利用卫生设施死亡数据优化G-DRG支付设计
Opportunities, risks and policy implications of using health facility mortality data to inform DRG payment design under Ghana’s NHIS
该研究探讨了在低收入国家(LMICs)利用卫生设施死亡数据优化诊断相关组(DRG)支付设计的机会与挑战。G-DRG系统在加纳的实施面临数据不足和分类不精确的问题,导致支付未能充分反映临床风险。通过分析死亡数据,可以识别DRG分类中的不足,优化支付机制,促进基于价值的购买模式。
Provider payment reforms remain central to the pursuit of universal health coverage (UHC), particularly in low- and middle-income countries (LMICs) that face the dual challenge of rising health needs and constrained fiscal space1. Payment mechanisms shape provider behavior by influencing the volume, type, and quality of services delivered, as well as the overall efficiency and sustainability of financing systems2.
Over the past two decades, countries have moved away from passive, input-based reimbursement models toward strategic purchasing approaches that link financing to service outputs, efficiency, and quality. In high-income settings, payment innovations have been used to align provider incentives with value rather than volume3. Low- and middle-income countries (LMICs) are increasingly following this trajectory- adapting global lessons to local contexts marked by limited fiscal space, fragmented financing, and increasing epidemiological complexity1. The World Health Organization and the World Bank have both emphasized that reforms to provider payment systems are critical for improving allocative and technical efficiency, ensuring accountability for results, and containing costs without compromising access or equity4.
In high-income countries, DRG systems typically incorporate explicit measures of severity of illness and risk of mortality through secondary diagnoses, comorbidity indices, and patient-level cost accounting. Systems such as All Patient Refined DRGs (APR-DRGs) or severity-adjusted DRG variants rely on detailed clinical documentation, standardized ICD coding practices, and validated comorbidity measures, with tariffs calibrated using routinely updated hospital cost accounting systems5. By contrast, Ghana’s G-DRG system operates in a context of limited diagnostic granularity, incomplete coding of comorbidities, insufficient linkage between clinical outcomes and payment design, and constrained availability of patient-level cost data. As documented in other low- and middle-income settings, DRG systems under these conditions tend to rely on simplified case groupings and historically negotiated tariffs, limiting their ability to adequately capture clinical severity and outcome risk6,7.
In the context of such approaches to DRG calibration and nascent facility-level mortality data systems, mortality data can serve as a diagnostic signal to identify poorly specified DRG groupings, disaggregate severity gradients, and identify systematic mismatches between reimbursement levels and clinical risk. Importantly, we do not propose that observed mortality should directly determine reimbursement; rather, mortality can be used to identify DRGs in which existing classifications inadequately capture differences in clinical severity and expected resource use. Similar uses of outcome indicators as signals for payment system refinement have been documented in early DRG reforms in middle-income countries8,9.This distinction is therefore critical for Ghana and for other countries seeking to strengthen strategic purchasing while their health information and costing systems continue to mature. This paper explores the opportunity for leveraging the principles of DRG cost calibration utilized in high income contexts but adapting these to low-income contexts that may not have all the requisite data frameworks for such precise costing using outcomes data.Applying these improvements in the development of provider payment mechanisms reflects a broader shift toward value-based purchasing, in which routine health data such as mortality, morbidity, and utilization statistics can be used to optimize tariffs, refine case mix classifications, and strengthen accountability for outcomes6.
The availability of improved health information systems and facility-level reporting therefore has expanded opportunities to refine payments not only on inputs and activities but also on the quality and results of care delivered. In Ghana, the Diagnosis-Related Group (G-DRG) payment system was introduced in 2008 and represented a landmark effort to move away from open-ended fee-for-service reimbursement model under the National Health Insurance Scheme (NHIS) and towards a more structured approach that could enhance transparency, predictability, and cost containment10. The G-DRG system covers inpatient and bundled outpatient services, while pharmaceuticals remain reimbursed separately under a fee-for-service approach11,12. The NHIS tariff manual, which is periodically updated, sets out reimbursement rates for over 550 clinical groupings, ranging from obstetric care to chronic disease management13.
Designing a robust DRG system requires attention to several interlinked factors—including accurate clinical classification, reliable cost and utilization data, and appropriate adjustment for case severity and comorbidities. In the absence of such refinements, DRGs may mask meaningful differences in patient complexity and inadvertently create incentives for under-provision of health services or upcoding14. Patterns of in-facility deaths and case fatality rates across DRG categories can help identify heterogeneous or poorly specified groups, flag quality differentials, and identify areas where existing classifications may inadequately capture clinical risk15. Using mortality data to inform DRG refinement could therefore support more accurate case-mix classification while also strengthening accountability for outcomes and shifting to a stronger value-based purchasing model7. Ghana’s G-DRG system has, however, faced challenges and tariffs have sometimes failed to reflect the true cost of treating severe or complex conditions, with criticisms for insufficient alignment with disease burden or quality outcomes6.
