Indicators to complement global monitoring of safely managed on-site sanitation to understand health risks

Inadequate sanitation is linked to multiple faecal exposure pathways and health risks, and the presence of a toilet alone does not ensure health protection. The SDG 6.2 indicator for safely managed sanitation improves upon MDG-era monitoring by considering management from toilet to final treatment, but global indicators developed by the JMP do not capture several key safety dimensions highlighted by WHO’s Guidelines on Sanitation and Health. There is uncertainty and limited research on the scope and formation of indicators for safely managed sanitation, particularly for on-site systems predominant in low- and middle-income countries. This paper asks whether complementary, practical indicators—collectable via household surveys—can better identify health-relevant risks not covered by global monitoring. Using large-scale primary data from seven countries, the study evaluates five complementary indicators along the sanitation service chain to determine how much they change the classification of on-site systems deemed safely managed and in which contexts each indicator is most critical.

The paper situates its contribution within debates on SDG 6.2 monitoring. Prior work critiqued shared sanitation classification and highlighted challenges in urban sanitation monitoring and means of implementation, but offered limited review of safe management indicators across the on-site chain. Rose et al. framed sanitation safety through a community lens without proposing indicator alternatives; Beard et al. called for revised facility categories but did not assess chain-wide safety indicators. An Austrian national framework proposed complementary indicators largely applicable to sewered, high-income contexts. WHO and JMP encourage complementing global indicators with context-specific national metrics. Evidence underscores overlooked exposure pathways: mechanical transmission by flies and rodents; groundwater contamination associated with on-site sanitation in certain hydrogeological conditions; and occupational risks to sanitation workers during emptying. These gaps motivate complementary indicators tailored to on-site systems prevalent in LMICs.

Design and setting: The study draws on SNV’s standardised performance monitoring framework applied in 2018–2019 across seven countries in Asia and Africa. A total of 31,784 households were surveyed (26,436 urban; 5,348 rural) in 18 urban and 13 rural districts (with some district centres). Data were collected via mobile tools (AKVO Flow) by trained enumerators alongside local governments. Sampling and ethics: Multi-stage sampling selected wards/districts as PSUs; secondary sampling units were country-specific (e.g., villages in Indonesia), with proportional allocation and systematic household selection. Surveys were translated, consent obtained, and data anonymised. UTS HREC approved data use (ETH20-5620; 6 July 2021). Indicators: In addition to JMP global indicators (improved facility; contained without overflow/discharge to surface; disposal in-situ if not emptied; ever emptied), five complementary indicators targeted key exposure pathways:

• Animal access to excreta: observation/self-report on rat and fly access (e.g., slab presence, water seal, pan cover, vent screening). Improved facilities accessible to rats/flies were flagged unsafe.
• Groundwater contamination risk: applied only to systems classified as contained by global criteria. A risk matrix (AGROSS, British Geological Survey) combined household-reported containment depth with secondary data on groundwater depth and soil type at sub-district/neighbourhood scale to classify high vs low risk. Infiltration depth thresholds: <5 m always unsafe; >20 m safe; 5–20 m unsafe in coarse sands/gravel/fractured rock. Households lacking depth data (≈3%) were excluded from this indicator.
• Timely emptying (unemptied): for never-emptied systems, compared years-in-use to a containment-type- and country-specific timely emptying threshold derived from average volumes, users, and sludge accumulation rates (literature- and programme-based): indicative values ~40 L/c/yr for dry/composting pits; 60 L/c/yr for wet pits; 80 L/c/yr for septic tanks. Unemptied systems exceeding thresholds were flagged overdue.
• Timely re-emptying (emptied): for previously emptied systems, compared time since last emptying to thresholds as above to flag overdue re-emptying.
• Emptying health and safety: assessed whether someone entered the pit/tank during emptying and whether minimum PPE (boots, gloves, mask) was used; both entering and inadequate PPE were considered unsafe. Analysis: Indicators were first computed per step and then combined cumulatively across the on-site chain up to emptying (transport and treatment could not be reliably assessed by household surveys). Prevalence ratios (SPSS v28) examined associations between contextual variables (urban/rural; wealth; groundwater depth; dry vs wet; pit vs tank; age >5 years; depth) and indicator failure, with significance at two-sided p<0.05. Results were presented as country averages (to avoid skew from unequal sample sizes) and disaggregated further in supplementary materials.

