The Persistent Preference for Traditional Bonesetting in Ethiopia: Risk Factors and Missed Opportunities

Abstract

Introduction:

Many patients in Ethiopia prefer traditional bonesetting (TBS) over formal health services for musculoskeletal injuries, despite well-documented complications. We analyzed risk factors for initial and persistent TBS preference among patients who patronized TBS and subsequently presented to a tertiary hospital for definitive treatment.

Methods:

Over 15 months, we prospectively enrolled patients with musculoskeletal injuries who presented to 8 Ethiopian tertiary hospitals after patronizing TBS for the same injury. Multivariable regression evaluated associations between TBS preference and covariates including demographics, injury characteristics, reasons for seeking TBS, and hospitalization outcomes.

Results:

We enrolled 1,243 patients (mean age 27 ± 18.4 years; 67.6% male). Elbow (307, 24.7%) and leg/ankle/foot (245, 19.7%) injuries were most common; 978 (78.7%) were closed fractures/dislocations. At study enrollment, 1,003 (80.7%) patients had complications, most commonly stiffness/contracture (671, 54.0%) and malunion (346, 27.8%). Prolonged hospitalization (>7 days) occurred for 223 (17.9%) patients; 37 (3.0%) underwent amputation. TBS was initially preferred after injury by 808 (65.0%) patients, although 298 (24.0%) visited a health facility before TBS. Initial TBS preference was lower among patients with formal education (primary school: adjusted odds ratio [AOR] 0.64, [95% confidence interval [CI]: 0.41-0.99]; secondary school or above: AOR 0.56 [0.36-0.87]) and open fractures (AOR 0.41 [0.22-0.77]). For future injuries, 455 (36.6%) patients/guardians preferred TBS over health facility–based care. Controlling for initial TBS preference, future preference was associated with poverty (AOR 1.66, 95% CI 1.10-2.49, p = 0.015) and visiting a health facility before TBS (AOR 2.04, 95% CI 1.34-3.10, p < 0.001). Neither prolonged hospitalization (p = 0.734) nor amputation (p = 0.946) influenced future preference.

Conclusion:

More than one-third of patients preferred TBS for future injuries. Among patients in poverty, persistent TBS preference after reaching a tertiary hospital may reflect entrenched structural barriers to formal orthopaedic care. Initial health facility visits before TBS were associated with greater future TBS preference even after definitive hospital care, suggesting missed opportunities to build trust in formal care.

Level of Evidence:

Prognostic Level IV. See Instructions for Authors for a complete description of levels of evidence.

Introduction

Traditional bonesetting (TBS) is widely practiced in many low- and middle-income countries (LMICs) where access to modern orthopaedic care remains limited^1-8. Bonesetters operate outside formal health systems without regulatory oversight^9-11. TBS practices include manipulations, splinting, and application of herbal concoctions, which have been linked to serious complications including malunion, infection, and amputation^1,2,8,12-16. The drivers of TBS preference are consistent globally: accessibility, affordability, cultural trust, and dissatisfaction with formal care have been documented in sub-Saharan Africa, South and Southeast Asia, the Middle East, and Latin America^3,5-8,17-20.

Ethiopia, a low-income East African country of 132 million people, carries a high burden of musculoskeletal trauma driven by falls, road injuries, and interpersonal violence^21-26. Hospital capacity remains severely constrained, limiting formal orthopaedic access for many Ethiopians^27-30. In this context, many patients prefer TBS because it is accessible, affordable, and trusted more than formal health systems^10,31-33. Low income, rural residence, and limited education have been associated with TBS preference, yet the frequency and drivers of these preferences in Ethiopia remain unknown^10,31,34-36. Critically, persistent preference for TBS, after patients have encountered formal care, is poorly understood and may reflect missed opportunities to retain patients within formal health pathways.

The Bone-Setting Associated Disability (BOSAD) initiative is a prospective multicenter Ethiopian study of TBS practices, complications, and patient preferences. In a previous study from this initiative, rigid splinting and organic topical treatments by bonesetters independently increased the odds of amputation and prolonged hospitalization among 460 children¹². For patients of all ages who sought TBS before presenting to a tertiary hospital for definitive treatment, we analyzed risk factors for initial and persistent TBS preference.

