Fiduciary risk, corruption, and oversight in global health aid: governance interventions and accountability outcomes - a systematic review

¹School of Professional Studies, Clark University, Worcester, Massachusetts, USA, ²Department of Human Resource Management, Faculty of Management Sciences, University of Benin, Benin City, Nigeria, ³College of Public Health, Kent State University, Kent, Ohio, USA, ⁴Department of Hospital Epidemiology and Infection Control, The Johns Hopkins Hospital, Baltimore, MD, USA

^&Corresponding author
Grace Aiwonose Ibe, School of Professional Studies, Clark University, Worcester, Massachusetts, USA

Introduction    

Global health aid represents one of the most consequential conduits of international development finance. Since the early 2000s, a confluence of multilateral initiatives including the Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM), Gavi, the Vaccine Alliance, and major philanthropic actors such as the Bill & Melinda Gates Foundation, alongside bilateral programmes administered by the United States Agency for International Development (USAID) and the United Kingdom's Foreign, Commonwealth and Development Office (FCDO), has substantially increased the volume of external resources flowing toward health systems in low- and middle-income countries (LMICs). Conservative estimates place annual global health official development assistance (ODA) at approximately USD 40-44 billion during the 2015-2022 period, with additional substantial disbursements through philanthropic and non-governmental channels [1].

Despite the transformative potential of these financial flows, their integrity remains persistently threatened by fiduciary risk, defined in this review as the probability that resources intended for health outcomes are diverted, misappropriated, or inefficiently managed through corrupt practices or weak oversight systems. Corruption in the health sector constitutes a structural determinant of health system performance and population health outcomes. The World Health Organization estimates that approximately 10-25% of total global health expenditure is lost annually to inefficiency, leakage, and corruption, equivalent to roughly USD 500 billion globally [2]. In contrast, global health ODA represents a smaller but highly vulnerable subset of these financial flows, within which fiduciary risks can disproportionately affect service delivery in low- and middle-income countries.

Despite the breadth of literature on health sector corruption and aid governance, systematic synthesis integrating fiduciary risk, governance interventions, and accountability outcomes in global health aid remains limited. Existing reviews tend to be fragmented across disciplinary and thematic boundaries. Vian [3] provides a foundational taxonomy of health sector corruption but predates the expansion of contemporary governance instruments such as performance-based financing and digital financial management systems. Previous studies have also highlighted corruption in health systems and weaknesses in aid governance mechanisms in low- and middle-income countries [4].

More recent scoping and policy-oriented reviews have examined governance and accountability mechanisms in global health but remain largely intervention- or sector-specific, without systematically linking corruption typologies to governance responses and measurable accountability outcomes across aid delivery chains. Similarly, broader governance synthesis work highlights institutional and political determinants of aid effectiveness but does not provide an integrated empirical synthesis of fiduciary risk mechanisms within health aid systems.

Accordingly, these studies do not fully integrate the fiduciary, institutional, and accountability dimensions of global health aid governance. This review addresses these limitations by: (1) focusing explicitly on official development assistance (ODA)-financed global health aid rather than domestic health systems alone; (2) integrating corruption typologies with governance intervention effectiveness; and (3) applying structured quality appraisal using the Mixed Methods Appraisal Tool (MMAT) across heterogeneous study designs to strengthen comparability and synthesis rigour.

This systematic review aims to synthesize evidence on fiduciary risk and governance interventions in official development assistance (ODA)-financed global health aid, and to examine how these interventions relate to accountability outcomes across recipient-country contexts. This review addressed the following research questions: What are the principal typologies of fiduciary risk documented in global health aid, and how are they distributed across regions and aid mechanisms? Which governance interventions, financial management systems, independent audit institutions, performance-based financing, and social accountability mechanisms are associated with improved accountability outcomes, and under what conditions? What contextual or institutional factors moderate the effectiveness of these governance interventions across different settings?

Methods     

Study design and protocol

A formal review protocol was not prospectively registered. Key methodological decisions, including eligibility criteria, search strategy, and quality appraisal approach, were, however, specified in advance of full-text screening and were followed consistently throughout the review process. This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [5]. The review adheres to the methodological standards established by the Cochrane Collaboration and the Campbell Collaboration for governance and policy-relevant systematic reviews.

