Improving Management of Adult Ankle Fractures in Malawi

Abstract

Background:

The burden of musculoskeletal trauma is increasing worldwide, especially in low-income countries such as Malawi. Ankle fractures are common in Malawi and may receive suboptimal treatment due to inadequate surgical capacity and limited provider knowledge of evidence-based treatment guidelines.

Methods:

This study was conducted in 3 phases. First, we assessed Malawian orthopaedic providers’ understanding of anatomy, injury identification, and treatment methods. Second, we observed Malawian providers’ treatment strategies for adults with ankle fractures presenting to a central hospital. These patients’ radiographs underwent blinded, post hoc review by 3 U.S.-based orthopaedic surgeons and a Malawian orthopaedic surgeon, whose treatment recommendations were compared with actual treatments rendered by Malawian providers. Third, an educational course addressing knowledge deficits was implemented. We assessed post-course knowledge and introduced a standardized management protocol, specific to the Malawian context.

Results:

In Phase 1, deficits in injury identification, ideal treatment practices, and treatment standardization were identified. In Phase 2, 17 (35%) of 49 patients met operative criteria but did not undergo a surgical procedure, mainly because of resource limitations and provider failure to recognize unstable injuries. In Phase 3, 51 (84%) of 61 participants improved their overall performance between the pre-course and post-course assessments. Participants answered a mean of 32.4 (66%) of 49 questions correctly pre-course and 37.7 (77%) of 49 questions correctly post-course, a significant improvement of 5.2 more questions (95% confidence interval [CI], 3.8 to 6.6 questions; p < 0.001) answered correctly. Providers were able to identify 1 more injury correctly of 8 injuries (mean, 1.1 questions [95% CI, 0.6 to 1.6 questions]; p < 0.001) and to identify 1 more ideal treatment of the 7 that were tested (mean, 1.0 question [95% CI, 0.5 to 1.4 questions]; p < 0.001).

Conclusions:

Adult ankle fractures in Malawi were predominantly treated nonoperatively despite often meeting evidence-based criteria for surgery. This was due to resource limitations, knowledge deficits, and lack of treatment standardization. We demonstrated a comprehensive approach to examining the challenges of providing adequate orthopaedic care in a resource-limited setting and the successful implementation of an educational intervention to improve care delivery. This approach can be adapted for other conditions to improve orthopaedic care in low-resource settings.

Trauma is a leading cause of morbidity and mortality worldwide, disproportionately affecting low-income and middle-income countries^1–4. Trauma-related disability is especially devastating to poor individuals, as health care-related costs and decreased productivity can worsen impoverishment^5,6. Disability can be mitigated with quality surgical care, which remains inaccessible for most of the world’s poorest people^7–9.

Malawi is a low-income country in southeastern Africa of 19 million people, 83% living in rural areas, half living below the national poverty line, and 10% living with an untreated surgical condition of the upper or lower extremities^10–12. Ankle fractures were found to be the most common adult injury seen in orthopaedic outpatient departments of 4 Malawian public hospitals¹³.

In high-income countries such as the United States, a surgical procedure is recommended for many displaced and/or unstable ankle fractures^14–16. However, access to surgical procedures remains inadequate in Malawi. Orthopaedic care is provided in 24 district hospitals and 4 urban central hospitals. Over 90% of orthopaedic trauma care is delivered by orthopaedic clinical officers (OCOs), who are non-physician clinicians trained in nonoperative care^17,18. Patients with ankle fractures are usually evaluated by OCOs, who triage patients to nonoperative treatment or refer patients to an orthopaedic surgeon for a surgical procedure. At the time of this writing, 10 orthopaedic surgeons were practicing in Malawi, roughly 1 per 1.9 million Malawians. All are board-certified, fellowship-trained, and stationed at central hospitals, where orthopaedic surgical procedures are available. However, surgeons must consider resource availability and each patient’s unique circumstances when recommending a surgical procedure. Implants are entirely donated, with frequent shortages; patients often present late and face challenges to follow-up^13,19,20. In this context, we examined how ankle fractures are managed in Malawi and the barriers to evidence-based care.

Materials and Methods

Overview

Our study had 3 phases. In Phase 1, we assessed orthopaedic providers’ knowledge of ankle anatomy, injury classification, and ideal treatment methods. We surveyed providers on actual treatment practices and their rationale. In Phase 2, we observed Malawian providers’ treatment strategies for adult ankle fractures over 5 weeks at Kamuzu Central Hospital (KCH), the only public referral hospital in central Malawi. In Phase 3, we implemented an educational course addressing knowledge deficits and introducing a standardized protocol for ankle fracture management in Malawi. All phases were approved by the College of Medicine Research Ethics Committee (COMREC-P.03/19/2628,P.03/19/2629) in Malawi and by the institutional review board at Brigham and Women’s Hospital in the United States. Statistical analyses were performed with SAS version 9.4 (SAS Institute).