At the same time, Ghana’s health information system is evolving. The introduction of electronic claims (e-claims), improvements in hospital reporting, and mortality audits have expanded the availability of facility-based data, including cause-specific mortality. These advances are underpinned by the Ghana Health Information System Strategic Plan (GHISSP)16, which seeks to build an integrated, interoperable, and standards-based health information ecosystem. The GHISSP emphasizes strengthening data governance, interoperability, and linkages across routine health information systems, civil registration, and insurance databases to enhance the generation and use of quality data to drive smarter decision-making. While these datasets remain incomplete and uneven in quality, they present a largely untapped opportunity: using facility-level mortality information to refine DRG groupings and strengthen accountability for outcomes. Mortality-informed DRG configuration could improve the homogeneity of case groupings, better align financing with disease burdens, and support improvements in data quality. Yet the approach is not without risks. Weaknesses in mortality reporting, representativeness biases, and risks of gaming or manipulation could undermine the integrity of the payment system if not carefully managed17,18.
In this paper, we examine the potential of utilizing facility-based mortality data to optimize Ghana’s G-DRG payment system. We argue that while this approach holds significant promise for enhancing NHIS sustainability and accountability, its success depends on phased implementation, robust data governance, and deliberate efforts to mitigate equity risks.
The current landscape of Ghana’s G-DRG system
The G-DRG was introduced to address widespread dissatisfaction with the inefficiencies of fee-for-service reimbursement19. Under fee-for-service, providers had strong incentives to over-provide services, leading to escalating costs for the NHIS. The G-DRG was expected to reduce perverse incentives, promote cost control, and enable more predictable expenditure management10.
Under the National Health Insurance Scheme, the G-DRG payment mechanism applies to all NHIS-accredited public, private-for-profit, and faith-based providers delivering covered services20,21. Primary health centres, district hospitals, regional hospitals, and teaching hospitals are all reimbursed under the G-DRG system, although the complexity of cases treated varies substantially across levels of care12.
Tariffs are determined through periodic revisions led by the National Health Insurance Authority (NHIA), drawing on historical cost studies, claims utilization patterns, and stakeholder negotiations13. These revisions are constrained by fiscal pressures, limited availability of facility-level cost data, and broader macro-economic considerations. This is further compounded by weaknesses in clinical and intervention coding: incomplete and inconsistent use of ICD standards for diagnoses and causes of death, alongside limited adoption of the International Classification of Health Interventions (ICHI), reduces the system’s ability to distinguish case severity, resource intensity, and outcomes across DRG categories23. As a result, tariff calibration and claims verification rely on rough groupings that mask variation in clinical complexity and mortality risk, constraining effective DRG refinement7,24.
In principle, the DRG system was adopted in Ghana to reward efficiency: health facilities can deliver care at lower cost than the tariff and retain savings, while those that overspend bear the financial risk. However, several structural challenges have undermined the effectiveness of the G-DRG First, G-DRG tariffs have not consistently aligned with prevailing hospital charges and resource requirements, creating potential financing gaps for some services and case types25. This is particularly problematic for conditions such as obstetric emergencies, trauma, and advanced HIV/AIDS or TB co-infections, where resource use far exceeds the reimbursement rate19,22. Second, claims inflation has persisted, with providers accused of “upcoding” cases to higher-paying DRGs or exaggerating case severity22. Third, the G-DRG has been criticized for limited responsiveness to Ghana’s shifting disease burden, including the rise of non-communicable diseases alongside persistent infectious disease burdens6.
These limitations underscore the need for refinement. Facility mortality data could provide a new dimension to DRG configuration, offering insights into severity, outcomes, and resource needs that are currently absent from the country’s tariff-setting.
Opportunities for mortality-informed DRG refinement
The use of mortality data in optimizing the G-DRG system offers several potential benefits for Ghana’s NHIS.
Improving clinical homogeneity
A central objective of DRG systems is to group patients with similar clinical characteristics and expected resource use into homogeneous categories. In practice, however, DRG systems in many low- and middle-income countries aggregate clinically heterogeneous cases due to limited diagnostic detail, weak coding of comorbidities, and constrained cost data6,26. This heterogeneity undermines both cost accuracy and the ability of payment systems to reflect clinical risk.
Facility-level mortality data can serve as an informative signal of unobserved severity within DRG categories15,27. Persistently higher case fatality rates within specific DRGs may indicate that groupings mask substantial differences in disease severity, complications, or treatment intensity. In high-income settings, similar insights are generated through formal severity adjustment mechanisms embedded within DRG systems, which stratify cases by risk of mortality and complications28.