Eligibility criteria

Studies were included if they: (a) addressed fiduciary risk, corruption, financial mismanagement, or governance failures in the context of global health aid or health sector ODA; (b) examined oversight mechanisms, accountability interventions, or anti-corruption strategies applicable to health aid; (c) reported empirical findings using quantitative, qualitative, or mixed methods designs. In addition to primary empirical studies, systematic reviews and policy analyses addressing fiduciary risk or governance interventions in global health aid were eligible for inclusion. These sources were included to capture higher-order synthesis of governance evidence, particularly where primary studies had already been aggregated elsewhere and were treated analytically as a distinct category from primary empirical studies throughout data extraction and synthesis; and (d) were published in English in peer-reviewed journals or as grey literature from recognized international organizations (e.g., WHO, World Bank, OECD, Global Fund), as well as national audit institutions, bilateral agency reports, and reputable non-governmental organizations engaged in health systems governance and anti-corruption research. Studies were excluded if they addressed domestic health spending without an aid component, focused exclusively on humanitarian emergency response without a governance dimension, or lacked sufficient methodological transparency for quality appraisal. To ensure methodological rigor, grey literature sources were included only where sufficient detail on study design, data sources, and analytical approach was provided to permit quality appraisal using the MMAT framework.

Search strategy

Systematic searches were conducted in PubMed/MEDLINE, Web of Science (Core Collection), Scopus, and Google Scholar. The search strategy employed a Boolean combination of terms across three conceptual domains: (i) corruption and fiduciary risk ("corruption", "fiduciary risk", "misappropriation", "embezzlement", "fraud", "elite capture"); (ii) global health aid ("health ODA", "global health aid", "development assistance for health", "Global Fund", "PEPFAR", "Gavi"); and (iii) governance and accountability ("accountability", "oversight", "public financial management", "transparency", "audit", "social accountability"). Reference lists of included studies and key systematic reviews were hand-searched to identify additional relevant publications. The complete search strategy for PubMed/MEDLINE, including Boolean operators and field tags (e.g., [MeSH Terms], [Title/Abstract]), is provided in Annex 1A in accordance with PRISMA 2020 reporting standards. The search strategy was iteratively refined to balance sensitivity and specificity, with pilot searches conducted to ensure retrieval of key benchmark studies identified a priori. The PRISMA 2020 flow diagram illustrating study selection is presented in Figure 1.

Google Scholar was searched using an abbreviated version of the primary Boolean string, reflecting the platform's limits on search-string complexity and its lack of field-tag/MeSH support. Given Google Scholar's relevance-ranked (rather than exhaustive) retrieval, results were screened in order of relevance, to a maximum of the first 200 results per query combination, and screening was stopped early within that limit if no additional eligible studies were identified across 20 consecutive results consistent with recommended practice for using Google Scholar as a supplementary source in systematic reviews. All Google Scholar searches were documented, including search terms, screening date, and number of records screened, in Annex 1A. Searches across all four databases were conducted on 2-3 March 2026 and covered publications dated January 2000 to December 2023.

Study selection and data extraction

Title and abstract screening were conducted independently by two reviewers, with disagreements resolved through discussion and, where necessary, arbitration by a third reviewer. Full-text assessment was applied to all potentially eligible studies. Data were extracted using a standardized extraction form capturing study design, geographic context, study population, type of health aid examined, corruption or fiduciary risk modality, governance intervention(s) assessed, accountability outcomes reported, and key findings. A narrative synthesis approach was adopted, given the substantial heterogeneity in study designs, interventions, and outcomes, which precluded formal meta-analytic pooling [6]. Inter-reviewer agreement was assessed using Cohen's kappa statistic. For title and abstract screening, substantial agreement was observed (κ= 0.78; 95% CI: 0.71-0.85), with all disagreements resolved through discussion. Extracted data were cross-checked for consistency and accuracy, and discrepancies were resolved through consensus discussion.

Given the heterogeneity of study designs and outcome measures, a narrative synthesis approach was adopted following established guidance for complex intervention reviews. Studies were first grouped by fiduciary risk modality, and second by governance intervention category. Thematic categories within each grouping were derived inductively from recurring patterns identified during data extraction and were cross-checked independently by two reviewers to confirm consistency of categorization. Within each thematic group, findings were compared across study design, geographic context, and MMAT quality rating to identify convergent and divergent patterns. For instance, where studies reported conflicting findings, divergent effects of performance-based financing on reporting integrity were retained and discussed explicitly rather than resolved by simple majority count or omitted; greater interpretive weight was given to findings from studies rated high quality on the MMAT.

Quality assessment

Methodological quality was appraised using the Mixed Methods Appraisal Tool (MMAT) [7], which enables the assessment of qualitative, quantitative, and mixed-methods studies within a unified framework. Given the substantial heterogeneity in study designs, intervention types, and outcome measures across the included literature, formal quantitative aggregation of MMAT scores was not performed. Instead, the appraisal was used to inform the interpretation of findings within the narrative synthesis. Each study was assessed against the five MMAT criteria relevant to its methodological design. Studies demonstrating greater methodological clarity, internal consistency, and analytical transparency were accorded greater interpretive weight in the synthesis, while studies with identifiable limitations in study design, data completeness, or analytical rigour were interpreted with appropriate caution. No studies were excluded solely based on methodological quality, consistent with established guidance for narrative synthesis in complex policy and governance research. Publication bias was assessed narratively, with particular attention to the geographic concentration of studies in sub-Saharan Africa and the predominance of Global Fund- and PEPFAR-related programmes, which may reflect underlying patterns in funding flows and research prioritization.