Materials and Methods

Study Setting

This prospective cohort study was conducted at 8 Ethiopian tertiary hospitals. Hawassa University’s institutional review board approved the study (IRB/280/12). From July 2021 to September 2022, patients seen by orthopaedic providers in emergency or outpatient departments who had patronized TBS for the same injury were approached for enrollment. Informed consent was obtained from the patient or a parent/guardian, who completed enrollment surveys with the assistance of trained BOSAD data collectors fluent in the patient’s language. Medical interpreters were available when needed. We recorded demographics, socioeconomic background, injury characteristics, prior health facility visits, TBS treatments, and care preferences including reasons for seeking TBS. Patients were followed during hospitalization, with outcome and future TBS preferences collected at discharge.

Data Elements

Three outcomes were analyzed: (1) initial TBS preference at the time of injury, (2) visiting a health facility before patronizing TBS, and (3) TBS preference for a future injury. We examined 14 covariates associated with these outcomes: age, sex, highest household education level, occupation, religion, household size, poverty status, residence type, geographic region, injury mechanism, injured body part, injury type, time to presentation, and reasons for seeking TBS^{5,7,17,31,37-43}.

Age was categorized by developmentally meaningful stages: infant/toddler (0-5 years), school age (6-10 years), adolescent (11-17 years), young adult (18-35 years), middle-aged (36-55 years), and older adult (≥56 years). Household education was categorized as no education (cannot read or write), informal education (literate without formal schooling), primary school (grades 1-8), secondary school or higher, or unknown. Poverty status was based on per-capita household income relative to a threshold of 913 Ethiopian birr/month (World Bank poverty line: $2.15/day; purchasing power parity ratio 14.2)³⁵. Six geographic regions were included: Oromia; Amhara; Harari; Sidama; Southern Nations, Nationalities and Peoples’ Region; and Somali.

Injury mechanisms and body part classifications are listed in Appendix A1-2. Injury type was classified by the treating physician as closed fractures and/or dislocations, open fractures (Gustilo-Anderson classification), or nonosseous injuries. Time from injury to presentation was categorized as ≤30, 31 to 60, or >60 days. Patients ranked their top 5 reasons for seeking TBS from 10 options, grouped into 4 domains: practical reasons (faster service, cheaper service, accessibility), sociocultural reasons (social pressure, cultural beliefs, superstitious beliefs), fear of surgery/amputation, and dissatisfaction with formal care (health system dissatisfaction, mistrust of hospitals). A normalized weighted rank score was calculated per domain, with the highest-scoring domain designated the primary motivation for seeking TBS. Ties were broken by first-ranked reason.

Statistical Analysis

Multivariable models were constructed separately for each outcome using covariates with type III p < 0.1 in bivariate screening, with noncontributory covariates (p > 0.1) removed in final parsimonious models. Education, religion, poverty, geographic region, injury mechanism, injury type, and primary motivation for seeking TBS were retained a priori regardless of p-value. Multicollinearity was checked, and significance was set at p < 0.05. Post hoc descriptive secondary analyses examined reasons for seeking TBS among all patients, stratified by initial TBS preference; and among future TBS preferrers, stratified by prior health facility contact. Analysis was performed using R (version 4.3.1; R Foundation for Statistical Computing)⁴⁴.

Results

We enrolled 1,243 patients who had patronized TBS before study site arrival; the mean age was 27 ± 18.4 years; 68% were male. Ten geographic regions were represented; 67.8% (n = 843) lived below the poverty line. The median time from injury to presentation was 37 days (interquartile range 15-91). Ground-level falls (n = 383, 30.8%), falls from height (n = 316, 25.4%), and road traffic incidents (n = 208, 16.7%) caused most injuries. Injuries were mostly of the elbow/forearm/wrist (n = 622, 50.0%) and leg/ankle/foot (n = 245, 19.7%); with closed fractures (n = 978, 78.7%), open fractures (n = 77, 6.2%), and nonosseous injuries (n = 188, 15.1%). Most (93.2%) had no medical comorbidities. TBS treatments included topical application of inorganic (586, 47.1%) or organic (185, 14.9%) materials and rigid (481, 38.7%) or nonrigid (474, 38.1%) immobilization (Table I). Complications upon hospital presentation were observed in 1,003 (80.7%) patients. Prolonged hospitalization (>7 days) occurred for 223 (17.9%) patients and 37 (3.0%) underwent amputation (Table I). One patient developed fatal septicemia, arriving 10 days after sustaining a Gustilo-Anderson type IIIA open ankle fracture splinted by a bonesetter.

