The Indonesian Hospital Case: Strong Availability, Weak Waste Controls
This chapter examines what a national Indonesian hospital study actually found when it translated WASH-related facility conditions into domain scores and compared them with maternal sepsis mortality and patient satisfaction. The useful judgment is not whether the hospitals were simply “good” or “bad.” It is how to preserve the profile inside the average: strong measured availability in several domains, a pronounced waste-management bottleneck, and outcome associations that require caution rather than a simple success story.
What the study measured—and what it did not
The study used data from the 2019 Health Facility Research, known as Rifaskes, and analyzed 532 hospitals selected through systematic random sampling. The authors describe those hospitals as representing about 2,813 hospitals in Indonesia as of July 2018. The outcome data included the average number of maternities in the hospitals in 2018, the proportion of maternal deaths attributed to sepsis, and the percentage of patients reporting satisfaction in the hospital's most recent survey.
The WASH side of the analysis covered six service areas: water, sanitation, waste management, hand hygiene, environmental cleaning, and management. These were not a complete inventory of every WASH-FIT indicator. The researchers used the Rifaskes variables as proxies for WASH-FIT-related conditions because the national dataset did not contain every item needed for a full assessment. They calculated a score for each service and classified it as poor, adequate, or good. In this study, the good category meant a score above 75 percent.
That design gives the results a clear scope. The study describes a national hospital sample and its measured facility conditions. It does not observe every health-care facility in every country, and it does not directly measure all health-care-associated infections. Its maternal-sepsis measure is clinically important, but it is not interchangeable with a complete infection-surveillance system. Its patient-satisfaction measure is broader still: the authors could not establish whether respondents were rating WASH specifically or the hospital's services as a whole.
Those boundaries are not a reason to discard the case. They tell us what kind of case it is: a cross-sectional, secondary-data snapshot that can reveal patterns and priorities, but cannot show that a WASH intervention caused a later change in infection outcomes.
A mostly strong profile with one conspicuous bottleneck
The domain results resist a single headline. Water services were in the good category for 97.37 percent of hospitals. The corresponding figures were 84.02 percent for sanitation, 90.60 percent for hand hygiene, 79.32 percent for environmental cleaning, and 80.64 percent for management. If those five figures were all a reader saw, “WASH conditions were generally good” would seem reasonable.
Waste management changes the picture. Only 13.72 percent of hospitals were in the good category for waste management. Another 78.57 percent were adequate and 7.71 percent poor. The largest group was therefore not failing completely; it was operating below the study's good threshold. That distinction matters operationally. A hospital may have a waste unit, separate some materials, or possess treatment options while still lacking a complete, consistent, and safe chain from the point of generation to final disposal.
The pattern is more informative than an overall label. Water and hand-hygiene availability were relatively strong in the measured dataset; the study does not establish why. Waste management crosses more interfaces: clinical staff must separate waste correctly, containers and storage must be suitable, transport must be safe, treatment must occur, and hazardous materials must not be abandoned or mixed. A weakness at any of those points can remain hidden if the reader averages it with stronger domains.
This is why the case should not be summarized as “Indonesian hospitals had good WASH.” A more accurate statement is that the assessed proxy indicators were mostly in the good category, while waste management was predominantly adequate and rarely good. The difference is not stylistic. The second statement identifies where improvement may have the greatest practical value and avoids implying that strong water availability has solved the broader safety problem.
The study's structure is visible in this map. The six measured domains describe facility conditions; hospital characteristics shape the setting in which those conditions appear; maternal sepsis and satisfaction are modeled outcomes. The lines do not mean that a domain score directly caused an outcome. They show why the score must be interpreted within the institution that produced it.
A score may also be a marker of institutional capacity
The study found systematic differences among hospitals. Government hospitals generally had better water, sanitation, waste-management, and management services than private hospitals. General hospitals tended to have stronger sanitation, waste management, and management services than specialized hospitals. Lower-level hospitals generally had worse WASH conditions. At level D, for example, substantial shares of hospitals remained only adequate in sanitation, waste management, hand hygiene, environmental cleaning, and management, even though water services were good in almost all hospital levels.
These patterns are compatible with several mechanisms. Higher-level hospitals may have more beds, equipment, buildings, infrastructure, staff, and clinical services. Government hospitals may operate under different oversight, planning, funding, or training arrangements. General hospitals may have a larger service base and a different budget structure from specialized facilities. The study discusses monitoring, leadership, staffing, and financial systems as possible explanations, but those explanations are not fully tested by the data.
That distinction matters when comparing outcomes. Suppose a higher-level hospital has a better WASH score and a lower maternal-sepsis proportion. The score may be part of the explanation, but the same hospital may also have more staff, services, infrastructure, and referral capacity. The score and the outcome can travel together because they share an institutional cause, because one influences the other, or because of a mixture of both.