The Mixed Methods Appraisal Tool (MMAT) was applied to the primary empirical studies (quantitative, qualitative, and mixed methods designs) among the 67 included studies. The MMAT is not designed for appraising systematic reviews or policy analyses; the included policy analyses and systematic reviews were therefore assessed separately using a modified checklist addressing transparency of sourcing, clarity of the underlying evidence base, and methodological explicitness, adapted from AMSTAR-2 principles where applicable. These sources were used to contextualize and cross-check findings from primary studies within the narrative synthesis, rather than being pooled with or weighted equivalently to primary empirical evidence.

Studies rated high quality were accorded greater interpretive weight, particularly in instances of conflicting findings across studies. Findings supported only by lower-quality studies were interpreted cautiously and, where possible, flagged explicitly as requiring corroboration from further research. Exclusion reasons at full-text stage included: no governance or accountability dimension (n = 74); non-health ODA focus (n = 51); insufficient methodological transparency for quality appraisal (n = 38); excluded population (humanitarian emergency response only, no governance component) (n = 27) (Figure 1).

Results     

Characteristics of included studies

The 67 studies included in the final synthesis comprised 24 quantitative studies (35.8%), 21 qualitative studies (31.3%), 14 mixed-methods studies (20.9%), and 8 policy or systematic review articles (11.9%). Geographically, most empirical studies were conducted in Sub-Saharan Africa (n = 41; 61.2%), followed by South and Southeast Asia (n = 14; 20.9%), Latin America and the Caribbean (n = 7; 10.4%), and multi-regional studies (n = 5; 7.5%). The Global Fund and PEPFAR programmes were the most frequently examined aid mechanisms, appearing in 31 and 19 studies, respectively. The geographic distribution of included studies is presented in Figure 2. As the figure illustrates, the evidence base is heavily concentrated in Sub-Saharan Africa, reflecting both the geographic focus of major global health financing mechanisms and a likely publication bias toward anglophone and donor-funded research environments. Supplementary Annex 1B provides detailed characteristics of all included studies, including study design, geographic context, aid mechanism, fiduciary risk modality, governance intervention, key findings, and quality assessment results. Annex 1C provides the relevance scores and rationale for including the 67 studies.

Implications of geographic distribution

The observed concentration of studies in Sub-Saharan Africa has important implications for the transferability of findings. While the region represents the primary locus of large-scale health aid programmes, institutional, political, and administrative contexts differ substantially across regions. Governance structures, procurement systems, and donor engagement modalities in South and Southeast Asia and Latin America may differ in ways that affect both the manifestation of fiduciary risk and the effectiveness of accountability interventions. Consequently, caution is warranted in generalizing findings across regions. The relative underrepresentation of studies from Latin America and parts of Asia highlights a significant gap in the evidence base and underscores the need for more geographically diverse research to strengthen external validity. Regional classifications follow World Bank income group and regional designations. The "multi-regional" category comprises studies drawing on cross-national datasets spanning two or more regions. Programme attribution reflects the primary aid mechanism examined in each study; studies addressing multiple programmes were assigned to the most prominently featured (Figure 2).

Typology of fiduciary risk in health aid

The evidence supports a multidimensional taxonomy of fiduciary risk. The most frequently documented corruption modalities were procurement fraud and supply chain diversion (identified in 43 studies), which encompassed bid-rigging, collusion between procurement officers and suppliers, invoice inflation, and the systematic diversion of pharmaceutical commodities into parallel markets [8]. Ghost employee schemes and payroll fraud within health worker systems were documented across 19 studies, with particularly severe manifestations in fragile and post-conflict states where civil service management systems lacked basic data integrity controls [8,9]. Table 1 summarizes the principal fiduciary risk modalities identified in the reviewed literature, including their frequency, primary contexts, and assessed risk levels.

Budget misclassification, defined as the deliberate re-categorization or reclassification of expenditures to obscure the true utilization of health aid funds, was identified in 14 studies (20.9%) and represents a technically sophisticated form of fiduciary breach that is often difficult to detect through conventional audit mechanisms [10]. Its manifestations are particularly evident in donor-financed public financial management systems where expenditure lines are reallocated across budget heads to comply formally with reporting requirements while materially altering the intended use of funds. Concrete evidence of this modality has been documented in multiple Global Fund Office of the Inspector General (OIG) investigations between 2011 and 2015, where audit findings revealed cases of mis-posted expenditures, reclassification of ineligible costs under compliant budget lines, and retroactive adjustment of financial records to conceal non-compliant spending patterns. These cases illustrate that budget misclassification is not merely a theoretical category but a recurring operational strategy within weakly supervised financial systems.

