Hip Fracture Management in Older Adults in Resource‐Constrained Settings: Determinants, Care Pathways, and Outcomes—A Systematic Review and Meta‐Analysis

ABSTRACT

Background and Aims

Hip fractures in older adults cause substantial mortality and functional decline, particularly where access to timely surgery is limited. This review synthesized the determinants, management pathways, and outcomes of hip fractures in resource‐constrained settings and evaluated the association between non‐operative management (NOM) and mortality.

Methods

PubMed, Scopus, and Web of Science were searched from January 1, 2015, to January 25, 2026. Eligible studies included older adults with hip fractures managed in resource‐constrained settings and reported management determinants, care pathways, or clinical outcomes. Study selection and reporting followed PRISMA 2020 and MOOSE recommendations. Mortality comparing NOM with operative management was pooled using a random‐effects model. Risk of bias and certainty of evidence were assessed using design‐appropriate tools and a GRADE‐informed approach.

Results

Fourteen studies were included. NOM prevalence ranged from approximately 10% to more than 30%. Selection of NOM and delayed surgery reflected patient factors, including advanced age, frailty, comorbidity, impaired premorbid mobility, and cognitive impairment, alongside system constraints such as out‐of‐pocket costs, limited theater capacity, workforce and implant shortages, and delayed referral. Three LMIC/UMIC cohort studies contributed to the mortality meta‐analysis. NOM was associated with higher mortality than operative management (RR 2.76, 95% CI 1.14–6.70), with substantial heterogeneity (I² = 84.3%). Certainty of the evidence was very low because of the observational designs, heterogeneity, and confounding by indication.

Conclusion

NOM and delayed hip fracture care reflect both clinical vulnerability and health‐system constraints. Although NOM was associated with higher mortality, the limited observational evidence, substantial heterogeneity, and confounding by indication preclude causal interpretation. Strengthening financing, referral pathways, surgical capacity, implant availability, and perioperative care may improve equitable access to timely treatment.

Abbreviations

CGAs

comprehensive geriatric assessments

CI

confidence interval

DALYs

disability‐adjusted life years

ED

emergency department

ERAS

enhanced recovery after surgery

HICs

high‐income countries

ICU

intensive care unit

LICs

low‐income countries

LMICs

low‐ and middle‐income countries

LOS

length of stay

MDT

multidisciplinary team

NOM

non‐operative management

OR

odds ratio

PRISMA

preferred reporting items for systematic reviews and meta‐analyzes

RCTs

randomized controlled trials

THA

total hip arthroplasty

WHO

World Health Organization

1. Introduction

Hip fractures are one of the most significant injuries among the elderly, as it has a significant impact on survival, functional independence, and utilization of health systems [1, 2]. The incidence of hip fracture is increasing worldwide due to ageing populations, with the burden of hip fractures being particularly high in low‐ and middle‐income countries (LMICs) [3, 4] due to the often limited capacity of their health systems to cope with the increasing burden of trauma and surgical diseases (Cooper 2011; Gullberg 1997). In spite of the development of operative techniques and perioperative care, hip fractures are linked to significant morbidity and mortality in the short and long term, even in highly‐resourced environments (Haentjens 2010; Schnell 2010).

In high‐income nations (HICs), early operative intervention (usually by 24–48 h) has become the emerging norm in most hip fractures with evidence‐based backing on the benefits of timely surgery in the reduction of complications and better outcomes [5, 6, 7]. Non‐operative management (NOM) is often restricted in this case to a small and highly selected group of patients, including those with a high level of frailty, or terminal illness or with clear palliative objectives of care (Wijnen 2021). These are typically decisions that are formulated in the context of shared decision‐making, and backed by the availability of multidisciplinary perioperative and geriatric services. In contrast, the treatment of hip fractures in resource‐limited settings, such as many low and middle‐income countries (LMICs) and some upper middle income countries (UMICs), is often carried out in settings in which the provision of surgical intervention in a timely fashion is not assured [8, 9, 10]. Constraints like time in the operating theater, shortage of implants, shortage of anesthesia and surgical personnel, delayed referral and transfer mechanisms, and high out‐of‐pocket payments are common causes of delay in seeking operative care or preclude it entirely [11, 12] (Meara 2015). Consequently, NOM and the prolonged delay of surgery could reveal structural barriers, as opposed to patient preference or a deliberate palliative approach (Dash 2015; Dela 2021). The global surgery view of hip fracture thus acts as a practical sentinel condition: the quality, affordability, and availability of prerequisites of surgery, and the quality of that surgery depend not just on the clinical risk of the patient but also on the quality and availability of the necessary surgery itself [13, 14].

Emerging evidence from resource‐constrained settings suggests that NOM and delayed surgery is associated with poor outcomes including high early and late mortality (Amrayev 2018; Paruk 2020; Tabu 2023). However, it is difficult to interpret because NOM is frequently confounded by indication: patients managed non‐operatively are often older, frailer and have a higher burden of comorbidities compared with those who are being treated by surgery. In addition, the meaning of “NOM” differs depending on the context—“NOM” might mean palliative management in line with patient goals or it might mean “default” non‐operative care due to a system failure—raising important clinical, ethical, and policy questions about equity and access to essential surgical care [15, 16, 17].