• Coverage context: On average across countries, 81% used improved sanitation; 79% used improved on-site systems; 10% practised open defecation. Shared improved (classified as limited) averaged 17% urban and 6% rural. In rural areas, 89% used pits and 11% tanks; urban areas reported a mix of pits and tanks.
• Animal access (complementary vs global improved): 14% of respondents used improved toilets accessible to rats/flies. Urban improved coverage reduced by 18% when considering animal access (rural: 8%). Country reductions ranged from 1–2% (Laos; urban Nepal) to 28–29% (Tanzania; Zambia), with within-country peaks up to 51% (Zhemgang, Bhutan). Poorer households and dry toilets showed higher prevalence of access.
• Containment (global): 66% used contained systems; 14% uncontained (8% outlet to surface; 4% flooded/overflow; 1% both). Urban uncontained averaged 20% (Bangladesh 57% uncontained, largely due to outlets). Uncontained more prevalent in urban, wet containments, tanks, deeper containments, and deeper groundwater; flooding/overflow higher in dry toilets, pits, and poorer households.
• Groundwater risk (complementary): Among systems classified as contained, 35% on average posed high groundwater contamination risk (0% Bhutan; 78% Tanzania). Adjusting to only areas where ≥25% households used groundwater for drinking reduced average to 24%, eliminating risk in Tanzanian cities (groundwater not used for drinking) and slightly lowering risk in Nepal and Zambia.
• Emptying status (global): 10% of improved on-site systems had ever been emptied (lowest 1–4% in Zambia, Bhutan, Laos; highest 32% in Bangladesh). Emptying more likely in older, wet, and urban systems.
• Overdue emptying (complementary): Of the 67% with unemptied improved containments, 21% were overdue for emptying (beyond timely threshold). Largest reductions: Indonesia 42%, urban/rural Nepal 27%; smallest: Zambia 6%. Overdue status was more prevalent in urban, wet, and non-poor households among never-emptied systems.
• Overdue re-emptying (complementary): Very low overall impact (0.4% of improved on-site systems), max ≈0.8% in Indonesia.
• Emptying practices and OHS (complementary): Only 8% of improved systems were emptied without someone entering; entering more common when emptied by households/tenants (24%) vs manual providers (15%) vs mechanical (3%). Rural areas and wet containments had greater entering risk. Adequate PPE use was reported for only 3% of improved systems that had been emptied; Bangladesh showed the largest reduction (≈29% emptied without minimum PPE). Inadequate PPE more prevalent in urban areas and poorer households.
• Cumulative impact across the on-site chain (excluding transport/treatment): Using global indicators, 56% had safely managed on-site sanitation up to emptying; using complementary indicators, this fell to 16% overall. Urban areas saw a larger drop (to just over one-fifth of the global estimate) than rural (to about one-third). Country examples: Bangladesh 26% (global) vs 2% (complementary); Tanzania 52% vs 1%; Bhutan least impacted.
• Contextual associations (prevalence ratios): Urban settings were associated with higher groundwater risk, overdue emptying, and inadequate PPE; rural areas with higher likelihood of entering to empty. Shallow groundwater (<5 m) increased groundwater and animal-access risks but reduced overdue emptying and entering. Dry containments/pits had higher animal access and groundwater risks but lower uncontained status and entering. Older containments (>5 years) were more often uncontained and overdue but less prone to animal access.

Findings show that relying solely on SDG global indicators can substantially overestimate the safety of on-site sanitation from a public health perspective. Incorporating five complementary indicators—animal access, groundwater risk, timely emptying, timely re-emptying, and emptying OHS—reduced the estimate of safely managed on-site services from 56% to 16% across seven countries. Groundwater risk had the largest average impact (35% of contained systems reclassified unsafe), followed by overdue emptying (21%) and animal access (14%). Emptying-related indicators had lower overall impact due to low emptying prevalence, but pose significant risks where emptying occurs (e.g., very low PPE use). Context strongly influences which risks dominate; thus, monitoring and investment decisions should be made at sub-national levels with disaggregated data. Adopting these indicators can better direct resources to mitigate key exposure pathways (e.g., focusing on hydrogeology-driven groundwater risks or improving emptying safety and service availability) and aligns with CWIS and WHO sanitation safety planning principles. The results argue that the global definition of safely managed sanitation should not be equated with health safety and that complementary, context-appropriate indicators are needed for decision-making.

This study demonstrates that integrating five practical, survey-based complementary indicators into monitoring systems provides a more comprehensive assessment of health risks from on-site sanitation. Applying these indicators reduced the proportion of systems classified as safely managed by global criteria to less than one-third of the original estimate, with groundwater contamination risk, animal access to excreta, and overdue emptying driving most reclassification. National and local monitoring frameworks should adopt these tested indicators, selecting those most relevant to context (e.g., hydrogeology, technology mix, service availability). Future research should refine indicator methods with locally specific data (e.g., sludge accumulation rates, groundwater risk matrices), develop complementary indicators for transport and treatment stages, and explore innovative data collection (sensors, spatial mapping, citizen science). Strengthening these approaches can improve investment targeting, reduce environmental pathogen loads, and inform post-2030 monitoring objectives.

• Household surveys cannot accurately assess transport and treatment safety; cumulative analysis excluded these stages. The global indicators for these steps (proportion delivered; treatment of solids/liquids) were not directly verifiable here.
• Sampling, while diverse, was not nationally or globally representative; findings reflect surveyed cities/districts only.
• Several measures rely on self-report (e.g., emptying, PPE use, containment depth/age), which may introduce recall or social desirability bias; some depth data were missing (≈3%) and excluded for groundwater risk.
• The groundwater risk matrix applies generalized thresholds; local hydrogeological variability and other contamination sources may confound attribution to sanitation.
• Timely emptying thresholds used literature- and programme-based assumptions for volumes, users, and sludge accumulation; limited local data may affect accuracy.
• Analysis of contextual variables used independent prevalence ratios and did not control for confounding; causality cannot be inferred.
• Some global indicator components (e.g., effluent outlets) were not assessed in all rural contexts within SNV’s framework, potentially underestimating uncontained systems in those areas.