Petty bribery and informal payments at the point of care, while extensively documented in the broader health systems literature [11], were identified in only 12 of the included studies (17.9%) in the specific context of aid-funded service delivery. This suggests a persistent analytical fragmentation between corruption studies focused on domestic health systems and those examining externally financed health programmes, with the latter under-examining micro-level corruption dynamics despite their relevance to service delivery outcomes.

Elite capture of aid resources, documented in 11 studies (16.4%), emerged as a structurally significant modality characterized by the appropriation of health aid benefits by politically connected actors, senior bureaucrats, or local power elites. Reported mechanisms include diversion of donor-funded supplies to private facilities owned by politically exposed persons, manipulation of beneficiary selection criteria in targeted health programmes, and preferential allocation of donor-funded infrastructure projects to politically influential districts. Evidence from sub-Saharan Africa indicates that elite capture frequently operates through formal administrative channels rather than overt illegality, making detection difficult within standard audit frameworks.

Several studies highlighted the structural role of parallel implementation architectures, particularly Project Implementation Units (PIUs), in generating fiduciary vulnerabilities while simultaneously weakening national Public Financial Management (PFM) systems [12]. This mechanism operates through three interrelated pathways: first, duplication of procurement systems whereby PIUs operate independent tendering processes outside national procurement authorities; second, employment distortions arising from non-competitive PIU staffing structures that often offer salary differentials significantly above civil service scales, thereby incentivizing skilled personnel migration from public institutions; and third, fragmented financial control systems characterized by the use of multiple unintegrated accounting platforms that increase the risk of unaudited cash advances and expenditure opacity. Collectively, these mechanisms produce a governance paradox in which donor-designed risk mitigation strategies generate parallel fiduciary channels that are themselves weakly regulated.

Figure 3 summarizes the frequency of the principal fiduciary risk modalities identified across the included studies and classifies each modality according to its assessed level of risk. Procurement fraud and supply chain diversion emerged as the dominant fiduciary risk, appearing in 43 of the 67 included studies (64.2%), indicating that procurement-related vulnerabilities remain the most pervasive governance challenge in donor-funded health programmes. Ghost employee and payroll fraud schemes were identified in 19 studies (28.4%) and were particularly concentrated in fragile and post-conflict settings where payroll verification systems were weak or absent. Budget misclassification, elite capture, and informal payments were documented less frequently but represented important recurrent patterns of fiduciary failure across multiple health financing contexts.

The comparatively lower frequency of phantom project activities (9 studies, 13.4%) and drug theft and diversion (8 studies, 11.9%) should not be interpreted as evidence of limited importance. Rather, these forms of corruption are inherently more difficult to detect through conventional audit and financial reporting mechanisms and may therefore be underrepresented in the published literature. Overall, the distribution illustrated in Figure 3 demonstrates that procurement-related corruption consistently dominates the evidence base, while several other fiduciary risks remain less visible despite potentially substantial effects on programme performance and accountability (Figure 3).

Governance interventions and accountability outcomes

The review identified four principal categories of governance intervention: (i) financial management systems and digital technologies; (ii) independent oversight and audit institutions; (iii) multilateral conditionality and performance-based funding; and (iv) social accountability mechanisms.

Financial Management Information Systems (FMIS), Integrated Financial Management Information Systems (IFMIS), and digital payment platforms were consistently associated with improvements in financial traceability, expenditure control, and audit trail integrity across the reviewed studies [13]. Evidence from Rwanda, Tanzania, Ghana, and Kenya, where IFMIS deployment has been particularly extensively documented, indicates measurable improvements in budget execution reporting, real-time expenditure tracking, and reductions in unverified transactions following system implementation. These country examples are illustrative rather than exhaustive, reflecting variation in implementation maturity across contexts. However, multiple studies emphasized that the effectiveness of FMIS/IFMIS platforms is strongly contingent on enabling system conditions, including sustained human resource capacity for system operation, reliable electricity and ICT infrastructure, and consistent political commitment to transparency reforms [14]. In contexts where these conditions are weak or unevenly distributed, digital financial management systems often function as reporting tools without substantially altering underlying fiduciary risk structures.

Supreme Audit Institutions (SAIs) and independent fiscal oversight bodies were examined in 22 studies spanning the period 2000-2023, with more recent evidence strengthening earlier findings on institutional constraint. The literature consistently indicates that audit effectiveness is shaped by three core factors: institutional independence from the executive, adequacy of technical and financial resources, and the degree of enforcement follow-through by legislative and judicial actors [15,16]. The evidence also indicates persistent implementation gaps in which audit findings related to health-sector and donor-funded programme irregularities are not consistently acted upon, limiting deterrent effects [15,16]. Related work on the budget process also highlights the interaction of formal and informal institutional practices in accountability [17].