Determinants of hip fractures management in resource constrained settings are therefore multidimensional [18, 19]. Patient characteristics (age, frailty, comorbidity, cognitive status, premorbid mobility) interact with system factors (financing mechanisms, public vs. private pathways, theater and implant availability, referral delays) and provider‐level considerations (risk tolerance, perceived futility, availability of orthogeriatric or multidisciplinary support) to shape decisions and outcomes (Xiang 2023; Dela 2021). However, these determinants have not been synthesized with the outcomes comprehensively and in a manner that would be directly applicable to clinicians, healthcare planners and policymakers who would like to find modifiable leverage points.

Existing reviews on management of hip fracture have focused largely on high‐income settings or on particular operative techniques or on perioperative timing standards that assume stable surgical capacity [5, 6]. Reviews discussing NOM often set it in the context of it being a palliative choice in HICs, which may not be translatable into settings where NOM is often limited by affordability or supply side limitations. As a result, there remains a fragmented evidence base in relation to (i) management of hip fractures in resource‐constrained settings, (ii) reasons for NOM and delay, and (iii) associated outcomes.

Based on projected increases in older populations in LMICs and UMICs, the number of hip fracture‐related deaths and disability is likely to rise without specific strengthening of essential surgical systems (Cooper 2011). An integration of findings between outcomes and determinants is required to provide support to pragmatic policy decisions, including the financing of protection, redesign of referrals, strengthening theater and implant supply, and development of perioperative services, and to determine priorities on future research in environments where randomized trials are rare.

Thus, the purpose of this systematic review was to summarize the determinants and outcomes of hip fracture management strategies in older adults in resource limited settings. In particular, we aimed to (i) identify patient‐, provider‐, and system‐level determinants influencing selection of operative versus NOM and surgical timing; (ii) describe outcomes associated with different management strategies; and (iii) quantitatively estimate the association between NOM and mortality where data permitted. This review seeks to connect clinical outcomes with health‐system variables to guide evidence‐based practice and health‐system planning for equitable hip fracture care in resource‐limited settings.

2. Materials and Methods

2.1. Reporting Framework

This systematic review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyzes (PRISMA) 2020 statement and, because the synthesis primarily included observational evidence, the Meta‐analysis of Observational Studies in Epidemiology (MOOSE) guidelines [20, 21]. A protocol was developed a priority, which defined the review question, eligibility criteria, outcomes and synthesis methods.

2.2. Review Question and Objectives

The review addressed the following question: Among the elderly with hip fractures in resource constrained settings, what determinants and outcomes are associated with different management strategies (operative management, NOM, and care pathways including surgical delay)?

Objectives were to:

• 1.Identify patient‐, provider‐, and system‐level determinants associated with selection of management strategies, particularly NOM and delayed surgery.
• 2.Describe clinical outcomes across management strategies (primary: mortality; secondary: complications, function, and process outcomes).
• 3.Quantitatively synthesize differences in mortality between NOM and operative management where data were sufficiently comparable.

2.3. Definitions and Conceptual Framework

Resource‐constrained settings were defined as those studies that took place in low‐ and middle‐income countries (LMICs) and upper‐middle‐income countries (UMICs) (World Bank classification) and/or settings that were specifically mentioned as having constrained surgical capacity (e.g., limited theater capacity, implant shortage, significant financial constraint).

Management strategies were defined as:

• Operative management: any definitive surgical intervention for hip fracture (e.g., arthroplasty, intramedullary nailing, dynamic hip screw).

• NOM: conservative management without definitive surgery, including palliative NOM where specified.

• Care pathway/timing exposures: delays to admission or surgery, categorized using author‐defined thresholds (e.g., > 48 h) or continuous time‐to‐surgery metrics.

Determinants were classified into patient factors (age, frailty, comorbidity, premorbid mobility), fracture factors (type/displacement), system factors (financing, sector, theater/implant availability, referral pathways), and provider factors (decision drivers).

2.4. Eligibility Criteria

2.4.1. Eligibility Criteria Were Defined Using a PECO/PICO‐Aligned Framework

2.4.1.1. Population

Older adults with hip fractures as defined by study authors, with older adult operationalized as ≥ 50 years (or study‐defined thresholds, typically 50–60 years), including fragility hip fractures.

2.4.1.2. Exposure/Intervention and Comparators

The qualified studies evaluated at least one of the following criteria:

• NOM versus operative management.

• Determinants associated with selection of NOM, operative management, or surgical delay.

• Outcomes associated with operative strategy or care pathway factors (e.g., early vs delayed surgery).

2.4.1.3. Outcomes

Primary outcome: all‐cause mortality, which was documented at the timepoints that were reported in each study (e.g., 30‐day, 90‐day, 1‐year), and specifically mortality between NOM and operative management where pooling of data was possible.

Secondary outcomes: complications (e.g. pneumonia, pressure ulcers, thromboembolism), length of stay, destination of discharge, readmissions, mobility/functional outcomes, and healthcare process outcomes (e.g. time to surgery).