TABLE I.

Descriptive Statistics of Cohort (N = 1,243)

Characteristic	N (%)a	Characteristic	N (%)a
Age (years)	27.4 ± 18.4	Mechanism of injury
Age groups		Fall from ground level	383 (30.8)
0-5 years	90 (7.2)	Fall from heighti	316 (25.4)
6-10 years	227 (18.3)	Interpersonal violencej	186 (15.0)
11-17 years	143 (11.5)	Road traffic incidentk	208 (16.7)
18-35 years	401 (32.3)	Otherl	150 (12.1)
36-55 years	277 (22.3)	Injured body partm
≥ 56 years	105 (8.4)	Shoulder	100 (8.0)
Sex		Arm	42 (3.4)
Male	840 (67.6)	Elbow	307 (24.7)
Female	403 (32.4)	Forearm	146 (11.7)
Highest reported household education level		Wrist	169 (13.6)
No educationd	218 (17.5)	Hand	45 (3.6)
Informale	70 (5.6)	Hip	64 (5.1)
Primary schoolf	444 (35.7)	Thigh	59 (4.7)
Secondary schoolg	473 (38.1)	Knee	66 (5.3)
Unknownh	38 (3.1)	Leg/ankle/foot	245 (19.7)
Occupation		Injury type
Gov/private employee	228 (18.3)	Soft tissue only	188 (15.1)
Merchant	55 (4.4)	Closed fracture/dislocation	978 (78.7)
Daily laborer/farmer	258 (20.8)	Grade 1 open fracturep	26 (2.1)
Housewife	144 (11.6)	Grade 2 open fracturep	24 (1.9)
Unemployed	13 (1.0)	Grade 3 open fracturep	27 (2.2)
Student	431 (34.7)	Treatments received before study enrollment
Other	114 (9.2)	TBS topical treatments
Religion		None	472 (38.0)
Ethiopian orthodox	481 (38.7)	Organic materialr	185 (14.9)
Muslim	449 (36.1)	Inorganic materials	586 (47.1)
Protestant	294 (23.7)	TBS immobilization
Catholic	9 (0.7)	No immobilization	288 (23.2)
Other	10 (0.8)	Rigid immobilizationt	481 (38.7)
Median household size (incl. patient)	6.0 (4.0-7.0)	Other nonrigid immobilizationu	474 (38.1)
Median household income (1,000 birr)n	3.0 (2.0, 5.2)	Complications documented upon patient presentation to a study sitev
Poverty statuso		Any complication	1,003 (80.7)
Below poverty line	843 (67.8)	All complicationsw
Above poverty line	400 (32.2)	Joint stiffness/contracture	671 (54.0)
Residence type		Malunion	346 (27.8)
Rural	586 (47.1)	Muscle atrophy/weakness	264 (21.2)
Urban	657 (52.9)	Prolonged hospital stay (>7 days)	223 (17.9)
Geographic regionb		Infected skin	101 (8.1)
Amhara	249 (20)	Chronic dislocation	100 (8.0)
Benishangul-Gumuz	1 (0.1)	Skin maceration or blistering	92 (7.4)
Dire Dawa	4 (0.3)	Nonunion	83 (6.7)
Gambella	2 (0.2)	Neurovascular injury	76 (6.1)
Harari	36 (2.9)	Pressure sore	75 (6.0)
Oromia	443 (35.6)	Chronic osteomyelitis	67 (5.4)
Sidama	176 (14.2)	Amputation	37 (3.0)
SNNPRc	215 (17.3)	Delayed union	33 (2.7)
Somali	116 (9.3)	Septicemia	31 (2.5)
Tigray	1 (0.1)	Complex regional pain syndrome	27 (2.2)
Medical comorbidities		Wet gangrene	26 (2.1)
No condition recorded	1,159 (93.2)	Volkmann ischemic contracturex	24 (1.9)
Hypertension	42 (3.4)	Missed compartment syndromex	24 (1.9)
Diabetes	13 (1.0)	Septic joint	22 (1.8)
Cardiovascular disease	3 (0.2)	Avascular necrosis	18 (1.4)
Chronic kidney disease	1 (0.1)	Degloved wound	17 (1.4)
Other	25 (2.0)	Dry gangrene	6 (0.5)
Time to presentationq
≤30 days	558 (44.9)
31-60 days	234 (18.8)
>60 days	451 (36.3)