Performance-Based Financing (PBF) frameworks, including those implemented by the Global Fund and World Bank Results-Based Financing programmes, were evaluated in 18 studies. Across multiple settings, PBF interventions were associated with improvements in selected service delivery and financial reporting indicators, including reductions in drug stockouts, improved reporting completeness, and enhanced facility-level accountability documentation [18]. However, a consistent finding across studies is the emergence of unintended behavioural responses, particularly data manipulation, indicator inflation, and strategic reporting practices in environments where verification systems are weak or externally imposed rather than institutionally embedded [19-21]. This suggests that PBF may, in some contexts, reconfigure rather than eliminate corruption risks, shifting them from procurement and input-level fraud toward output and reporting distortions.

Social accountability mechanisms, including community scorecards, citizen report cards, participatory expenditure tracking, and civil society monitoring initiatives, represent the most rapidly expanding governance intervention category in the post-2010 literature. Evidence from Mozambique, Uganda, Tanzania, and Sierra Leone indicates that these mechanisms can improve service responsiveness, enhance citizen reporting of corruption, and strengthen local-level transparency in health service delivery [22-24]. However, effect sizes are generally modest and context-sensitive, with stronger impacts observed in settings where civil society organizations possess operational autonomy and where local power structures are less hierarchically entrenched. Several studies also highlight the risk of elite capture of participatory accountability processes, particularly in contexts characterized by strong local patronage networks, thereby limiting their effectiveness as standalone governance tools [25].

Cross-cutting synthesis of governance interventions

Across the four categories of governance interventions, a consistent pattern emerges: no single intervention operates as a standalone solution to fiduciary risk in global health aid. Instead, the evidence suggests a layered and interacting governance architecture, in which the effectiveness of each intervention is contingent upon the presence and functionality of others [26,27]. Financial Management Information Systems (FMIS/IFMIS) primarily strengthen ex-ante and real-time financial traceability, expenditure control, and audit trail integrity; however, their effectiveness is dependent on enforcement within audit and oversight institutions. Where Supreme Audit Institutions (SAIs) are weak, under-resourced, or politically constrained, improved financial data does not automatically translate into enforcement action, thereby limiting FMIS impacts largely to enhanced reporting functionality rather than behavioral deterrence.

Conversely, audit and oversight institutions demonstrate stronger potential for ex-post accountability enforcement, but only when reliable financial data systems exist to generate complete and auditable records [26]. In the absence of robust FMIS or equivalent integrated digital financial infrastructure, audit institutions often operate reactively, constrained by fragmented records, weak expenditure visibility, and limited forensic traceability of donor-funded transactions [26]. Performance-based financing (PBF) systems introduce a distinct accountability logic centered on output verification, incentives, and results orientation. However, the evidence indicates that PBF effectiveness is highly dependent on (i) the integrity and completeness of underlying financial reporting systems and (ii) the presence of independent verification and audit mechanisms capable of validating reported outputs. Without these safeguards, PBF frameworks risk shifting corruption rather than eliminating it, moving fiduciary risk from procurement and input-level diversion toward data manipulation, indicator inflation, and strategic reporting behaviors.

Social accountability mechanisms operate primarily through citizen-mediated oversight, participatory monitoring, and bottom-up accountability pressure, thereby introducing an external corrective layer to formal institutional weaknesses [26]. However, their effectiveness is strongly moderated by the responsiveness and absorptive capacity of formal governance institutions. Where audit institutions, administrative systems, and political actors do not act upon citizen-generated accountability signals, social accountability mechanisms tend to produce transparency and awareness gains without sustained systemic correction [26].

Taken together, the evidence supports a complementarity hypothesis rather than a substitution model of governance reform. FMIS strengthens information integrity; audit institutions enforce compliance; PBF systems incentivize performance; and social accountability mechanisms introduce external transparency pressure from the demand side [27]. However, these mechanisms are mutually reinforcing only when embedded within a minimum threshold of institutional capacity, rule-of-law stability, and political commitment to accountability reform. In contexts where this institutional threshold is absent, governance interventions tend to operate in isolation, producing fragmented accountability systems that improve procedural compliance and reporting completeness without materially reducing underlying fiduciary risk [27]. This finding reinforces the broader conclusion that effective fiduciary governance in global health aid requires a multi-layered, system-integrated accountability architecture, rather than discrete reform instruments implemented independently of institutional context [27].

Contextual moderators of governance effectiveness

A cross-cutting finding of the review is that no governance intervention operates with uniform effectiveness across contexts. The evidence identifies four principal moderators of accountability outcomes: political will and executive commitment to anti-corruption norms; the pre-existing institutional baseline of the recipient country's PFM system; the degree of donor coordination and harmonization; and the integrity of the broader rule-of-law environment. Studies consistently demonstrated that governance interventions achieved greatest impact in contexts characterized by moderate institutional quality sufficient to operationalize new systems combined with genuine political incentives for accountability [27].