2.4.1.4. Study Designs

The categories of designs that were eligible were observational cohort studies (prospective or retrospective), extractable effect estimate of a registry analysis, cross‐sectional studies that provide published determinants, and clinician survey determinants that are pertinent to selection of management strategies.

2.5. Exclusion Criteria

Studies were excluded if they were: review articles/editorials/letters with no primary data, protocols with no outcomes, not focusing on hip fractures or older adults, or not conducted in low resource settings. Where high‐income evidence was retained, this was considered to be the contextual comparator evidence, which was processed separately from the primary pooled model.

2.5.1. Information Sources and Search Strategy

The study team searched Scopus, PubMed and Web of Science from from 1st Januray 2015 to 25th January 2026. using predefined search strings involving terms pertaining to hip fracture, NOM/conservative management, operative treatment, delays/timing, determinants, outcomes, and resource‐constrained/LMIC related concepts.

2.5.2. Study Selection Process

All retrieved records were collated and duplicates removed. Titles/abstracts were screened and then followed by full‐text assessment. The PRISMA flow was:

• Records identified: 86 (Scopus n = 73; PubMed n = 7; Web of Science n = 6)

• Duplicates removed: 10.

• Records screened (title/abstract): 76.

• Records excluded at title/abstract: 59.

• Full texts assessed: 17.

• Full texts excluded: 5.

• Studies included overall: 14 (Table 1; categorized by role in synthesis).

• Studies contributing to primary LMIC/UMIC mortality pooling: 3 (Amrayev2018; Paruk2020; Tabu2023).

• Contextual comparator included only in sensitivity analysis: 1 (Wijnen2021).

Table 1.

Characteristics of included studies and role in synthesis.

Article ID	Country	Income classification*	Study design	Sample size	Primary focus	Role in synthesis
Amrayev 2018	Kazakhstan	UMIC	Retrospective cohort	398	NOM vs operative mortality	Primary synthesis (meta‐analysis contributor)
Baidoo 2020	Ghana	LMIC	Prospective cohort	76	Mortality predictors	Primary synthesis
Dash 2015	India	LMIC	Prospective multicentre cohort	1031	Management patterns and outcomes	Primary synthesis
Dela 2021	South Africa	UMIC	Prospective multicentre cohort	1996	Access and surgical delay	Primary synthesis
Gu 2022	China	UMIC	Retrospective cohort	NR	Postoperative mortality	Primary synthesis
Gurger 2019	Turkey	UMIC	Retrospective cohort	188	1‐year mortality predictors	Primary synthesis
Hsu 2017	Taiwan	HIC	National database cohort	> 10,000	Socioeconomic mortality inequality	Primary synthesis (timing/inequity context)
Paruk 2020	South Africa	UMIC	Prospective cohort	200	12‐month mortality	Primary synthesis (meta‐analysis contributor)
Raichandani 2022	India	LMIC	Retrospective cohort	266	2‐year mortality	Primary synthesis
Tabu 2023	Kenya	LMIC	Prospective cohort	158	Fragility fracture outcomes	Primary synthesis (meta‐analysis contributor)
Wijnen 2021	Netherlands	HIC	Retrospective cohort	1279	Palliative NOM outcomes	Contextual comparator (excluded from primary pooled analysis; sensitivity analysis only)
Huang 2024	Multi‐country	Mixed‐income	Multicentre observational	NR	Income‐related management inequities	Contextual comparative evidence (not part of primary synthesis)
Xiang 2023	International	Mixed‐income	Cross‐sectional clinician survey	~400	Determinants of NOM	Determinants‐only evidence (qualitative synthesis only)
Armstrong 2022	Multi‐country (LMIC)	LMIC	Mixed‐methods	NR	Barriers to evidence‐based hip fracture care	Determinants‐only evidence (qualitative synthesis)

^*Income classification based on World Bank categorization at time of study conduct.

Screening was conducted by two independent reviewers and conflicts resolved by a third reviewer.

2.5.3. Data Items and Data Extraction

A standardized extraction framework was employed. Extracted variables included: study characteristics (country, income classification, setting, design, sample size, follow‐up), population details (age/sex, fracture type, premorbid function, frailty/comorbidity indices), management strategy definitions (NOM definition, operative strategy, timing metrics), outcomes (mortality timepoints and secondary outcomes), effect estimates (crude/adjusted RR/OR/HR/MD/SMD), adjustment variables/confounders, and contextual determinants (financing barriers, referral pathways, system constraints, provider decision drivers). Where raw counts were reported, unadjusted effect sizes have been calculated.

2.5.4. Risk of Bias Assessment

Risk of bias using tools appropriate to design:

• Newcastle‐Ottawa Scale (NOS) for Cohort Studies.

• ROBINS‐I for comparative non‐randomized evidence with high probability of confounding (e.g. palliative NOM vs. surgery).

• JBI tools for cross sectional studies and clinician surveys.

The major sources of bias, especially confounding by indication in comparisons of NOM, were reported in writing. The tabulation of risk‐of‐bias summaries was displayed in graphic form.