Phase 1: Knowledge Assessment

We designed a paper-based knowledge assessment (Appendix A) in English, the language of medical instruction in Malawi. This included a figure of ankle osseous and ligamentous anatomy to label, a multiple-choice question on syndesmosis anatomy, and 8 clinical vignettes with radiographs demonstrating the following fractures: Weber A lateral malleolar, Weber C lateral malleolar, bimalleolar, trimalleolar, lateral malleolar with increased medial clear space (bimalleolar-equivalent), ligamentous Maisonneuve, bimalleolar fracture-dislocation, and open trimalleolar fracture-dislocation. Multiple-choice questions tested injury identification and knowledge of ideal treatment methods. We additionally surveyed actual treatment methods used by participants. When ideal and actual treatments differed, we solicited participants’ explanations.

All orthopaedic providers working at or near KCH were invited to participate. We disclosed the subject matter immediately prior to participation, obtained informed consent, and served as proctors to ensure that external resources and peers were not consulted. No identifying information was solicited; only participants’ job titles and degrees were recorded. All participants received $10 cash compensation paid in Malawian kwacha.

Phase 2: Observations of Treatment Strategies

Between May 27 and June 28, 2019, a research assistant (A.K.) shadowed providers at KCH and invited all adults with ankle fractures encountered during routine care to participate in the study. All patients with open and closed injuries and those with acute, subacute, and chronic injuries were included. Pediatric patients (<18 years of age) and patients with polytrauma were excluded. Written informed consent was obtained in the Malawian language Chichewa and English. Patients were compensated $10 paid in Malawian kwacha.

Age, sex, residence, comorbidities, occupation, education, injury mechanism, and prior treatments were recorded for each patient using a secure smartphone-based tool (CommCare; Dimagi). The Malawian provider caring for each patient classified the injury and gave his or her intended treatment plan and rationale. Providers were not identified; only job titles and degrees were recorded.

Deidentified radiographs for each patient were collected and were analyzed post hoc by 3 U.S.-based foot and ankle fellowship-trained orthopaedic surgeons and 1 trauma fellowship-trained Malawian senior orthopaedic surgeon. Blinded to Malawian providers’ classifications and treatment plans, the reviewers independently classified each injury and gave treatment recommendations. The percent agreement and Fleiss kappa coefficient of agreement were calculated between U.S. surgeons. The differences between U.S. surgeons were discussed and were reconciled by consensus. We then compared classification and treatment recommendations between the Malawian providers, the Malawian surgeon, and U.S. surgeons, calculating the percent agreement and Cohen kappa coefficient of agreement.

Phase 3: Educational Course

Based on Phases 1 and 2, a team of Malawian and U.S. faculty, including the study investigators, designed an educational course to review anatomy and injury classification, to review evidence-based treatment guidelines, and to present a standardized protocol for ankle fracture management feasible in Malawi (Appendix B). Jointly implemented by Malawian and U.S. faculty, the course took place on September 26, 2019, during the annual Malawi Orthopaedic Association gathering for continuing professional development. It included lectures, case-based discussions, and practical skills training (Appendix C).

Resembling the assessment in Phase 1, pre-course assessments were performed on the morning prior to the course and post-course assessments were performed on the morning after the course (Appendix D). All providers in attendance were invited to participate and written informed consent was obtained. Course instructors served as proctors to ensure that assessments were completed without the help of external resources or peers. No identifying information was solicited; only participants’ job titles and degrees were recorded. Unique participant numbers were used to match pre-course and post-course assessments, and change in performance was analyzed using paired t tests. All participants were compensated $10 paid in Malawian kwacha.

Results

Phase 1: Knowledge Assessment

Fifteen OCOs, 2 orthopaedic residents, and 1 attending physician participated. Practice experience ranged from 5 months to 29 years (median, 9.5 years [interquartile range (IQR), 5 to 11 years]). Overall, of the 35 questions, the median score was 61% (IQR, 49% to 70%) (21 questions), and performance varied widely, with participants answering 37% (13 questions) to 91% (32 questions) correctly.