Figure 4 provides a conceptual synthesis of these four structural moderators and their influence on governance intervention effectiveness. Political will and executive commitment to anti-corruption norms emerged as the most frequently identified moderating factor, appearing in 31 studies. This finding reinforces the broader governance literature, which consistently demonstrates that technically robust institutional reforms have limited effectiveness in the absence of sustained political commitment to accountability.

The pre-existing Public Financial Management (PFM) institutional baseline represents the second major moderating factor. Across the reviewed studies, this baseline served as a practical capacity ceiling, determining the extent to which governance interventions could improve accountability outcomes. Across the reviewed studies, Financial Management Information Systems (FMIS), strengthened audit and oversight institutions, and social accountability mechanisms consistently produced stronger accountability outcomes in countries with stronger institutional capacity than in highly fragile settings where foundational governance systems remained weak.

The review further identified donor coordination and the broader rule-of-law environment as critical contextual conditions shaping governance effectiveness. Donor harmonization reduced duplication of reporting requirements and strengthened fiduciary oversight, whereas fragmented donor arrangements increased administrative burdens and weakened national accountability systems. Likewise, strong legal institutions and effective enforcement mechanisms enhanced the implementation of governance reforms, while weak judicial environments constrained accountability even when technically sound interventions were introduced. Collectively, these findings demonstrate that governance interventions operate interactively rather than independently, with deficiencies in any one structural condition capable of limiting overall effectiveness (Figure 4)

The donor coordination architecture emerged as a particularly significant structural determinant of governance effectiveness in global health aid. In settings characterized by high donor fragmentation measured by the number of disbursing entities, the degree of off-budget financing, and the proliferation of non-harmonized reporting requirements, recipient country Public Financial Management (PFM) systems face substantial administrative burdens that systematically weaken fiduciary controls and increase transaction complexity. This fragmentation contributes to parallel reporting structures, duplication of audit requirements, and reduced coherence in national budgeting and expenditure oversight processes [28,29]. These findings are consistent with broader literature emphasizing aid accountability and the political and institutional conditions under which donor-recipient relationships and governance reforms operate. The evidence base for the review comprised 67 included studies, whose detailed characteristics are presented in Supplementary Annex 1B and whose relevance assessments are provided in Annex 1C.

Global policy commitments to improve donor coordination and alignment have evolved through successive international frameworks. The Paris Declaration on Aid Effectiveness (2005) and the Accra Agenda for Action (2008) established early principles of harmonization, ownership, and alignment with national systems. These were subsequently reinforced by the Busan Partnership for Effective Development Cooperation (2011), which marked a shift toward broader development effectiveness and inclusive partnership structures, and further complemented by the Addis Ababa Action Agenda (2015), which emphasized financing coherence, domestic resource mobilization, and strengthened accountability in development finance flows. Despite these successive frameworks, empirical studies consistently document persistent deviations from aid effectiveness principles within the health sector, particularly in relation to off-budget financing and parallel implementation structures [29]. Taken together, the evidence indicates that while global aid effectiveness frameworks have progressively strengthened normative commitments to coordination, their implementation in global health aid remains uneven. The persistence of fragmented donor practices continues to impose structural inefficiencies on recipient PFM systems, thereby weakening fiduciary oversight capacity and reinforcing systemic vulnerability to misallocation and weak accountability enforcement.

Discussion     

Principal findings in context

This systematic review synthesizes a substantial body of evidence on fiduciary risk, governance interventions, and accountability outcomes in global health aid. Three principal conclusions emerge from the synthesis. First, fiduciary risk in health aid is structurally produced rather than merely incidental, arising from the interaction of institutional fragility in recipient systems, perverse incentive structures created by donor conditionality, and organizational vulnerabilities inherent in complex multi-principal aid delivery chains. This structural interpretation challenges dominant donor narratives that locate corruption primarily within recipient-country institutional failure and instead highlights the need for a more reflexive analysis of donor-side contributions to fiduciary vulnerability [30,31].

Compared with existing reviews of health sector corruption and aid governance, such as Vian (2008) and subsequent thematic syntheses focused on either corruption typologies or sector-specific interventions, this review extends the literature by explicitly integrating fiduciary risk, governance interventions, and accountability outcomes within a single analytical framework applied to official development assistance (ODA)-financed health aid systems. While previous reviews have examined these domains separately, few have systematically linked corruption modalities to governance responses and their observed accountability effects across aid delivery chains in a single synthesis. On this basis, the present review contributes an integrated, cross-domain perspective rather than claiming exclusivity as the "most comprehensive" evidence base.