2.6. Data Synthesis and Statistical Analysis

All included studies were incorporated into the narrative synthesis. Determinants of management strategy were grouped into patient‐, provider‐, and health‐system‐level domains, while outcomes were organized according to operative management, NOM, and care‐pathway factors such as surgical delay. Meta‐analysis was undertaken when at least two studies reported sufficiently comparable mortality data for NOM and operative management.

The primary analysis was prespecified to include studies conducted in low‐ and middle‐income countries or upper‐middle‐income countries. Inclusion of the high‐income contextual comparator was prespecified as a sensitivity analysis. Planned subgroup analyzes according to national income classification and surgical timing were not performed because fewer than two sufficiently comparable studies were available within the relevant subgroups. No post hoc subgroup analyzes were undertaken.

Study‐specific risk ratios (RRs) and 95% confidence intervals (CIs) were calculated from arm‐level mortality data and pooled on the logarithmic scale using an inverse‐variance random‐effects model. Between‐study variance (τ²) was estimated using the DerSimonian–Laird method. Confidence intervals were calculated using the conventional Wald method. Statistical heterogeneity was assessed using Cochran's Q test and quantified using I² and τ². For Cochran's Q, p < 0.05 was considered evidence of heterogeneity. Small‐study effects were to be assessed only when at least 10 studies contributed to an analysis; this criterion was not met. Findings were interpreted according to the magnitude and precision of effect estimates rather than statistical significance alone.

Statistical analyzes were performed using R version 4.4.2 with the metafor package. Python was used exclusively for figure generation and visualization. Because the included comparative studies were observational and susceptible to confounding by indication, pooled estimates were interpreted as associations rather than causal effects.

2.7. Ethical Considerations

Ethical approval and informed consent were not required because this systematic review and meta‐analysis used data from previously published studies and involved no direct participant recruitment or identifiable individual‐level data.

3. Results

3.1. Study Selection

The database search identified 86 records: 73 from Scopus, seven from PubMed, and six from Web of Science. After duplicate removal and eligibility screening, 14 studies were included in the review. Three LMIC/UMIC cohort studies contributed to the primary mortality meta‐analysis, while one high‐income study was included only in the prespecified sensitivity analysis. The remaining studies contributed to the narrative synthesis of management determinants, care pathways, and outcomes. The study‐selection process is presented in Figure 1.

Figure 1.

Prisma flow chart.

3.2. Study Characteristics

The included evidence comprised prospective and retrospective cohort studies, registry‐based analyzes, and mixed‐methods or cross‐sectional studies. Study settings included LMICs, UMICs, selected high‐income contexts, and multicountry cohorts. Sample sizes ranged from 76 participants to more than 10,000 individuals. The characteristics and roles of the included studies are summarized in Table 1, while their population, fracture, and management characteristics are presented in Table 2.

Table 2.

Population, fracture, and management characteristics of key included studies.

Study	Population and fracture profile	Baseline characteristics and data gaps	Management strategy and NOM definition	Comparator, timing and role in synthesis
Amrayev et al. [22]	Older adults with hip fractures; fracture subtype not reported	Age‐ and sex‐stratified characteristics were incompletely reported	NOM versus operative management; NOM defined as conservative care without surgery	Operative management; timing not reported; primary meta‐analysis contributor
Paruk et al. [23]	Public‐sector hip fracture cohort	Twelve‐month follow‐up reported; frailty indices were not fully extractable	NOM versus operative management; detailed NOM definition not reported	Operative management; 12‐month mortality; primary meta‐analysis contributor
Tabu et al. [24]	Adults admitted with fragility hip fractures	COVID‐19‐era cohort; some outcome timepoints required clarification	NOM versus operative management; detailed NOM definition not reported	Operative management; surgical timing not consistently reported; primary meta‐analysis contributor
Dash et al. [25]	Older adults with fragility hip fractures in a multicentre Indian cohort	Fracture‐subtype breakdown was incompletely reported	Operative and non‐operative management patterns evaluated	Timing and comparator details were not sufficiently comparable for meta‐analysis; qualitative synthesis
Baidoo et al. (2020)	Single‐center Ghanaian hip fracture cohort	Comorbidities were assessed; detailed age distribution was incompletely reported	Management‐specific comparison was not the primary focus	Mortality predictors evaluated; qualitative outcome synthesis
Dela et al. (2021)	Multicentre South African low‐trauma hip fracture cohort	Access and health‐system factors were evaluated	Care‐pathway and access comparison rather than a direct NOM‐versus‐surgery comparison	Public versus private care and delay to surgery; determinants synthesis

3.3. Determinants of Management Strategy and Surgical Delay

The reported determinants of management strategy operated across provider and health‐system levels. Health‐system constraints included limited operating‐theater capacity, implant shortages, financing barriers, and differences in access between public and private healthcare pathways. Provider‐level considerations included perceived operative futility, concerns regarding perioperative risk, and resource diversion during the COVID‐19 pandemic. These determinants are summarized in Table 3.

Table 3.

Reported determinants of non‐operative management and delayed surgery.