Participants scored highest in anatomy (median, 81% [IQR, 62% to 92%]), identifying a median of 6 of 7 osseous structures and 4 of 5 ligamentous structures correctly. Eight participants (44%) correctly identified syndesmotic anatomy. Participants correctly answered a median of 63% (IQR, 53% to 75%) of questions on osseous injury identification and 50% (IQR, 33% to 67%) of questions on ligamentous injury identification. Participants scored lowest in the knowledge of ideal treatment (median, 43% [IQR, 29% to 57%]) (Fig. 1).

Fig. 1.

Phase-1 knowledge assessment scores by question category. Scores for each participant were calculated as the percentage of questions answered correctly, by question category and overall. Questions were categorized as pertaining to anatomy, osseous injury identification, ligamentous injury identification, and knowledge of ideal treatment principles. In the box-and-whisker plots, the left-most boundary of the box indicates the 25th percentile, the black line within the box marks the median, the diamond marks the mean, and the right-most boundary of the box indicates the 75th percentile. Whiskers to the left and right indicate the minimum and maximum.

With regard to actual treatment practices, for 1 injury type (the bimalleolar-equivalent fracture), 12 (67%) of 18 participants agreed on the same treatment (manipulation under anesthesia, followed by casting or splinting). For the other 6 of the 7 injury types, there was no majority agreement (>9 of the 18 participants) on the actual treatment practice. Participants demonstrated no consensus on whether the following injuries required referral for a surgical procedure: Weber C, bimalleolar-equivalent, and ligamentous Maisonneuve (Fig. 2). Participants also identified several challenges in providing ideal treatment, most commonly resource limitations (17 [94%] of 18 participants) and inadequate training (13 [72%] of 18 participants) (Table I).

Fig. 2.

Phase-1 participants’ referral preferences for specific ankle fracture types. The percentage of participants who would (yes) or would not (no) refer a patient to central hospital for a surgical procedure is presented for each ankle fracture type tested in the knowledge assessment.

TABLE I.

Phase 1: Challenges in Delivering Ideal Treatment

Challenge*	No. of Participants (N = 18)
Resources are too limited to give every patient the ideal treatment	17 (94%)
I am not adequately trained to perform the ideal treatment	13 (72%)
I do not have the necessary equipment and/or material to safely perform the ideal treatment	8 (44%)
I do not have access to the facilities required to safely perform the ideal treatment	3 (17%)
I rarely see this kind of injury	1 (6%)

^*Challenges denote reasons that actual treatment practices differed from ideal treatment.

For additional results, including participant demographic characteristics, performance on identifying specific injury types and their ideal treatment, and actual treatment preferences, see Appendix E.

Phase 2: Observations of Treatment Strategies

We observed treatment of 52 adults with ankle fractures, all treated by OCOs with diplomas in clinical orthopaedics as their highest qualification. The mean age was 42.6 years; 60% were male, and 92% of patients came from the Lilongwe district. Patient characteristics are summarized in Table II.

TABLE II.

Phase 2: Patient Characteristics^*

Characteristic	Value
No. of patients	52
Age* (yr)	42.6 ± 10.4
Sex†
Female	21 (40%)
Male	31 (60%)
District†
Dowa	1 (2%)
Lilongwe	48 (92%)
Mzimba	1 (2%)
Salima	2 (4%)
Employment type†‡
Formal	27 (52%)
Informal	16 (31%)
Unemployed, retired, or student	9 (17%)
Occupation†
Farmer	4 (8%)
Housewife	1 (2%)
Laborer	2 (4%)
Office worker	1 (2%)
Other	20 (38%)
Small-scale business	16 (31%)
Student	1 (2%)
Unemployed	7 (13%)
Income level†§
0 to 10,000 kwacha/month	14 (27%)
10,001 to 20,000 kwacha/month	3 (6%)
20,001 to 30,000 kwacha/month	7 (13%)
30,001 to 40,000 kwacha/month	3 (6%)
40,001 to 50,000 kwacha/month	7 (13%)
>50,000 kwacha/month	18 (35%)
Education level†
No schooling	4 (8%)
Primary grade 1 to 4	8 (15%)
Primary grade 5 to 8	13 (25%)
Secondary or above	27 (52%)
Medical comorbidities†
Hypertension	5 (10%)
Asthma	2 (4%)
Diabetes	1 (2%)
None	44 (85%)
Time since injury* (days)	49.3 ± 60.6
Mechanism of injury†
Other	6 (12%)
Assault	1 (2%)
Fall	26 (50%)
Road traffic collision	19 (37%)
Laterality†
Left	24 (46%)
Right	28 (54%)
Neurovascular status†
Intact	49 (94%)
Sensory deficit	1 (2%)
Vascular injury	2 (4%)
Open fracture†	8 (15%)
Referred to KCH†	27 (52%)
Christian Health Association of Malawi hospital	7 (26%)
District hospital	8 (30%)
Health center	11 (41%)
Other	1 (4%)

^*The values are given as the mean and the standard deviation.