TBS = traditional bonesetting.

^aValues are given as the count, with the percentage of the total cohort in parentheses, except for age which is given as the mean ± SD, and household size and household income, which are given as the median with the Q1 and Q3 values in parentheses. Percentages may not sum to 100 because of rounding.

^bThere were no patients in this cohort from the following regions of Ethiopia: Afar, Addis Ababa.

^cSouthern Nations, Nationalities and Peoples’ Region.

^dCannot read or write, and with no formal schooling background.

^eAbility to read and write but did not attend formal grade school.

^fGrades 1-8.

^gGrades 9-12, college, technical school, professional degree, secondary degree or above.

^hIncludes Children <4 years old and those enrolled in nursery school or kindergarten, and for whom no other information for the other family members was provided.

ⁱFalls from height were from higher than ground level.

^jIncludes injury from a fight or assault, a bullet wound, a blast injury, or a stab injury.

^kIncludes a pedestrian struck by a motorcycle or motor vehicle, a driver of a motorcycle or motor vehicle involved in a collision, or a passenger of a motorcycle or motor vehicle involved in a collision.

^lIncludes a sports injury, a bicycle accident, a construction-site injury, or other mechanisms of injury.

^mBody part categories were condensed to address small sample sizes.

ⁿA median monthly income of 3,000 birr is approximately $210, converted using the 2022 World Bank purchasing power parity ratio for Ethiopia of 14.2 Ethiopian birr per US dollar.

^oPoverty status was assessed by calculating the income per person in each patient’s household on the basis of the total household income and the number of household members reported. The poverty line was set at 913 birr per person per month, equivalent to the 2022 World Bank poverty level of $2.15 per day and converted using the 2022 World Bank purchasing power parity ratio for Ethiopia of 14.2 Ethiopian birr per US dollar.

^pGustilo-Anderson open fracture classification.

^qTime from initial injury to arrival at one of the 8 study site hospitals.

^rOrganic topical materials that had been applied included mud, leaves or other plant materials, and butter.

^sInorganic topical materials included petroleum jelly or some other chemical or drug.

^tRigid immobilization included use of bamboo, wooden, or metal splints.

^uOther nonrigid immobilizations included wrapping with a towel, leather splint, or some other soft dressing.

^vComplications documented upon patient presentation to a study site are ordered by decreasing overall frequency.

^wPercentages do not sum to 100 because of patients having multiple complications.

^xMissed compartment syndrome refers to a late-presenting compartment syndrome with tissue damage but not yet having progressed to a Volkmann ischemic contracture.

Most patients or their guardians (808, 65.0%) reported an initial preference for TBS at the time of injury, 298 (24.0%) visited a health facility before TBS, and 455 (36.6%) preferred TBS over hospital-based care for a future injury. The most common reasons for seeking TBS were easy accessibility (n = 808, 65.0%), social pressure (n = 733, 58.9%), and faster service (n = 526, 42.3%) (Table II).

TABLE II.

Traditional Bonesetter Preferences and Reported Reasons for Those Preferences

Characteristic	N (%)*
Initial TBS preference at the time of injury†	808 (65.0)
Visited a health facility first before TBS†	298 (24.0)
Prefers TBS for a future injury	455 (36.6)
Why did you choose TBS for this injury?
Easily accessible	808 (65.0)
Social pressure	733 (59.0)
Faster service	526 (42.3)
Cultural beliefs	502 (40.4)
Cheaper service	391 (31.5)
Fear of surgery	213 (17.1)
Fear of amputation	198 (15.9)
Health system dissatisfaction	116 (9.3)
Superstitions/religious beliefs	69 (5.6)
Healthcare worker ignorance	12 (1.0)

TBS = traditional bonesetting.