Study	Setting and study design	Determinant domain	Reported determinants
Armstrong et al. [26]	Multicountry LMIC study; mixed‐methods design	Health‐system level	Limited operating‐theater capacity, implant shortages, and financial barriers to surgical care
Xiang et al. [27]	International study; clinician survey	Provider and health‐system levels	Perceived lack of operative benefit and diversion of surgical resources during the COVID‐19 pandemic
Dela et al. (2021)	South Africa; multicentre observational study	Health‐system level	Differences in access between public and private healthcare pathways and delays to definitive surgery

3.4. Mortality Outcomes

Mortality reporting varied across studies in terms of follow‐up timepoint, management comparison, and effect measure, limiting direct comparison of the individual estimates. Study‐level mortality outcomes and effect estimates are presented in Table 4, while the reported covariate‐adjustment strategies and residual data limitations are summarized in Table 5. Only studies providing sufficiently comparable arm‐level mortality data were included in the quantitative synthesis.

Table 4.

Mortality outcomes and effect estimates reported by the included studies.

Study	Outcome domain	Comparison or population	Timepoint	Reported estimate	Contribution to synthesis
Amrayev et al. [22]	Management‐stratified mortality	NOM versus operative management	1 year	65% versus 21%	Management‐stratified data contributed to the primary meta‐analysis
Tabu et al. [24]	Overall mortality	Entire hip‐fracture cohort	30 days	3.7%	Overall estimate included in the descriptive synthesis; separately extracted management‐stratified data contributed to the primary meta‐analysis
Paruk et al. [23]	Overall mortality	Entire hip‐fracture cohort	1 year	33.5%	Overall estimate included in the descriptive synthesis; separately extracted management‐stratified data contributed to the primary meta‐analysis
Baidoo et al. (2020)	Mortality predictor	Male versus female participants	4 years	HR 5.24 (95% CI NR)	Included in the narrative synthesis of mortality predictors
Paruk et al. [23]	Delay‐associated mortality	Mortality per additional day to surgery	1 year	HR 1.02 per day (95% CI NR)	Included in the narrative synthesis of surgical‐delay outcomes

Note: Overall mortality proportions were not pooled with management‐stratified estimates. The primary meta‐analysis used separately extracted arm‐level data comparing NOM with operative management. Effect estimates are presented as reported in the source studies.

Abbreviations: CI, confidence interval; HR, hazard ratio; NOM, non‐operative management; NR, not reported in the extracted dataset.

Table 5.

Reported covariate‐adjustment strategies in the included studies.

Study	Outcome or association assessed	Reported adjustment variables	Model or effect measure	Principal limitation
Amrayev et al. [22]	Mortality according to management strategy	Age, sex, ASA classification, and fracture type	Multivariable model; model family NR	Residual confounding by frailty, premorbid function, and treatment indication
Baidoo et al. [28]	Long‐term mortality predictors	Age, sex, creatinine level, and chronic obstructive pulmonary disease	HR; model family NR	Single‐center design and incomplete adjustment for frailty and functional status
Paruk et al. [23]	Mortality and delay to surgery	Age, sex, renal function, and surgical delay	HR; model family NR	Residual confounding by treatment selection, frailty, and functional status
Raichandani et al. (2021)	Long‐term mortality	Age, sex, and Charlson Comorbidity Index	Regression model; model family NR	Adjustment for fracture type and surgical timing was not reported
Dela et al. [29]	Access to care and surgical delay	Healthcare sector, province, age, and sex	Regression model; model family NR	Potential residual confounding by socioeconomic status, clinical severity, and referral pathway

3.5. Quantitative Synthesis of Mortality

Three LMIC/UMIC cohort studies—Amrayev et al., Paruk et al., and Tabu et al.—contributed to the primary meta‐analysis. In the random‐effects model, NOM was associated with higher mortality than operative management (RR 2.76, 95% CI 1.14–6.70; Figure 2). Between‐study heterogeneity was substantial (Q[2] = 12.77, p = 0.002; I² = 84.3%; τ² = 0.406). Given the limited number and observational design of the contributing studies, the pooled estimate was interpreted as an association rather than a causal effect.

Figure 2.

Forest plot of mortality following non‐operative versus operative management of hip fractures. Random‐effects meta‐analysis of three LMIC/UMIC cohort studies demonstrated a pooled risk ratio of 2.76 (95% CI 1.14–6.70). Between‐study heterogeneity was substantial (I² = 84.3%; τ² = 0.406).

3.6. Sensitivity and Subgroup Analyzes

The prespecified sensitivity analysis incorporating the high‐income contextual comparator altered the magnitude but not the direction of the pooled association. Because the comparator represented a clinically distinct palliative NOM pathway and had serious risk of confounding, the LMIC/UMIC‐restricted model was retained as the primary analysis. Planned subgroup analyzes by national income classification and surgical timing were not performed because the available studies were insufficiently comparable. The feasibility and outcomes of the prespecified analyzes are summarized in Table 6.

Table 6.

Prespecified primary, sensitivity, and subgroup analysesrisk of bias and certainty of evidence.