Second, the evidence on governance interventions, while substantial in volume, remains limited in causal inferential rigour. The predominance of observational, cross-sectional, and case study designs, combined with significant heterogeneity in intervention specification and outcome measurement, constrains the strength of conclusions that can be drawn regarding the effectiveness of specific accountability mechanisms. This limitation is particularly important given the scale of investment in governance reform programmes in global health aid, meaning that large financial commitments are often informed by evidence that would not meet the methodological standards typically expected in experimental or quasi-experimental evaluation contexts.

Third, consistent evidence of context dependence in governance intervention effectiveness underscores the limitations of blueprint approaches to accountability reform. The scaling of social accountability mechanisms, performance-based financing (PBF) frameworks, and financial management information systems (FMIS/IFMIS) based on findings from specific contexts without sufficient consideration of institutional transferability has contributed to a pattern of reform implementation failures. These failures are frequently attributed in the literature to weak implementation fidelity, but the evidence reviewed here suggests that they also reflect more fundamental constraints related to institutional readiness and political economy conditions [32].

Theoretical implications

The findings of this paper speak directly to ongoing theoretical debates in global health governance, development economics, and anti-corruption scholarship. From a principal-agent perspective [33,34], the review documents how health aid systems generate multiple nested principal-agent relationships between donors and recipient governments, between governments and implementing agencies, and between agencies and service providers, each of which creates information asymmetry and moral hazard opportunities that can be exploited for corrupt gain. The proliferation of PIUs, while representing a donor attempt to compress this chain, effectively creates additional agency layers without resolving the underlying information and incentive problems.

Figure 5 illustrates the principal-agent structure underlying global health aid delivery across major multilateral and bilateral financing mechanisms, including the Global Fund, PEPFAR, Gavi, and World Bank Results-Based Financing programmes. The framework depicts the sequential relationships between donors, recipient governments, implementing agencies, and frontline service providers, while recognizing that many programmes maintain direct oversight arrangements that bypass national government systems. The figure also identifies the predominant fiduciary risks occurring at each interface, illustrating how corruption opportunities arise across multiple stages of aid implementation rather than at a single point within the delivery chain.

The framework further highlights the governance paradox associated with Parallel Implementation Units (PIUs). Although PIUs are commonly established to strengthen fiduciary oversight and reduce implementation risk, the reviewed evidence indicates that they may simultaneously weaken national institutional capacity by creating parallel financial management, procurement, and reporting systems outside government structures. This conceptual synthesis supports the principal-agent interpretation advanced by Eisenhardt [33] and Pratt and Zeckhauser [34], demonstrating that additional oversight layers do not necessarily eliminate information asymmetries but may instead create new accountability challenges within complex aid systems (Figure 5).

Social accountability scholarship has increasingly moved toward a political economy framing that situates community monitoring within broader power structures rather than treating citizen voice as a technocratic mechanism. The review supports this political economy turn, finding that the effectiveness of social accountability mechanisms is fundamentally conditioned by the political settlement within which they operate, particularly the degree to which elites have incentives to tolerate or respond to accountability pressures. This finding is theoretically coherent with the adaptive governance literature [35], which argues that governance reform must engage with the underlying political economy rather than seeking to circumvent it through technical interventions.

Policy implications

The evidence synthesized in this review carries several significant policy implications for donors, recipient governments, and multilateral governance bodies. However, the feasibility and sequencing of these implications vary substantially across institutional contexts, particularly in fragile and low-capacity settings. For donors, the findings strongly support a gradual reorientation away from parallel fiduciary control systems toward strengthening national Public Financial Management (PFM) institutions, even where this involves accepting higher short-term fiduciary exposure as a trade-off for long-term system capacity development. The persistent paradox of Project Implementation Unit (PIU) proliferation, whereby donor-driven implementation structures bypass national systems while simultaneously weakening them, remains a central structural constraint to sustainable governance reform and warrants policy redesign rather than incremental adjustment [35].