The statistical tests reinforce the need for context. Waste-management service was associated with all the listed hospital characteristics. Region and hospital type were associated with four WASH statuses, and hospital level with three. These are not minor background differences. They tell the reader that the facilities were not interchangeable observations drawn from one uniform environment. A comparison that ignores ownership, type, level, or geography risks treating institutional capacity as if it were a single WASH variable.
The practical implication is not to adjust away every context variable and make the score meaningless. It is to ask a more useful question: what intervention would change the WASH condition without assuming that the same intervention will work in every hospital? A waste-management plan for a tertiary government hospital may depend on different staffing, transport, and treatment arrangements than one for a lower-level or remote facility. The domain score identifies a need; the institutional context shapes the feasible response.
When the outcome association points the wrong way
The maternal-sepsis analysis is the study's most important warning against causal shortcuts. The researchers used the proportion of maternal deaths due to sepsis as the dependent variable and treated poor WASH status as the reference category in linear regressions. The reported relationships were not uniformly protective in the direction a reader might expect.
Adequate sanitation had a positive standardized association with the maternal-sepsis proportion, with β = 0.724. Adequate hand hygiene had β = 0.712, and good hand hygiene had β = 0.866. By contrast, adequate and good environmental cleaning had negative coefficients of −0.501 and −0.503, and adequate and good management had negative coefficients of −0.654 and −0.446. The study also found hospital level associated with the outcome.
A careless reading could turn the positive hand-hygiene coefficients into the claim that better hand hygiene increases maternal mortality. The study does not support that claim. The authors explicitly describe the sanitation and hand-hygiene findings as contrary to their hypothesis and offer reasons for caution. The sanitation variables captured the availability of outpatient and staff toilets; they did not establish safely managed sanitation or show that the facilities prevented pathogen transmission. The hand-hygiene variables captured facilities, not whether workers and visitors used them correctly and frequently.
Several explanations could therefore coexist. Hospitals treating more severe or complicated cases may invest in visible WASH improvements precisely because their risks are higher. Referral patterns may send the most difficult cases to better-equipped hospitals. A facility may have hand-hygiene stations while adherence remains poor. Reporting practices may differ between hospitals. Or unmeasured factors may influence both the WASH score and maternal outcomes. The supplied data do not identify which explanation dominates.
The correct lesson is narrower and stronger: a counterintuitive coefficient shows that the measured proxy and the outcome do not form a simple causal test. It does not disprove the biological or operational importance of hand hygiene. It shows why the next measurement must include practice, exposure, severity, timing, and other determinants rather than treating availability as the whole intervention.
Patient satisfaction is a different kind of outcome
Patient satisfaction adds a useful contrast because it is relevant to care quality but not a clean infection measure. The study reports an average satisfaction proportion of 83.04 percent across the hospitals. Among the WASH-related variables, only adequate environmental cleaning was significantly associated with satisfaction, and the association was positive relative to the poor reference category. The authors suggest that procedures might help cleaning staff work properly, which could contribute indirectly to a cleaner environment and better experience, but they also say that the explanation needs further research.
The model's low R²—0.049—makes the boundary clearer. The measured WASH variables explained only a small share of the variation in satisfaction. That is plausible because the authors note that satisfaction may reflect medical services, administrative services, and other supporting services. A broad satisfaction score should not be presented as evidence that WASH reduced infections, nor should a weak association be interpreted as proof that WASH does not matter to patients.
The same principle applies to the maternal-sepsis model, which had R² = 0.155. The model captures some variation, but most variation remains outside it. Statistical significance for a coefficient and explanatory power for a model answer different questions. A coefficient asks whether a measured relationship is distinguishable from a null association under the model's assumptions. R² describes how much variation the model accounts for. Neither one, by itself, establishes that changing a facility condition would change the clinical outcome.
The study's own limitations make the conclusion appropriately modest. It could measure the presence of WASH facilities and selected procedures, but not compliance with handwashing or cleaning protocols. It could not fully capture disability access, amenities, or safely managed sanitation. It relied on secondary data and could not control how the original data were collected or entered. These limits leave room for a useful management conclusion—waste management needs attention—without supporting the stronger claim that the measured WASH profile caused or prevented infections.
What this case establishes
The Indonesian hospitals provide three defensible lessons. First, a domain profile is more useful than a single overall label: strong water and hand-hygiene availability can coexist with a serious waste-management gap. Second, institutional differences shape both WASH conditions and outcomes, so a score may partly reflect resources, oversight, staffing, and hospital role. Third, an association that runs against the expected protective direction should trigger better measurement of proxies, practice, timing, and confounding—not a reversal of the underlying safety principle.
The case therefore brings the next question into focus. A facility can have a water source, toilets, hand-hygiene facilities, cleaning procedures, and a waste-management unit. What must happen between those conditions and a patient outcome for infection prevention to be plausible, observable, and testable? The answer lies in the missing links between facilities and infection outcomes.