^*Percentages do not sum to 100 because of rounding.

^†Categories are not mutually exclusive.

Bivariate and multivariable analyses are presented in the Appendix B1-3. Adjusted odds of initial TBS preference were higher for patients aged 11 to 17 years (adjusted odds ratio [AOR]: 1.63, [95% confidence interval: 1.01-2.63]), 36 to 55 years (AOR 1.60 [1.09-2.35]), with interpersonal violence (AOR 1.59 [1.01-2.50]) or other injury mechanisms (AOR 4.26 [2.46-7.39]). Primary (AOR 0.64 [0.41-0.99]) and secondary education or above (AOR 0.56 [0.36-0.87]), Muslim religion (AOR 0.51 [0.35-0.76]), and open fracture (AOR 0.41 [0.22-0.77]) were associated with lower odds. Relative to practical reasons for seeking TBS, patients with sociocultural reasons had increased odds (AOR 1.40 [1.03-1.90]) and dissatisfaction with formal care had reduced odds (AOR 0.45 [0.21-0.96]) of initial TBS preference (Fig. 1). Patients who did not initially prefer TBS more frequently cited health system dissatisfaction and quicker service as reasons for seeking TBS, compared to those initially preferring TBS (Appendix C1).

Fig. 1.

Risk factors for initial preference for traditional bonesetting. Forest plot constructed from the final parsimonious multivariable model (see Methods). Religion: the “Other” category includes 1 patient recorded as Catholic and 2 recorded as “other.” Poverty status was defined using household per capita monthly income relative to the poverty line of 913 Ethiopian birr per person per month, equivalent to US$2.15 per day using the 2022 World Bank purchasing power parity conversion factor for Ethiopia. Geographic region: Oromia includes 2 patients from nearby Gambella; Amhara includes 1 patient from nearby Benishangul-Gumuz and 1 from nearby Tigray; SNNPR; Somali includes 1 patient from nearby Dire Dawa. Mechanism of injury: the “Other” category includes bicycle accident, construction site injury, and other mechanisms. Primary motivation for seeking TBS: derived from patient-ranked reasons grouped into 4 domains: practical reasons (accessibility, affordability, timeliness), sociocultural reasons (cultural, superstitious, or social beliefs), fear of surgery or amputation, and dissatisfaction with formal care. The domain with the highest normalized weighted rank score was designated each patient's primary motivation; ties were broken by first-ranked reason. Estimates for subgroups with small sample sizes should be interpreted with caution given wide confidence intervals. SNNPR = Southern Nations, Nationalities and Peoples’ Region, and TBS = traditional bonesetting.

Visiting a health facility before TBS was less likely for children aged 0 to 5 years (AOR 0.39 [0.17-0.89]) and women (AOR 0.71 [0.51-1.00]). Muslim patients had higher odds of visiting a health facility first (AOR 1.60 [1.01-2.54]). Higher-energy injury mechanisms (fall from height: AOR 1.76 [1.15-2.69]; interpersonal violence: AOR 1.88 [1.17-3.03]; motor vehicle collision: AOR 1.86 [1.17-2.97]), leg/ankle/foot injuries (AOR 1.94 [1.21-3.10]), and forearm/hand/wrist injuries (AOR 1.63 [1.05-2.53]) were associated with higher odds of visiting a health facility first. Dissatisfaction with formal care, relative to practical reasons for seeking TBS, was associated with visiting a health facility before TBS (AOR 4.54 [2.17-9.48]) (Fig. 2).

Fig. 2.