For multilateral financing institutions such as the Global Fund and Gavi, the review highlights the critical importance of sequencing governance interventions [36-42] Performance-based financing (PBF) frameworks should not be implemented in the absence of a robust verification infrastructure. Priority investments should therefore focus on third-party verification systems, independent audit strengthening, and data integrity assurance mechanisms before large-scale PBF deployment. The evidence indicates that without these preconditions, PBF systems risk displacing rather than reducing corruption, shifting fiduciary risk from procurement-level fraud toward more sophisticated forms of data manipulation and performance reporting distortion [43-50]. For recipient governments, the findings underscore the importance of institutional safeguards, including Supreme Audit Institution (SAI) independence and strengthened parliamentary oversight capacity as foundational conditions for accountability in health aid governance. However, the effectiveness of social accountability mechanisms is highly contingent on political and civic space conditions. In many fragile and hybrid governance contexts, the operating space is constrained by regulatory, political, or security-related restrictions, limiting the feasibility of citizen-led accountability initiatives. This introduces an important policy tension: while social accountability is theoretically valuable, its operational viability is uneven and context-dependent, and in some settings may require donor-supported protection of civic engagement space before it can function effectively. Given this heterogeneity in feasibility, policy recommendations should be understood as prioritized and conditional rather than uniform. Interventions can be broadly categorized into three tiers: i) tier 1 (high feasibility, strong evidence strength): FMIS/IFMIS implementation with adequate capacity investment; strengthening of Supreme Audit Institutions; improvement of audit follow-up mechanisms [51-67]. ii) Tier 2 (conditional feasibility): performance-based financing systems requiring pre-existing verification and data integrity infrastructure; structured donor coordination reforms aligned with national PFM systems. iii) Tier 3 (context-dependent/low feasibility in fragile settings): social accountability mechanisms requiring enabling civic space; advanced participatory governance tools dependent on stable institutional environments [68-70]. This stratification reflects the empirical reality that governance interventions in global health aid are not uniformly transferable across contexts [71-82]. A summary prioritization matrix detailing intervention type, evidentiary strength, and minimum enabling conditions is therefore recommended for inclusion as a supplementary appendix to enhance policy usability and decision-making relevance.

Limitations

This review has several limitations. Although key methodological decisions were specified before full-text screening and applied consistently, the review protocol was not prospectively registered, which is acknowledged as a limitation of this review. The review was also limited by language restriction to English-language studies, which may introduce language bias and underrepresent evidence from non-Anglophone regions. Despite the inclusion of grey literature from major multilateral organizations, publication bias may persist due to the underreporting of negative or null findings in governance interventions [83-87].

Methodological heterogeneity across included studies prevented meta-analysis and limits the precision of comparative effectiveness conclusions, making findings primarily suitable for narrative synthesis. In addition, incomplete reporting of inter-rater reliability statistics and the supplementary nature of the data extraction tool may limit full reproducibility, although standard systematic review procedures were followed. The broad time span of included studies (2000-2023) introduces temporal heterogeneity, as global health aid governance has evolved significantly over this period, potentially limiting the direct applicability of earlier findings to contemporary systems. Finally, the evidence base is geographically concentrated in Sub-Saharan Africa, limiting generalizability to other regions such as South/Southeast Asia and Latin America, where aid structures and governance contexts may differ substantially [88-91].

Conclusion     

Fiduciary risk and corruption in global health aid are structurally produced governance challenges that cannot be addressed through isolated technical interventions. This systematic review synthesizes evidence on governance interventions targeting key fiduciary risks, including procurement fraud, payroll manipulation, drug diversion, and budget misclassification, which arise from institutional fragility, information asymmetry, and misaligned donor-recipient incentives [92-95]. Financial management systems, audit institutions, performance-based financing, and social accountability mechanisms make context-specific contributions to improved accountability, but the largely observational and heterogeneous evidence base limits the ability to draw strong causal inferences. Intervention effectiveness is consistently shaped by institutional capacity, political will, and broader rule-of-law conditions. Overall, the evidence supports a layered accountability model integrating digital financial controls, independent oversight, harmonized donor practices, and civil society engagement. However, its effectiveness depends on minimum institutional thresholds, and reliance on parallel donor systems may undermine domestic accountability capacity. Future research should prioritize stronger causal designs, including longitudinal and quasi-experimental studies, to improve the evidence base for governance reform in global health aid.

Additional evidence reinforces these patterns: studies of humanitarian and health-sector governance identify recurring corruption and fiduciary vulnerabilities [96,97], while research on aid contracts and donor transitions highlights the importance of institutional conditions and accountability arrangements in shaping governance outcomes [98,99]. Country-level audit evidence also documents weaknesses in public financial management and expenditure accountability [100].

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Grace Aiwonose Ibe: conceptualization, study design, literature search, data extraction, analysis, and primary manuscript drafting. Omoyebagbe Rosaline Dania: literature screening, data extraction, quality appraisal (MMAT), and contribution to manuscript drafting and revision. Kinjal Nayak: methodology development, data synthesis, interpretation of findings, and critical review of the manuscript. Larry James Baker: supervision, conceptual guidance, validation of findings, and critical revision of the manuscript for intellectual content.

Table and figures     

Table 1: typology of fiduciary risk modalities in global health aid

Figure 1: PRISMA 2020 flow diagram of study selection

Figure 2: geographic distribution of included studies by region (N = 67)

Figure 3: frequency of fiduciary risk modalities identified in the included studies

Figure 4: contextual moderators influencing governance intervention effectiveness

Figure 5: principal-agent framework for global health aid delivery and associated fiduciary risks

Annexes     

Annex 1: supplementary materials

References