trend graphs (Martinez, 2016). Trend graphs were developed to illustrate the differences in sanitation provision over time (Martinez, 2016; Safi et al., 2022); however, in the present study, they were used to compare differences in the safe management of sanitation services between households and schools in AMA (SFDG1 vs. SFDG2) and to explore the impact of school sanitation on citywide sanitation (SFDG1 vs. SFDG3).

Figure 2 shows the SFDG for AMA in 2022 based on household sanitation. In AMA, 1% of the population was practicing open defecation, and the excreta of approximately 99% of the population was safely contained through various sanitation technologies (Figure 2), including sewers, biodigesters, septic tanks, holding tanks, ventilated pit latrines, and ordinary pit latrines (Colan-Consult, 2020; GSS, 2021b). While 98% of the captured excreta was emptied (Figure 2) only 69% of this excreta reached the treatment facilities (Colan-Consult, 2020), which were approximately 85% efficient (KI-014; SSGL, 2017). This means that the excreta of only 62% of the population in AMA was safely managed (Figure 2).

The number of school-aged children actually attending schools in AMA was 235,761 (Table 3), which was 74% of the school-aged population (GSS, 2021a; GSS, 2021c). School enrollment in AMA is higher than the average rate of 66% for Sub-Saharan African countries (African Union & UNICEF, 2020; White, 2021; Onukwue, 2022), and this is probably due to FCUBE (GSS, 2019). Ghana’s education system has three levels (Figure 1), but the primary and junior levels are combined and collectively referred to as basic schools (KI-001; KI-005; GES, 2020). Schools are either managed by the state (public) or privately owned (private), and the financing, operation, and maintenance of school sanitation infrastructure differ between these two categories (KI-005; GES, 2020).

Only 3.5% of the overall school population in AMA are residential students (Table 3); this is important in terms of sanitation as they would use the school sanitation facilities more often than the day students who live at home during the school term. Five senior high schools had residential students who only returned home at the end of each term. It was estimated that the residential students spend 275 days per year in school, including weekends during the school term (GES, 2020; Sky News Gh (SNG), 2022), while the day students spend 185 days per year in school (HT-026; KI-012; Ghana Students, 2022). Residential students only account for 0.6% of the AMA population, so no adjustments were made to the population used to generate SFDG3. Knowing the amount of time that the day students spend in school is essential to understanding their sanitation practices. Typically, students in Ghana spend 9 h in school from 07:30 to 16:30 (GEO, 2022). Studies have shown that students under the age of 12 are awake for between 12 and 14 h a day, while those aged between 12 and 18 years of age are awake for between 14 and 16 h daily (Reid et al., 2002; Nahmod et al., 2017). If an average of 14 waking hours is considered, then the day students spend over 64% of their waking hours in school on school days (GEO, 2022; Reid et al., 2002; Nahmod et al., 2017).

The GES policy provides guidelines for the use of adequate and safe sanitation facilities and recommends the use of specific technologies, such as septic tanks and ventilated improved pit (VIP) latrines (GES, 2014). In AMA, septic tanks and VIPs are the most common technologies for both schools and households (Figure 3). These are also typical sanitation technologies used in schools in Sub-Saharan Africa (Harvey and Adenya, 2009).

Nearly half of the students reported using the school toilets only when needed, and approximately one-third of the students reported using the toilets either once or twice a day, but 28% of the students interviewed did not use the school toilets at all (Figure 4). Some headteachers (19%, n = 26) noted that not all school students used the school toilets (HT-001 to HT-026). Hence, as with household toilets, the availability of school toilets does not equate to usage. These data were fed into the estimation of the parameter C in Equations 2 for the day students (Table 4).

The students stated that their school toilet usage was dependent on the level of toilet cleanliness (64%, n = 39); the other factors highlighted by the students included security and privacy (8%, n = 39), student-to-toilet ratio (15%, n = 39), and type of toilet technology (49%, n = 39); it should be noted that many of these factors are interrelated (Garn et al., 2014; Schmitt et al., 2017; Shao et al., 2021). Similar findings have been reported based on studies in China and Nigeria, where one of the main barriers to school sanitation usage included toilet cleanliness (Abigail et al., 2012; Shao et al., 2021). This was supported by the structured observations cleanliness scored lower than all other parameters for public and private schools (Figure 5).

The structured observations (Figure 5) were categorized into two themes physical infrastructure (accessibility, menstrual health management (MHM) features, privacy, and security), and operation and maintenance (O&M) (water availability and cleanliness).