Factors associated with visiting a health facility before traditional bonesetting. Forest plot constructed from the final parsimonious multivariable model (see Methods). Religion: the “Other” category includes 1 patient recorded as Catholic and 2 recorded as “other.” Poverty status was defined using household per capita monthly income relative to the poverty line of 913 Ethiopian birr per person per month, equivalent to US$2.15 per day using the 2022 World Bank purchasing power parity conversion factor for Ethiopia. Geographic region: Oromia includes 2 patients from nearby Gambella; Amhara includes 1 patient from nearby Benishangul-Gumuz and 1 from nearby Tigray; SNNPR; Somali includes 1 patient from nearby Dire Dawa. Mechanism of injury: the “Other” category includes bicycle accident, construction site injury, and other mechanisms. Body part categories were condensed because of small sample sizes. Primary motivation for seeking TBS: derived from patient-ranked reasons grouped into 4 domains: practical reasons (accessibility, affordability, timeliness), sociocultural reasons (cultural, superstitious, or social beliefs), fear of surgery or amputation, and dissatisfaction with formal care. The domain with the highest normalized weighted rank score was designated each patient's primary motivation; ties were broken by first-ranked reason. Estimates for subgroups with small sample sizes should be interpreted with caution given wide confidence intervals. SNNPR = Southern Nations, Nationalities and Peoples’ Region, and TBS = traditional bonesetting.

Future TBS preference was higher among patients living in poverty (AOR 1.66 [1.10-2.49]) and lower among rural dwellers (AOR 0.56 [0.38-0.83]). Initial TBS preference (AOR 66.09 [34.56-126.39]) and visiting a health facility before TBS (AOR 2.04 [1.34-3.10]) increased future preference odds. Neither prolonged hospitalization (AOR 1.05 [0.65-1.71]) nor amputation (AOR 0.97 [0.37-2.50]) influenced future preference (Fig. 3). Among future TBS preferrers (n = 455), patients who visited a health facility first more frequently cited health system dissatisfaction and cultural beliefs as reasons for seeking TBS, while those who went directly to TBS more commonly cited quicker service and social pressure (Appendix C2). Interaction analyses between religion and geographic region are presented in the Appendix D1-3.

Fig. 3.

Risk factors for future preference for traditional bonesetting. Forest plot constructed from the final parsimonious multivariable model (see Methods). Religion: the “Other” category includes 1 patient recorded as Catholic and 2 recorded as “other.” Poverty status was defined using household per capita monthly income relative to the poverty line of 913 Ethiopian birr per person per month, equivalent to US$2.15 per day using the 2022 World Bank purchasing power parity conversion factor for Ethiopia. Geographic region: Oromia includes 2 patients from nearby Gambella; Amhara includes 1 patient from nearby Benishangul-Gumuz and 1 from nearby Tigray; SNNPR; Somali includes 1 patient from nearby Dire Dawa. Mechanism of injury: the “Other” category includes bicycle accident, construction site injury, and other mechanisms. HF = health facility; prolonged stay = hospitalization at a Bone-Setting Associated Disability study center for >7 days. Primary motivation for seeking TBS: derived from patient-ranked reasons grouped into 4 domains: practical reasons (accessibility, affordability, timeliness), sociocultural reasons (cultural, superstitious, or social beliefs), fear of surgery or amputation, and dissatisfaction with formal care. The domain with the highest normalized weighted rank score was designated each patient's primary motivation; ties were broken by first-ranked reason. Estimates for subgroups with small sample sizes should be interpreted with caution given wide confidence intervals. SNNPR = Southern Nations, Nationalities and Peoples’ Region, and TBS = traditional bonesetting.

Discussion

Among 1,243 patients who patronized TBS and subsequently presented to an Ethiopian referral hospital for definitive treatment, 37% still preferred TBS for a future injury. This preference was not attenuated by prolonged hospitalization or amputation. Persistent TBS preference was most pronounced among patients living in poverty, those with an initial TBS preference, and those who had visited a hospital before TBS.

Building on our prior pediatric analysis¹², this cohort's demographics and injury patterns were consistent with prior Ethiopian trauma reports^21-26,45,46. Complications in 81% of patients, including malunion, stiffness, and amputation, commonly occur with delayed fracture care, and are widely reported in TBS-treated cohorts across LMICs^1,2,8,12-16. TBS-related delays or treatments may necessitate amputations preventable with timely care¹⁶, and amputation is devastating in Ethiopia given limited prosthetics and rehabilitation resources^47-49. While TBS-attributable morbidity cannot be established without a comparator, specific TBS practices independently increased odds of amputation and prolonged hospitalization in a prior pediatric analysis from this initiative¹². Engaging TBS providers to avoid dangerous practices and facilitate timely referral may reduce preventable complications.