^†The values are given as the number of patients, with the percentage in parentheses.

^‡Formal employment is registered, regulated, and protected by a country’s existing legal or regulatory framework. Formal workers have work contracts, benefits, social protection, and workers’ representation. In contrast, informal employment is defined as self-employment or wage employment that is not registered, regulated, or legally protected. Informal workers do not have formal work contracts, benefits, social protection, or workers’ representation. Informal workers face higher risks of poverty than workers in the formal economy³⁰.

^§10,000 Malawian kwacha ≈ $13.60 in U.S. dollars.

Forty-nine patients (94%) had adequate radiographs for post hoc analysis. There was substantial to near-perfect agreement between U.S. surgeons in injury characterization except syndesmotic injury identification, which had moderate agreement at 57% (Fleiss kappa, 0.35 [95% confidence interval (CI), 0.19 to 0.51]) (Table III). By U.S. surgeon consensus, there were 3 Weber-A fractures (6%), 33 Weber-B fractures (67%), and 13 Weber-C fractures (27%). Isolated lateral malleolar fractures were most common (22 [45%]), followed by bimalleolar fractures (16 [33%]) and then trimalleolar fractures (6 [12%]). Five patients (10%) had fracture-dislocations.

TABLE III.

Phase 2: Agreement in Fracture Characterization and Treatment Recommendations

	Among U.S. Surgeons*		Between Malawian OCOs and U.S. Surgeons†
	Percent Agreement	Fleiss Kappa‡	Percent Agreement	Cohen Kappa‡
Injury classification
Weber classification	86%	0.79 (0.66 to 0.93)
Osseous injury	78%	0.77 (0.67 to 0.87)
Syndesmotic injury	57%	0.35 (0.19 to 0.51)
Medial clear space	77%	0.65 (0.43 to 0.86)
Fracture characteristics
Nondisplaced	78%	0.68 (0.52 to 0.84)	80%	0.60 (0.38 to 0.82)
Displaced	73%	0.63 (0.47 to 0.79)	78%	0.57 (0.35 to 0.80)
Unstable	88%	0.83 (0.67 to 0.99)	65%	0.38 (0.09 to 0.66)
Dislocated	92%	0.76 (0.60 to 0.92)	80%	0.29 (0 to 0.66)
Treatment recommendation (operative vs. nonoperative)	90%	0.86 (0.69 to 1.00)	65%	0.38 (0.09 to 0.66)

^*Three foot and ankle fellowship-trained orthopaedic surgeons examined deidentified radiographs for each patient in Phase 2 and gave blinded, post hoc injury characteristics and treatment recommendations.

^†Consensus fracture characteristics and treatment recommendations among U.S. surgeons were then compared with the fracture characteristics and intended treatment plans formulated by the Malawian OCOs caring for the patients. Malawian OCOs were not asked to provide a Weber classification or describe the osseous and ligamentous injury on radiographs.

^‡The values are given as the kappa value, with the 95% CI in parentheses.

There was moderate agreement between Malawian OCOs and U.S. surgeons when identifying nondisplaced and displaced fractures, fair agreement when identifying unstable and dislocated injuries, and fair agreement when recommending nonoperative treatment compared with operative treatment (Table III). Malawian OCOs planned nonoperative management for 34 patients (69%) and planned to refer 15 patients (31%) for surgery. U.S. surgeons recommended nonoperative treatment without manipulation for 17 patients (35%). All had isolated nondisplaced lateral malleolar fractures. Operative treatment was recommended for 32 patients (65%), all of whom had displaced, unstable, and/or dislocated fractures (Fig. 3). The Malawian surgeon and the U.S. surgeons had near-perfect agreement in treatment recommendations at 94% (Cohen kappa, 0.87 [95% CI, 0.72 to 1.00]).

Fig. 3.

Comparison of Malawian OCOs’ treatment plans and U.S. surgeons’ consensus post hoc treatment recommendations for adult ankle fractures seen at KCH during Phase 2. The Malawian OCO caring for each patient provided the intended treatment plan. Three foot and ankle fellowship-trained orthopaedic surgeons based in the United States, blinded to the treatment plans of the Malawian OCOs, gave post hoc treatment recommendations based on deidentified patient radiographs.