This is due to the differences in funding requirements. For parameters related to physical infrastructure, both private and public schools had fair to good scores ranging from 50% to 100%, while features related to O&M were rated low at <40% (Figure 5). The lack of funding for O&M was the reason given for the lack of cleanliness by headteachers from both public and private schools (HT-002; HT-003; HT-013; HT-014). This impacted the sanitation system in public schools as many were not emptied when full, resulting in leakage into the environment (KI-004). This was supported by the structured observations which noted that 38% (n = 18) of the sanitation systems in public schools was not emptied when full. Half of the headteachers from the public schools (n = 18) noted that their schools had experienced overflowing sanitation containment systems. This information fed into the estimates for the emptying and transportation of fecal sludge in Table 4.

Private schools outperformed public schools in every category, as shown in Figure 5 this was thought to be due to the lack of O&M funding in public schools and the additional pressure on their sanitation systems related to increased student numbers due to FCUBE (GSS, 2019). The low performance of the O&M-related features compared to the infrastructure features (Figure 5) illustrates that although thought and investment were put into building school sanitation infrastructure, the public schools were struggling to manage their O&M due to a lack of continuous funding. This is in line with a UNICEF and WHO (2022) study that found that although there has been significant capital investment in the construction of public-school sanitation, there is a lack of sustainability as emptying, transportation, and treatment of fecal sludge are often neglected due to a lack of continuous funding. Similar situations have been reported in other developing countries, where O&M could not be sustained over a long period of time despite using novel approaches, such as community involvement, preventive maintenance programs, and packaged operation and maintenance (Chatterley et al., 2013; Duijster et al., 2022).

The data in Table 4 were used to generate SFDG2 and SFDG3.

The population equivalent for residential students SUP _eR was calculated using the data in Table 4 and Equations 2, 3. A sensitivity analysis was performed on the calculated population equivalent for day students SUP _eD (Equations 2, 3) due to the uncertainty of the data for the parameter C _D (proportion of children actually using sanitation whilst in school). The tested C _D values were related to the original assumption of C _D = 0.5 in Table 4. Three scenarios were tested by varying the C _D value, and the results are found in Table 5. Scenario C with C _D = 0.75 was used in SFDG3 rather than C _D = 0.5 as assumed in Table 4 as students are in school for 9 h out of their 14 waking hours (Table 4), which is equivalent to 64% of their waking time. Although AMA has a relatively high percentage of the population attending school (18.4%, Table 3), this only accounts for approximately 4% of the excreta flow from AMA owing to usage patterns, such as usage frequency and time spent in school.

Sensitivity analysis was performed on the assumptions made for the emptying and transportation of fecal sludge from schools in Table 4. Five scenarios were tested around these assumptions, and the results are shown in Table 6. The population used to test these scenarios was the total student population of 235,761. The baseline or reference values for these five scenarios were the SFDG1 emptying and transport efficiencies (Scenario 1, Table 6). In Scenarios 2 and 3 (Table 6), the emptying efficiency decreased by 10% and 20%, respectively, while the transport efficiency remained constant; this unexpectedly caused the percentage of safely managed excreta to increase from 54% to 65% because the unemptied fecal sludge was considered to be safely contained, which is related to groundwater contamination (Table 4) rather than the overall public health risk. The assumptions noted in Table 4 (Scenario 4, Table 6) related to emptying and transport resulted in a 50:50 split between safely and unsafely managed excreta. It was observed that for each drop of 10% in transport efficiency, the overall safe management rate dropped by 4% (Scenarios 1, 4, and 5, Table 6). After assessment, the data from Scenario 4 was used to generate the SFDGs.

SFDG2 (Figure 6) shows the excreta flows from schools in AMA; 71% of the excreta deposited in schools was safely contained using onsite and offsite sanitation technologies. Very few of the schools were connected to sewers (Colan-Consult, 2020; Boadi, 2019, KI-003), as the sewer network mostly serviced the central business district (Colan-Consult, 2020; Boadi, 2019, KI-003), where few schools are located.

While 27% of the excreta deposited at the schools was not contained and leaked into the immediate environment (Figure 6). A total of 68% of the excreta was emptied, but 40% of this flow ended up in the environment untreated (Figure 6). Therefore, 29% of the excreta from schools reached a treatment plant, and only 41% of the excreta from the schools was safely managed (Figure 6).

Comparisons were performed between school (Figure 6) and household sanitation (Figure 2) in AMA for each stage of the SSC (Figure 7). This shows that school sanitation service delivery was worse at every stage of the SSC compared to household sanitation. Surprisingly, Figure 7 shows that open defecation in schools is higher than in households, as some schools do not have adequate sanitation facilities (Figure 3). The emptying and transport stages of the SSC are worse in schools than in households owing to a lack of continuous funding and investment for O&M. This means that (on average) the excreta disposed of in schools was less safely managed than that disposed of in households in AMA (Figure 7).