Prespecified analysis	Eligible studies	Purpose	Feasibility and analytical disposition
NOM versus operative management for mortality	Amrayev et al. [22], Paruk et al. [23], and Tabu et al. [24]	To estimate the association between management strategy and mortality in resource‐constrained settings	Performed. The primary random‐effects meta‐analysis was restricted to LMIC/UMIC studies
Inclusion of the high‐income contextual comparator	Wijnen et al. (2021), added to the three primary studies	To assess the influence of including a clinically distinct high‐income palliative NOM cohort	Performed as a sensitivity analysis. Inclusion altered the magnitude but not the direction of the association; the LMIC/UMIC‐restricted model was retained as primary
Early versus delayed surgery	Dash et al. [25], Paruk et al. [23], and Hsu et al. (2017)	To evaluate surgical delay as a potentially modifiable care‐pathway factor	Not pooled. Delay definitions, effect measures, and mortality timepoints were insufficiently comparable; findings were synthesized narratively
LMIC versus UMIC stratification	Studies conducted in eligible LMIC and UMIC settings	To explore whether estimates differed according to national income classification	Not performed. Too few comparable studies were available within each subgroup; findings were described by setting

Overall risk‐of‐bias judgments ranged from low to serious. Most cohort studies were rated as having moderate or low‐to‐moderate risk of bias, principally because of treatment‐selection differences and incomplete adjustment for relevant prognostic factors. The high‐income contextual comparator was judged to have serious risk of bias because of substantial baseline differences between the operative and palliative NOM groups. Overall and domain‐level assessments are presented in Tables 7A and 7B and Figure 3.

Table 7A.

Overall risk‐of‐bias judgments for the included studies.

Study	Appraisal tool	Overall judgment	Principal risk‐of‐bias concerns
Amrayev et al. [22]	NOS	Moderate	Selection bias, confounding by indication, and limited covariate adjustment
Dash et al. [25]	NOS	Moderate	Case‐mix differences and limited control of confounding
Paruk et al. [23]	NOS	Low–moderate	Residual confounding and treatment‐selection effects
Tabu et al. [24]	NOS	Moderate	Selection concerns and incomplete adjustment for relevant confounders
Baidoo et al. [28]	NOS	Low–moderate	Single‐center design and residual confounding
Dela et al. [29]	NOS	Low	Potential residual confounding related to healthcare sector
Gu et al. (2022)	NOS	Moderate	Retrospective design and incomplete covariate adjustment
Gurger et al. (2019)	NOS	Moderate	Retrospective design and limited adjustment for confounding
Hsu et al. (2017)	NOS	Moderate	Potential residual socioeconomic confounding
Huang et al. (2024)	NOS	Moderate	Baseline comparability and incomplete adjustment
Raichandani et al. (2021)	NOS	Moderate	Retrospective design and limited control of relevant confounders
Wijnen et al. (2021)	ROBINS‐I	Serious	Confounding by indication and substantial baseline frailty differences between management groups
Xiang et al. [27]	JBI cross‐sectional checklist	Moderate	Selection and self‐report bias
Armstrong et al. [26]	JBI mixed‐methods appraisal tools	Moderate	Sampling limitations and incomplete reporting of reflexivity

Table 7B.

Newcastle–Ottawa scale assessment of the included cohort studies.

Study	Selection, maximum 4	Comparability, maximum 2	Outcome, maximum 3	Total score, maximum 9	Overall judgment
Amrayev et al. [22]	★★★	★★	★★	7	Moderate
Baidoo et al. [28]	★★★	★★	★★★	8	Low–moderate
Dash et al. [25]	★★	★★	★★	6	Moderate
Dela et al. [29]	★★★	★★	★★★	8	Low
Gu et al. (2022)	★★	★★	★★	6	Moderate
Gurger et al. (2019)	★★★	★★	★★	7	Moderate
Hsu et al. (2017)	★★★	★★	★★	7	Moderate
Huang et al. (2024)	★★	★★	★★	6	Moderate
Paruk et al. [23]	★★★	★★	★★★	8	Low–moderate

Figure 3.

Summary of overall risk of bias across included studies. Risk of bias was assessed using the Newcastle–Ottawa Scale for cohort studies and ROBINS‐I for non‐randomized comparative studies. Bars represent overall judgment per study.

Using a GRADE‐informed approach, certainty of the pooled mortality evidence was judged very low because of the observational study designs, substantial heterogeneity, limited number of pooled studies, and confounding by indication.

NOM, non‐operative management. NR, not reported. Studies contributing directly to the mortality meta‐analysis are identified in the final column. Data gaps indicate variables that were unavailable or insufficiently reported in the source studies.

4. Discussion

4.1. Principal Findings

This systematic review was a synthesis of evidence on determinants and outcomes of hip fracture management in older adults in resource constrained environments. NOM was common (between 10% and > 30% of cases in LMIC/UMIC cohorts) in the 14 included studies and often found in settings that were defined by financing barriers and limited capacity to perform surgery. Three LMIC/UMIC cohort studies contributed to the primary mortality meta‐analysis, in which NOM was associated with higher mortality than operative management (RR 2.76, 95% CI 1.14–6.70). Nevertheless, the substantial heterogeneity and serious potential for confounding by indication limit causal interpretation of this association.