Of the 17 patients for whom treatment recommendations differed between the Malawian OCOs and the U.S. surgeons, all met operative criteria per U.S. surgeons but were treated nonoperatively by Malawian providers. Four were treated with casting or splinting after manipulation, and 13 were treated with casting or splinting only. All were Weber-B or C injuries, classified as displaced or unstable by U.S. providers. Malawian OCOs stated that resources were too limited to provide ideal treatment for 9 (53%) of the 17 patients, including all 4 treated with casting or splinting after closed reduction. One of the 17 patients should have been treated surgically, according to the Malawian OCOs, but nonoperative treatment had already been started at a district hospital and was continued. For 7 (41%) of the 17 patients, Malawian OCOs stated that nonoperative treatment was ideal; however, the Malawian surgeon recommended a surgical procedure for 6 of these 7 patients. None of the 7 fractures were reported as displaced or unstable by Malawian OCOs.

For additional results comparing fracture classification and treatment plans and recommendations made by Malawian OCOs and U.S. surgeons, see Appendix E.

Phase 3: Educational Course

Sixty-one participants completed pre-course and post-course assessments. Fifty-five of these 61 participants were orthopaedic providers and made up 40% of all practicing orthopaedic providers in Malawi at that time. Providers represented 31 hospitals, including all 4 central hospitals and 16 of 25 district hospitals.

Scores improved for 51 (84%) of 61 participants. The mean overall score improved from 32.4 points (66% of 49 points) pre-course to 37.7 points (77%) post-course (Fig. 4-A). This represented a significant 5.2 additional questions (95% CI, 3.8 to 6.6 questions; p < 0.001) answered correctly. A significant improvement was found for OCOs with <10 years of experience (6.8 questions [95% CI, 4.1 to 9.6 questions]; p < 0.001), OCOs with ≥10 years of experience (5.4 questions [95% CI, 3.3 to 7.5 questions]; p < 0.001), and OCOs with a BSc (2.1 questions [95% CI, 0.4 to 3.8 questions]; p = 0.022). All participants had a significant improvement in performance except for those whose scores before the course were in the top quintile. The largest improvements were seen for those in the pre-course performance quintile 1 (11.6 questions [95% CI, 6.8 to 16.4 questions]; p < 0.001) and quintile 2 (6.8 questions [95% CI, 2.7 to 10.9 questions]; p < 0.001) (Fig. 4-B).

Fig. 4.

Figs. 4-A and 4-B Phase-3 knowledge assessment performance by participant group. Fig. 4-A The mean overall scores. The mean pre-course and post-course assessment scores (percentage of total questions answered correctly) were calculated for the cohort as a whole, by job title or years of experience (OCOs with <10 years of experience, OCOs with ≥10 years of experience, OCOs with a BSc degree, orthopaedic surgeons, and others) and by pre-course performance quintile. Fig. 4-B The mean change in performance. The change in number of questions answered correctly between the pre-course and post-course assessments was calculated for each participant. The mean change in performance was calculated by participant group. The asterisk denotes a significant change (p < 0.05) in mean scores between the pre-course and post-course assessments, calculated using paired t tests.

Overall, the cohort demonstrated a significant improvement in all question categories. The cohort was able to identify 1 more injury correctly of the 8 injuries that were tested (1.1 questions [95% CI, 0.6 to 1.6]; p < 0.001) and to identify 1 more ideal treatment of the 7 that were tested (1.0 question [95% CI, 0.5 to 1.4 questions]; p < 0.001). Greater than two-thirds were able to correctly identify all injuries in the post-course assessment except for bimalleolar fractures (48% pre-course and 54% post-course).

See Appendix E for additional results with regard to participant demographic characteristics, mean change in performance by question category, and performance on identifying specific injury types and their ideal treatment before and after the course.

Discussion

Patients sustaining ankle fractures in Malawi may be treated suboptimally because of resource limitations, knowledge deficits, and lack of treatment standardization. The expansion of surgical capacity is essential if the Malawian health-care system is to meet the needs of a rapidly growing population subject to a high trauma burden¹⁹.

In Phases 1 and 2, we found that resource limitations caused providers to deviate from evidence-based treatment practices and treat unstable fractures nonoperatively. In our previous assessment of trauma care capacity in Malawi, we found that most district hospitals lacked basic resources including radiography, likely impeding timely diagnosis and treatment decision-making. Two of 4 central hospitals (the only facilities offering surgical procedures for ankle fractures) reported limited availability of operating rooms, surgeons, and suture¹⁹. These resource limitations undoubtedly limit surgical availability and should be addressed nationally through infrastructure development and improved resource procurement and allocation¹⁹.