The data from SFDG2 were converted to school usage population equivalents using Equations 1–3. The final SFDG3 (Figure 8) shows the sanitation service delivery at the citywide level, including both household and school sanitation, using data from Table 4. Trend graphs (Figure 9) were then used to compare SFDG1 (Figure 2) to SFDG3 (Figure 8) to observe the impact of school sanitation on citywide sanitation. Impacts highlighted by the trend graphs (Figure 9) include the following:

• Uncontained excreta increased from 1% to 2% due to a lack of or poorly maintained containment in schools. It should be noted that the situation in the schools may be worse than that illustrated in the SFDG, as it was assumed that students who did not use the school toilets used their toilets at home and that fecal sludge is safely managed if it is not collected.

Limited data were required to produce SFDG2 (Figure 6) for the schools (Table 3) compared to the data required to split the excreta flow and integrate it into the citywide SFDG (Table 3, SFDG3). SFDG2 (Figure 6) could be used to highlight the main challenges in assessing and monitoring the progress of school and healthcare sanitation facilities in relation to the SDGs across the SSC. The data collected and the process developed to split excreta flows highlights the complexity of exploring sanitation beyond a household level on a citywide scale. The process highlighted in this work could be used as a guide for future studies, as it could help reduce the time required for data collection and analysis.

Although 18.4% of the people in AMA attend school, and considering the state of school sanitation in the city, the citywide sanitation services were only adversely impacted by 1%, which is the sensitivity limit of the SFDG tool used. This means that in cities with a lower percentage of the population attending school and a low number of residential students, the impact will be insignificant at the citywide level. This is because the school excreta flows of the day students (SUP _eD , Equation 2) only account for between 6% and 16% (Scenarios A-C, Table 5) of their annual excreta flow. This is due to the time spent in school during the year (t _D = 0.34, Table 4). For residential students, this value was 76% of their annual excreta flow due to the time spent in school each year (t _R = 0.75, Table 4). Hence, there will be a higher impact of school sanitation in cities with a higher number of residential students compared to the total city population. This indicates that the main factors influencing the institutional impact on sanitation at the citywide level are the percentage of the population at the institution, and the proportion of time spent in the institution per year. This means that larger residential institutions, such as hospitals, barracks, and universities, will have a bigger impact on citywide sanitation. The present research demonstrates the effects of where people spend their day on city sanitation flows. As approximately 70% of the people in AMA are employed or involved in income-generating activities (AMA, 2020), with the majority of people working outside their homes, it is assumed that these people will not be using household sanitation for the majority of their waking hours. According to Mazeau et al. (2014), 63% of men and 41% of women in one area of Accra use two or three different toilets every day. It is hypothesized that this flow of excreta is likely to have the biggest impact on the sanitation landscape of the city in relation to non-household toilets.

The findings of this study hide the fact that the excreta flows in schools are seasonal. Although these flows are insignificant at the citywide level when considering annual flows, there will be significant health and environmental impacts during the school terms. Thus, it may be more important to consider the school population as a whole, as in SFDG2 (Figure 6), and compare it to the household city sanitation situation (Figure 7) with regard to impact.

Fieldwork for this study was conducted between August and September 2022 coinciding with the schools’ summer holidays, which limited access to the schools, staff, and students. Gaining official data on private school sanitation was challenging as this information was not publicly available, so interview data were used. The SFD tool was designed for situational analysis at the citywide level and for use in advocacy and decision-making; it was not designed for accuracy or precision, so the sensitivity of the tool was only 1% of the population. The tool does not display decimal outputs but rather rounds them to the nearest integers. The SFDGs presented in this work are based on assumptions that have been justified (Table 4), which is linked to the use of SFDGs. More detailed information would therefore be required for an accurate or precise understanding of school sanitation flows at the city level. In this study, excreta was considered as a whole rather than its fractions, namely, urine and feces; if the patterns of defecation and urination were considered separately, the present findings may be significantly altered.

With regard to the residential students in AMA, no adjustments were required for the area's population. For large institutions, such as residential schools and regional hospitals that people commute to from outside the area, more data is required in regards to the locations of their homes so that the excreta flows can be added to the areas where they temporarily reside. It should be noted that some children from AMA also attended schools outside the study area, which was not considered in this analysis.

To obtain an overview of the sanitation conditions in schools across AMA, stratified convenience sampling was used with at least one school sample per district. In total, 18 public and 8 private schools were sampled, and the conditions of the school sanitation systems were very different for these two types of schools (Figure 5). This means that the school SFDGs and data will be skewed toward the poorer conditions in the public schools. Hence, we recommend that the different types of schools be taken into consideration when sampling in the future to replicate this study. The official statistical records may also be incorrect in some countries as there may be financial incentives for school registration, so the number of enrolled students may be inflated.