Practical reasons favoring TBS, including perceived accessibility, affordability, and timeliness, were reported by 55% of patients, consistent with prior reports across LMICs^{27,31,41,42,50-54}. Sociocultural factors, including cultural attribution of injuries to supernatural causes and social influence, were associated with 40% higher odds of initial TBS preference^{31,36,41,53,54}. Muslim patients had lower odds of initial TBS preference, consistent with evidence that religion shapes traditional medicine utilization in Ethiopia^55-57. Fear of hospital-based care also influences TBS preference^{31,42,43,53,54}. While fear of amputation and surgery drove initial TBS use in our cohort, sustaining an amputation was not associated with future TBS preference. The persistence of TBS preference despite adverse outcomes underscores the need to understand its underlying drivers.

Even after definitive tertiary hospital care, patients with prior health facility contact had twice the odds of future TBS preference compared with those who went directly to TBS. Negative first-contact experiences may underlie this association. Dissatisfaction with formal care was associated with lower odds of initial TBS preference, suggesting some patients sought TBS secondarily after an unsatisfactory healthcare encounter. Dissatisfaction was associated with nearly 5 times the odds of prior health facility contact. The association between health system dissatisfaction and TBS use has been reported in Ethiopia and across LMICs^{31,34-36,40-42}. The initial health facility visit may represent a missed opportunity to build trust in formal care and retain patients who did not originally prefer TBS. Improving first-contact care quality and cultural responsiveness warrants prospective investigation.

Prior studies across LMICs have identified low income, rural residence, and limited education as predictors of TBS preference^34,41,44. Consistent with these reports, formal education reduced the odds of initial TBS preference by 40% to 45%, and poverty increased the odds of future TBS preference by approximately 70%^41,42. Despite higher reported TBS patronage among rural patients³⁷, rural residence was associated with lower odds of future TBS preference, suggesting preferences may shift after overcoming geographic barriers to reach a tertiary hospital. Persistent TBS preference among patients in poverty may reflect entrenched structural barriers to accessing formal care. Strategies to improve orthopaedic care access for patients in poverty warrant further investigation.

TBS remains prevalent and trusted throughout Ethiopia; this cohort represents only those who subsequently sought tertiary hospital care. Educational programs guiding bonesetters toward safe management of lower-acuity injuries and referral of complex cases may reduce preventable complications¹⁰. A systematic review identified broad stakeholder support for TBS-formal healthcare integration, with training programs as the most feasible entry point⁵⁸. Bonesetters have shown receptivity to formal training, with structured programs reducing complications, expanding access to care, and providing a framework for oversight^9,16,38. Government registration of bonesetters with a defined scope of practice and referral requirements could reduce preventable complications while preserving accessibility^9,10. A bonesetter training program has been implemented as part of the BOSAD initiative and outcomes will be reported in future.

This study had several limitations. First, injury characteristics, TBS treatment methods, and complications were assessed at study enrollment and are subject to recall bias. Second, our cohort excluded patients who patronized TBS without subsequent hospital presentation and those who did not survive hospital transfer, potentially introducing sampling bias. Third, the absence of a non–TBS-seeking comparator group limited causal attribution of complications to TBS. Future matched cohort studies to establish TBS-attributable morbidity are planned within the BOSAD initiative. Finally, the study was limited to large Ethiopian academic hospitals, which may limit generalizability to smaller remote facilities.

Despite these limitations, this multicenter study provides the largest assessment of TBS preferences among patients who patronized TBS before tertiary hospital presentation. These findings may be relevant to other resource-limited settings where TBS remains common. More than one-third of patients persistently preferred TBS for future injuries. Two distinct drivers appear to underlie this preference. Among patients with prior health facility contact, negative first-contact experiences may have eroded trust in formal care. Among patients in poverty, entrenched structural barriers may limit access to it. Further investigation into both is needed to improve orthopaedic care delivery in Ethiopia.

Funding

The AO Alliance Foundation funded this study.

Appendix

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (https://links.lww.com/JBJSOA/B303). This content was not copyedited or verified by JBJS.