Phases 1 and 2 also identified that important knowledge deficits existed, largely among OCOs who are trained primarily in nonoperative treatment¹⁸. This resulted in potentially operative injuries failing to be referred to an orthopaedic surgeon. As gatekeepers to surgical procedures, OCOs must be able to properly assess an injury and determine whether it meets surgical criteria to effectively deliver care. In Phase 3, we addressed these knowledge deficits, especially with regard to the assessment of injury instability and treatment decision-making. Education is critical to building and maintaining surgical capacity^21,22. Educational courses collaboratively designed by high-income country and low-income and middle-income country partners have been shown to improve provider knowledge and skills, facilitating improved care and dissemination of knowledge^23,24. Participants in our educational course demonstrated an 11% mean score improvement, comparable with the results of similar orthopaedic courses conducted in Haiti^24,25. Persistent knowledge deficits, especially with regard to Weber-B lateral malleolar, bimalleolar, and bimalleolar-equivalent fractures, should be addressed in future studies.

We also developed and disseminated a standardized protocol for ankle fracture management. Care standardization is essential to improve quality and safety²⁶ and can improve resource utilization by ensuring that patients with injuries that should be treated operatively are reliably referred to central hospitals where surgical procedures are available. Recognizing that this may increase the burden on central hospitals, we designed our protocol to safely treat as many patients nonoperatively as possible. We also standardized referral practices to encourage timely transfer of surgical patients to avoid more challenging and costly treatments associated with delayed care. Our protocol (Appendix B) was intended to exist within the reality of Malawi’s current health-care system and to be revised and adapted in the future. Provider adherence and the protocol’s effects on resource utilization, provider workload, and patient outcomes must be investigated in future studies.

This study had several limitations. First, social desirability bias may have caused respondents to overreport or underreport adherence to evidence-based treatment, assuming that the study investigators desired these responses. Second, Phases 1 and 2 were conducted at KCH only, possibly limiting generalizability. We focused on care delivery at KCH, where high rates of delayed presentation and limited treatment standardization were previously observed^13,27, in an effort to examine challenges where they were perhaps greatest, and thus have the greatest impact on care nationally when implementing our educational course. Third, our post hoc treatment recommendations based on radiographic review were subject to personal bias. Although not necessarily representative for all orthopaedic surgeons in high-income countries, operative treatment was recommended for open, unstable, and/or displaced ankle fractures, which was justifiable on the basis of current evidence²⁸. There was substantial to near-perfect agreement between the 3 U.S. surgeons in all domains, except in identification of syndesmotic injury. In the absence of gross diastasis, the diagnosis of subtle syndesmotic injury on radiographs is challenging and preoperative and/or intraoperative stress testing is often required. Stress radiography, which is not commonly performed in Malawi, might have improved interrater agreement when assessing the syndesmosis²⁹.

Fourth, Phase 2 was conducted over a 5-week period during Malawi’s dry season, possibly limiting sample size. Because fewer patients present during the dry season¹³, resources may have been more available to perform surgical procedures than in the rainy season¹⁹. Access to surgery for an ankle fracture may be even lower than what was observed in this study. A retrospective study might have been preferable but was impossible in this setting because detailed patient records are not kept at KCH or at most Malawian public hospitals. Despite the logistical challenges of conducting clinical research in this setting, the information gathered demonstrated important disparities between Malawian and U.S. providers’ recommendations. This underscores the need to conduct clinical research in resource-limited settings to understand why evidence-based guidelines in high-income countries may be incompletely implemented or inapplicable in low-income and middle-income countries.

Lastly, it remains unknown whether knowledge gained by Malawian providers will be retained and will translate to improved care. Further investigation of knowledge retention and adherence to the standardized protocol is necessary.

In conclusion, we demonstrate a comprehensive investigative approach to examining the challenges of providing adequate orthopaedic care for a common injury in a resource-limited country and the successful implementation of an educational intervention to improve care capacity and to standardize treatment for ankle fractures.

Appendix A –. Phase 1 provider knowledge assessment

Correct answers are indicated in bold, where applicable.

Appendix B –. Standardized protocol for ankle fracture management in Malawi

Appendix C –. Ankle fracture educational course agenda

Appendix D –. Phase 3 pre- and post-course participant knowledge assessment

Correct answers are indicated in bold, where applicable.