The substantial heterogeneity observed (I² high; τ² non‐trivial) likely reflects genuine between‐setting variability in patient selection, frailty burden, timing and access to surgery, peri‐operative support, and mortality ascertainment. In these circumstances, a random‐effects model is methodologically suitable because it presumes that the actual effect can vary in settings instead of being constant [30, 31].

4.2. Interpreting NOM: Clinical Selection Versus Structural Constraint

Confounding by indication is a major limitation of the mortality comparison. Patients selected for NOM were generally older, frailer, more comorbid, and more likely to have high ASA classifications, impaired premorbid mobility, or cognitive impairment—factors that independently increase mortality risk. Because the pooled analysis was based on unadjusted arm‐level data and adjustment strategies differed across studies, the higher mortality observed with NOM may partly reflect poorer baseline prognosis rather than an independent effect of treatment choice. The pooled estimate should therefore be interpreted as an association and not as evidence that NOM itself caused the excess mortality [30, 32].

Nonetheless, there is also LMIC/UMIC‐based evidence that suggests that NOM is not a purely clinical decision that is made on frailty or palliative objectives. Rather, it can be caused by structural constraints such as the lack of operating theaters, the shortage of implants, staffing, and delays in referral pathways and disastrous out‐of‐pocket costs (Dash 2015; Dela 2022; Armstrong 2022). In this context, it is critical to distinguish between:

• Palliative NOM, aligned with patient goals in extremely frail individuals, and

• Access‐constrained NOM, occurring when potentially operable patients cannot obtain timely surgery.

This difference has a policy and ethical implication. When NOM is an indicator of barriers at the system level and not patient preference, excess mortality can indicate unfair access to necessary surgical care and not ineffectiveness of treatment.

4.3. Multi‐Level Determinants of Management Strategy

Management determinants were grouped into three domains that are interdependent.

4.4. Patient‐Level Determinants

Selection of NOM and delayed surgery were affected by advanced age, comorbidity burden, cognitive impairment, renal dysfunction and low premorbid mobility (Paruk 2020; Baidoo 2021). The same factors are powerful predictors of baseline mortality risk, which increases the confounding by indication.

4.5. System‐Level Determinants

There were also continuous complaints about system constraints such as the lack of theater capacity, shortage of implants, a lack of timely referral systems, and the necessity of direct payment (Dash 2015; Armstrong 2022; Dela 2022). These limitations have a plausible role in explaining why the prevalence of NOM is large in restrictive environments amidst where operative care would be clinically reasonable.

Globally, the findings correspond to the larger evidence which has shown that avoidable mortality on conditions that can be treated remains where surgical systems lack infrastructure, workforce and funding security [33, 34].

4.6. Provider‐Level Determinants and Decision Ecology

Clinician perspectives indicate that operative decision thresholds shift in environments with limited capacity, particularly during system stressors such as pandemic‐related theater reductions (Xiang 2023; Tabu 2023). Making decisions under constraint should then be understood as situation‐specific as opposed to standard clinical practice.

4.7. Predominance of High‐Income Country Evidence in Hip Fracture Care

The wider hip fracture literature is dominated by evidence obtained from high‐income healthcare systems. Systematic reviews of performance indicators have charted over 170 studies in over 30 countries, but only a minority of these come from LMICs, for relatively small numbers of patients [35]. Core indicators—such as time to surgery, early mobilization, and orthogeriatric review—have largely been developed and validated within systems where theater access, implants, and workforce availability are structurally assumed.

Likewise, numerous systematic reviews and meta‐analyzes investigating thresholds of surgical timings (e.g. within 24–48 h), anesthetic practice and perioperative optimization combine mainly high‐income observational cohorts [13, 36, 37, 38, 39, 40]. These analyzes implicitly assume that early surgery is operationally possible.

Umbrella reviews of hip fracture practice guidelines show that recommendations to expedited surgery and multidisciplinary perioperative management are mostly consensus‐based and are within the context of well‐resourced systems [41]. Evaluations of systemwide improvement programs are also focused on optimizing adherence to already established benchmarks in high‐capacity hospitals rather than interrogating baseline surgical availability [42, 43, 44].

Only recently, dedicated reviews have been made trying to examine hip fracture surgery in resource‐limited environments identifying a relatively small global evidence base and emphasizing major contextual differences in financing, implant procurement and delays related to patient payment [45]. These analyzes point out that direct transfer of HIC‐derived benchmarks to constrained systems may be inappropriate without adaptation.

Taken together, existing literature optimizes care within stable, high‐capacity systems to a large degree. By contrast, our review specifically addresses settings in which surgical access itself is variable, thereby extending the evidence base to contexts that have historically been underrepresented in guideline formation.

4.8. Mortality and Plausible Mechanisms

Mortality among NOM patients was consistently high across included studies. Although causal inference is limited, biologically plausible pathways exist linking NOM to adverse outcomes. Long‐term immobilization leads to higher rates of venous thromboembolism, pneumonia, delirium, pressure injury, malnutrition, and functional decline, all of which are independently linked with mortality among geriatric trauma patients [5, 6]. On the other hand, operative fixation allows mobilization, pain management and prevention of complications when performed in a properly supported and well‐timed peri‐operative setting [5, 6].