In Ghana, mortality patterns could be used to identify DRGs requiring further clinical stratification, with observable patient characteristics subsequently used to define severity subclasses and associated payment weights. Consistent with established approaches to DRG design and tariff setting, such analyses would be conducted across sufficiently large patient cohorts and facilities and used to inform periodic refinement of DRG classifications, rather than facility-specific payment adjustments. Comparable approaches have been used in early DRG reforms in middle-income countries, where outcome indicators informed iterative revisions to case groupings rather than direct tariff adjustments8,9.
Aligning payments with clinical severity and resource needs
Mortality-informed DRG refinements can also help identify where existing payments inadequately reflect differences in clinical severity and resource needs. In Ghana, conditions associated with high mortality such as severe malaria, neonatal sepsis, advanced cancers, and complicated obstetric cases17 are often under-reimbursed relative to their resource requirements19.
International experience shows that DRG systems that fail to account for severity and outcome risk tend to underpay hospitals treating sicker patients, leading to risk selection and inequitable resource allocation14. In several advanced economies, routine hospital data have been used to identify differences in severity and expected resource needs that inform more refined DRG classifications and payment weights14.
While Ghana’s data environment does not yet support formal severity weighting, triangulating mortality patterns with utilization and claims data can help prioritize DRGs for revision and ensure that tariffs better reflect the epidemiological transition toward more complex and chronic conditions.
Strengthening accountability for quality
Beyond cost calibration, mortality patterns within DRGs can serve as an accountability and quality monitoring tool. Unexpectedly high mortality rates within specific DRGs may signal gaps in clinical quality, referral pathways, or resource availability rather than differences in case mix alone29,30.
In high-income health systems, DRG-linked mortality indicators are routinely used for hospital performance benchmarking and quality improvement, even when not directly tied to payment24. Similar approaches have been piloted in middle-income settings, where outcome indicators complement claims audits to identify facilities or service lines requiring closer review8.
For Ghana, mortality can serve two distinct functions: informing DRG refinement by identifying inadequately specified case groups, and, when risk-adjusted, serving as an outcome indicator to identify potential quality concerns and support targeted quality-improvement efforts. Higher observed mortality should not result in higher reimbursement, and any introduction of risk-adjusted DRG payments should consequently be implemented alongside performance indicators that monitor quality of care and patient outcomes. This would help ensure that improved adjustment for clinical risk does not inadvertently weaken incentives for quality care, while supporting continuous quality-improvement measures.
Provider incentives and implementation risks
While facility-level mortality data offers potential value for refining DRG case-mix and improving severity calibration, their use in provider payment design introduces risks related to data quality, representativeness, and provider incentives. In settings where mortality data systems remain incomplete and uneven, these risks may distort reimbursement signals, exacerbate gaming, and undermine equity unless explicitly addressed through careful design and governance.
Recognizing these risks, mortality-informed DRG refinement in Ghana should prioritise phased implementation. We share a few anticipated challenges and risks with implementation.
Data quality risks to tariff calibration
Health facility mortality statistics in Ghana remain unreliable31. Many facilities lack adequately trained personnel to code causes of death, and misclassification is common due to poor clinical documentation and inconsistent application of ICD standards. Empirical evidence shows that facility death data in Ghana underreport certain causes while overreporting others, reflecting systemic weaknesses in cause-of-death certification and reporting. Without substantial improvements in data quality and completeness, incorporating mortality signals into DRG refinement risks introducing significant distortions into tariff calibration and provider reimbursement.
Representativeness bias
Facility-based deaths represent only a subset of overall mortality and are systematically skewed toward more severe cases and populations with access to health facilities32. In Ghana, rural populations and poorer households experience a substantial share of mortality outside formal health facilities, particularly for conditions sensitive to primary care access such as childhood malaria and maternal complications. Over-reliance on facility mortality data therefore risks over-weighting tertiary conditions while underrepresenting community-level disease burden. While modelling approaches—such as redistribution of ill-defined causes and calibration against community mortality data—can partially mitigate these biases, failure to account for representativeness limitations could misalign DRG refinements with true population health needs.
Perverse incentives and claims inflation
The use of mortality signals in DRG refinement also raises concerns about perverse incentives and strategic provider behaviour. If higher reimbursement is implicitly associated with higher observed mortality or severity, providers may have incentives to misreport diagnoses, exaggerate case severity, or selectively classify patients into higher-paying DRG categories. Directly linking reimbursement to observed deaths should therefore be avoided, as this could also create inappropriate incentives around clinical outcomes. Ghana’s NHIS has already experienced significant claims inflation under the existing G-DRG system, suggesting that mortality-informed DRG refinement, if translated into poorly designed payment adjustments, could intensify these behaviours unless accompanied by robust audit, verification, and accountability mechanisms22,26.