Appendix E –. Additional results

Phase 1

Table E1:

Phase 1 – Participant demographics

Total	18
Title
Orthopaedic Clinical Officer	15
Orthopaedic Resident	2
Orthopaedic Surgeon	1
Highest Level of Training
Diploma	13
BSc	2
MBBS	2
MD PhD	1
Place of Work
Kamuzu Central Hospital	10
District hospital	5
Mission hospital	2
Kamuzu Barracks	1

Figure E1: Percentage of phase 1 participants who A) correctly identified specific ankle fracture types, and B) correctly identified the ideal treatment of each fracture type.

The following ankle fracture types were tested: Weber A lateral malleolar, Weber C lateral malleolar, bimalleolar, trimalleolar, lateral malleolar with increased medial clear space (bimalleolar-equivalent), ligamentous Maissonneuve (only injury identification was tested), bimalleolar fracture-dislocation, and open trimalleolar fracture-dislocation.

Participants correctly answered a median of 63% of questions that tested identification of bony injuries (IQR 53-75%), and 50% of questions that tested identification of ligamentous injuries (IQR 33-67%). Sixteen participants (89%) correctly identified Weber A lateral malleolar and trimalleolar fractures, 15 (83%) correctly identified Weber C lateral malleolar fracture, 10 (56%) correctly identified bimalleolar fracture, and 9 (50%) or fewer correctly identified bimalleolar-equivalent, ligamentous Maissoneuve, bimalleolar fracture-dislocation, and open trimalleolar fracture-dislocation on x-ray.

Table E2:

Phase 1 – Actual treatment preferences

Injury type	Treatment preference	Number of participants (%)
Weber A lateral malleolus fracture	Casting/splinting without manipulation	9/18 (50)
	Immediate MUA, followed by casting/splinting only	7/18 (39)
Weber C lateral malleolus fracture	Temporary casting/splinting without manipulation, followed by ORIF in 1–2 weeks	5/18 (28)
Bimalleolar fracture	Immediate MUA, temporary casting/splinting after CR, then ORIF in 1–2 weeks	7/18 (39)
Trimalleolar fracture with syndesmotic and deltoid injury	Immediate MUA, temporary casting/splinting after CR, then ORIF in 1–2 weeks	9/18 (50)
Bimalleolar-equivalent fracture	MUA, followed by casting/splinting	12/18 (67)
Bimalleolar fracture-dislocation	Emergent operative exploration, compartment release, and ORIF	7/18 (39)
	Immediate MUA, temporary casting/splinting after CR, then ORIF in 1–2 weeks	6/18 (33)
Open trimalleolar fracture-dislocation	Immediate MUA and wound irrigation, temporary external fixation after CR, then wound debridement and ORIF within 24 hours	4/18 (22)

With regards to actual treatment preferences, for eight of the nine injury types presented, no clear consensus was demonstrated, with 9 participants (50%) or fewer reporting the same treatment preference. Only for bimalleolar-equivalent fractures was a clear dominant treatment modality demonstrated, with 12 of the 18 participants reporting preferred treatment with manipulation under anesthesia (MUA) and casting/splinting.

Phase 2

Figure E2: Post-hoc classification of ankle fractures for adult patients seen at Kamuzu Central Hospital during phase 2.

Deidentified x-rays for each patient seen during phase 2 were collected and independently analyzed post-hoc by three Foot and Ankle fellowship-trained orthopaedic surgeons based in the United States. Classification differences were discussed and reconciled to give a uniform classification of each injury representing consensus among the US-based providers.

Table E3:

Phase 2 - Treatment plans/recommendations made by Malawian OCOs and US surgeons

Treatment plans/recommendations	Malawian providers (ad hoc)	US providers (post-hoc)
Non-operative	34 (69)	17 (35)
Casting/splinting without manipulation	30 (61)	17 (35)
Casting/splinting after closed reduction	4 (10)	0 (0)
Operative	15 (31)	32 (65)
Temporary casting/splinting without manipulation	8 (16)	15 (35)
Temporary casting/splinting after closed reduction	4 (8)	17 (31)
No temporary immobilization	3 (6)	0 (0)
External Fixation	3 (6)	1 (2)
Open reduction and internal fixation	12 (24)	31 (63)

Malawian OCOs planned nonoperative management for 34 patients (69%), which included casting/splinting without manipulation for 30 patients (61%) and after manipulative closed reduction for 4 patients (8%). Operative treatment was planned/performed for 15 patients (31%). Prior to surgery, temporary immobilization by casting/splinting alone was recommended for 8 patients (16%), and casting/splinting after a closed reduction was recommended for 4 patients (10%).