Significantly, in limited systems operative advantage can be diminished by delayed presentation, limited anesthetic capacity and limited rehabilitation services. This is likely the reason behind the high heterogeneity in pooled analyzes.

4.9. Hip Fracture as a Sentinel Condition for Surgical System Capacity

Hip fracture represents a time‐sensitive, surgically treatable condition common among ageing populations. Outcomes therefore depend not only on individual clinical risk but also on the availability, timeliness, affordability, and quality of essential surgical services.

Our findings support the framing of hip fracture as a sentinel indicator of surgical system capacity. In the absence of timely fracture surgery, avoidable mortality and disability can be disproportionately represented by the older population. This framing is in line with the global surgery priorities that focus on making essential surgical systems in LMICs and UMICs stronger [33].

4.10. Explaining Heterogeneity

The high I² and non‐zero τ² likely reflect real clinical and system variability rather than statistical artifact. Contributing factors include:

• Variation in NOM indication (palliative vs access‐limited NOM).

• Differences in mortality timepoints (30‐day vs 1‐year).

• Case‐mix heterogeneity (fracture types, frailty, comorbidity).

• Differences in surgical delay thresholds.

• Variation in peri‐operative and rehabilitation support.

• Differences in outcome ascertainment (registry vs hospital‐based follow‐up).

Because only three studies contributed to pooling, heterogeneity estimates should be interpreted cautiously. Nevertheless, transparent reporting of Q, I², and τ² and the use of random‐effects modeling represent appropriate analytic practice [30, 31].

4.11. Certainty of Evidence and Interpretation

Using a GRADE‐informed framework, certainty of the evidence was judged very low because of the small number of observational studies, substantial heterogeneity, and confounding by indication [30, 46]. Although the confidence interval excluded the null value, uncertainty remains regarding the magnitude and causal interpretation of the association.

4.12. Policy and Practice Implications

The results justify a number of practical priorities of limited environments:

• Dedicated hip fracture pathways, prioritizing theater access and implant availability [5, 6].

• Financial protection mechanisms to reduce catastrophic out‐of‐pocket delays (Dela 2022; Armstrong 2022).

• Basic orthogeriatric co‐management principles, including delirium prevention, venous thromboembolism prophylaxis, and early mobilization planning.

• Clear documentation distinguishing palliative NOM from system‐driven NOM, improving ethical clarity and research interpretability.

4.13. Strengths and Limitations

The strengths of this study include; PRISMA 2020‐aligned reporting, implementation of MOOSE principles, clear distinction between contextual comparator evidence and primary inference, and clear reporting of heterogeneity and certainty [20, 21].

Limitations include observational design, confounding by indication, small numbers of pooled studies precluding publication‐bias testing, heterogeneity in NOM and delay definitions, and limited reporting of patient‐centered outcomes such as mobility recovery and quality of life.

4.14. Future Research

Future studies should use prospective multicentre registries with standardized measurement of age, frailty, ASA classification, comorbidity, cognitive status, premorbid mobility, treatment goals, and health‐system barriers. Comparative observational studies should apply methods such as propensity‐score matching or inverse‐probability weighting to improve baseline comparability between operative and NOM groups. Studies should also distinguish palliative NOM from access‐constrained NOM and report standardized mortality, functional recovery, quality‐of‐life, and complication outcomes. Quasi‐experimental evaluations are needed to determine whether improved financing, referral pathways, theater access, and implant availability reduce surgical delays and improve outcomes.

5. Conclusion

In resource‐constrained settings, NOM and delayed hip fracture care are common and associated with higher mortality although certainty of pooled estimates is very low due to bias, heterogeneity and imprecision. These findings reinforce hip fracture as a sentinel condition for surgical system capacity and highlight urgent opportunities for health‐system strengthening to improve equitable access to timely, effective care for ageing populations.

Author Contributions

Abdirahman Ibrahim Ismail: conceptualization, methodology, investigation, data curation, writing – original draft. Otim Aaron Thomas: conceptualization, methodology, data curation, investigation. Vivian Onyinye Ojiakor: conceptualization, investigation, validation, formal analysis, data curation. Nto Johnson Nto: methodology, validation, formal analysis, data curation, visualization. Augusine Oviosun: conceptualization, methodology, formal analysis, validation, visualization. Chinyere N. Anyanwu: visualization, writing – original draft, writing – review and editing, project administration, formal analysis. Wusa Makena: investigation, validation, formal analysis, software, data curation. Ugwu Okechukwu Paul Chima: visualization, project administration, resources, supervision, formal analysis. Emeka Godson Anyanwu: conceptualization, methodology, project administration, writing – original draft, writing – review and editing, visualization.

Funding

The authors have nothing to report.

Ethics Statement

Ethical approval was not required because this systematic review and meta‐analysis synthesized data from previously published studies and did not involve direct participation of human subjects or access to identifiable individual‐level patient data.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

Emeka Godson Anyanwu, the manuscript guarantor, affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.