Institutional capacity constraints
Implementing mortality-informed DRG refinement requires advanced actuarial modelling, epidemiological expertise, and robust IT infrastructure. At present, capacity within the NHIA to undertake such analyses remains constrained, though there is scope to strengthen these functions and draw on lessons from past reforms, including the Ashanti Region capitation pilot. Without adequate technical preparation, institutional coordination, and political buy-in, mortality-informed DRG refinements risk replicating earlier implementation challenges rather than delivering meaningful improvements.
Equity risks from differential institutional capacity
Finally, equity risks arise from differential institutional capacity across facilities. Hospitals with stronger digital infrastructure and coding capacity, often located in urban areas, may be better positioned to document severity accurately and benefit from mortality-linked DRG refinements. In contrast, rural and under-resourced facilities may be disadvantaged not because of poorer clinical performance, but due to weaker data systems and limited coding capacity. If unaddressed, these dynamics could exacerbate existing inequities in NHIS financing and access to care, disproportionately affecting facilities serving vulnerable populations33.
Looking forward: a phased and cautious approach
Ghana’s G-DRG system has been a major step in the country’s health financing reform journey, but persistent challenges- including tariff misalignment, rising costs, and weak linkage to outcomes have constrained its effectiveness. Facility-based mortality data represent a promising but underutilized resource that could help refine DRG groupings, better align tariffs with disease burden, and strengthen accountability for quality of care. However, international experience shows that integrating outcome or severity signals into DRG systems requires careful, incremental implementation to avoid unintended consequences such as gaming, cost escalation, or provider resistance.
To maximize benefits while minimizing risks, Ghana should therefore pursue a phased, evidence-based approach to mortality-informed DRG reform. First, mortality-informed DRG refinement should be piloted in a limited set of high-priority clinical areas such as maternal, neonatal, and selected infectious disease conditions, where mortality data are relatively more complete, and policy relevance is high. Similar piloting approaches that were used in the country when the NHIS explored the implementation of capitation as a provider payment mechanism for the NHIS34, and in Thailand and Australia, where severity and outcome adjustments were initially introduced in selected DRG groups before broader system-wide adoption35,36.
Second, mortality signals derived from routine facility data should be systematically triangulated with external sources, including Health and Demographic Surveillance System (HDSS) sites (Navrongo, Kintampo, Dodowa), civil registration and vital statistics (CRVS), and modeled estimates from WHO and the Global Burden of Disease. Such triangulation reflects international best practice in contexts where facility deaths are not fully representative of population mortality32.
Third, mortality-informed payment reform must be linked to investments in data system strengthening, including ICD-11 training, improved clinical documentation, cause-of-death reporting, and routine mortality audits. Experience from countries with advanced DRG-based systems suggests that payment reforms are most effective when paired with sustained investments in coding quality, audit capacity, and clinical governance24. In Ghana, such linkages present an opportunity by which improved data quality can support better tariff calibration, while financing incentives reinforce investments in health information systems.
Fourth, robust safeguards against gaming and unintended provider behavior are essential. Independent coding audits, transparent risk-adjustment models, and alignment with the NHIS e-claims platform can strengthen verification and oversight. Evidence from DRG reforms in both high- and middle-income countries underscores that outcome-linked payments, if poorly designed, may increase incentives for upcoding or selective admission unless accompanied by strong audit and governance mechanisms8,37.
Finally, sustained stakeholder engagement is critical for political feasibility and system legitimacy. The NHIA, Ministry of Health, Ghana Health Service, healthcare providers, clinicians and professional associations, policy and health financing experts, academia, and patient and civil society representatives should be involved in the design, piloting, and evaluation of mortality-informed DRG refinement. Ghana’s experience with earlier payment reforms, including the capitation pilot, highlights how limited consultation and trust deficits can undermine technically sound reforms21,22.
For Ghana, the next phase of improvements should include continuous learning in order to adopt and adapt carefully, ensuring that mortality data are used as a strategic tool for system improvement. If implemented carefully, mortality-informed DRG refinements could strengthen NHIS sustainability, improve provider payment design, tariff calibration and case-mix adjustment, and better align reimbursement with clinical severity and resource needs, while also catalyzing broader improvements in health information systems, clinical documentation, and mortality surveillance. In this respect, Ghana’s experience demonstrates a practical pathway for other countries in sub-Saharan Africa and similar low- and middle-income settings seeking to align provider payment mechanisms with the realities of disease burden and health system performance.
来源:Nature 健康护理主题 · nature.com