The US surgeons recommended non-operative treatment of casting without manipulation for 17 patients (35%). All had non-displaced fractures of the lateral malleolus only with no syndesmotic or medial clear space widening. This included all three Weber A lateral malleolus fractures, 13 Weber B fractures, and one Weber C fracture. Operative treatment was recommended for 32 patients (65%), all of which had displaced and/or unstable Weber B or C fractures. This included all 18 patients with syndesmotic injuries, and all 5 patients with medial clear space widening. Four had tibiotalar dislocations. Temporary immobilization by casting/splinting alone was recommended for 15 patients (31%), and casting/splinting after a closed reduction was recommended for 17 patients (35%). Definitive treatment with external fixation was recommended for 1 patient with an open bimalleolar fracture with syndesmotic widening and significant bone loss at the medial malleolus. For the remaining 31 patients (63%), definitive treatment with open reduction and internal fixation was recommended.

Figure E3: Percent agreement in fracture classifications by Malawian OCOs and US surgeons.

Percent agreement on fracture displacement and instability was examined for all patients, the subset of patients for whom Malawian OCOs and US surgeons made the same treatment recommendation (non-operative vs. operative), and the subset of patients for whom Malawian OCOs and US surgeons made different treatment recommendations. This last subset was further subdivided into patients for whom the Malawian OCOs stated their rationale for treatment was that resources were too limited to provide the ideal treatment, and patient for whom Malawian OCOs felt they were providing the ideal treatment.

For the patients who received the same treatment recommendations by Malawian and US providers, percentage agreement regarding fracture displacement and instability was high – 97% and 88% respectively. For the 9 patients who received different recommendations because Malawian providers felt that resources were too limited, agreement on fracture displacement remained relatively high at 67%, though agreement on fracture instability dropped to 44%. For the 7 patients who received different recommendations because Malawian providers felt non-operative care was the ideal treatment, percent agreement on fracture displacement and instability was 14% and 0%, respectively. All 7 patients’ fractures were characterized as displaced and/or unstable by the US surgeons; none were reported as displaced/unstable by Malawian OCOs.

Phase 3

Table E4:

Phase 3 - Educational course participant demographics

Total	61
Title
Orthopaedic Clinical Officer	52
Orthopaedic Surgeon	3
Other (Prosthetist, Physiotherapist, Student)	6
Highest Level of Training
Diploma	46
BSc	12
MBBS	3
Place of Work
Central hospital	27
District hospital	22
Mission hospital	8
Other facilities (barracks, private clinics, health centers)	4

Sixty-one providers completed pre- and post-course assessments. 55/61 were orthopaedic providers and made up 40% of all practicing orthopaedic providers in Malawi at that time. Providers represented 31 hospitals including all four central hospitals and 16/25 district hospitals.

Figure E4: Phase 3 knowledge assessment performance by question category.

A) Average scores for entire cohort. Scores were calculated as the percentage of questions answered correctly, overall and by question category. Questions were categorized as pertaining to anatomy, injury identification, syndesmotic/deltoid injury identification, knowledge of ideal treatment principles, post-treatment weightbearing recommendations, and referral recommendations. B) Average change in performance. Change in number of questions answered correctly between the pre- and post-course assessments was calculated for each participant. Average change in performance was calculated by question category. Error bars denotes 95% confidence intervals. All changes between the pre- and post-course assessments were statistically significant (p<0.05), as calculated using paired t-tests.

Figure E5: Percentage of phase 3 participants who A) correctly identified specific ankle fracture types, and B) correctly identified the ideal treatment of each fracture type.

Pre- and post-course performance are demonstrated. The following ankle fracture types were tested: Weber A lateral malleolar, Weber B lateral malleolar, Weber C lateral malleolar, bimalleolar, lateral malleolar with increased medial clear space (bimalleolar-equivalent), ligamentous Maissonneuve (only injury identification was tested), bimalleolar fracture-dislocation, and open trimalleolar fracture-dislocation.

Greater than two-thirds of the cohort was able to correctly identify all injuries in the post-course assessment except for bimalleolar fractures which were identified correctly by 48% pre-course and 54% post-course (Figure E5A). Participants improved in their ability to identify the correct ideal treatment of all injuries except Weber B lateral malleolar fractures, for which 79% of participants identified it correctly pre-course compared to 66% post-course. In the post-course assessment, greater than two-thirds of the cohort was able to correctly identify the ideal treatment of all injuries except bimalleolar, bimalleolar-equivalent, and open trimalleolar fracture-dislocations (Figure E5B).