Fracture laterality as an independent prognostic factor for clinical outcomes after hip fracture surgery Authors: Yasuhiko Takegami, Takahiro Imaizumi, Kikue Sato, Yusuke Osawa, Hiroto Funahashi, Shiro Imagama Published: 2026 ABSTRACT AbstractAimsFracture laterality is routinely recorded in large hip fracture registries for administrative purposes, yet rarely evaluated as a prognostic variable. We aimed to test whether laterality is independently associated with short-term outcomes after hip fracture surgery.MethodsThis retrospective cohort study used a regional, prefecture-wide administrative database based on Japan’s Diagnosis Procedure Combination system. We identified 72,008 patients who underwent primary surgery for hip fracture between 2015 and 2023. The primary outcome was in-hospital mortality. The secondary outcome was impaired independence in activities of daily living (ADL) at discharge, defined as a Barthel index (BI) score < 60. We used multivariable logistic regression to assess the association between fracture laterality and outcomes, adjusting for age, sex, BMI, smoking history, fracture type, admission BI, diabetes mellitus, and the Charlson Comorbidity Index (CCI).ResultsLeft-sided fractures were significantly more frequent than right-sided fractures (51.5% vs 48.5%, p < 0.001). In-hospital mortality did not differ significantly between right- and left-sided fractures in either unadjusted (2.0% vs 2.0%, p = 0.093) or adjusted analyses (adjusted odds ratio (aOR), 1.10; 95% CI 0.99 to 1.22; p = 0.075). In contrast, right-sided fracture was independently associated with impaired ADL at discharge (aOR 1.09; 95% CI 1.05 to 1.13; p < 0.001). In a prespecified sub-group analysis, right-sided fracture was associated with higher mortality among patients with high comorbidity burden (CCI ≥ 2; aOR 1.29; 95% CI 1.09 to 1.53; p = 0.004; p for interaction = 0.022). A sensitivity analysis using double-selection lasso logistic regression yielded materially unchanged estimates.ConclusionIn our cohort, right-sided hip fractures were independently associated with impaired ADL at discharge, and with higher in-hospital mortality among patients with substantial comorbidity. These associations persisted across sensitivity analyses but, given the observational design and the inability of administrative data to capture limb dominance, require confirmation in independent cohorts. Fracture laterality, which is a variable routinely captured at presentation but seldom analyzed, nonetheless merits consistent reporting and consideration in registry-based prognostic research.Cite this article: Bone Jt Open 2026;7(8):1064–1074.Keywords: Hip fracture, Laterality, Functional outcome, Mortality, Large cohort study, Administrative database, hip fracture surgery, clinical outcomes, hip fractures, comorbidities, Charlson Comorbidity Index (CCI), BMI, Multivariable logistic regression analysis, diabetes mellitus, logistic regression, retrospective cohort study Introduction Among osteoporotic injuries, hip fractures, encompassing both femoral neck and trochanteric fractures, cause the greatest disability, excess mortality, and socioeconomic burden worldwide.1-3 In Japan, where the population is ageing rapidly, the annual number of hip fractures is projected to increase significantly, highlighting the urgent need to identify all factors that may influence clinical outcomes and optimize management strategies.4 Recent large prospective cohort studies have advanced the understanding of prognostic factors after hip fracture,5,6 and a linked administrative data analysis has similarly expanded analytic possibilities in fracture outcome research more broadly.7 International registries are increasingly adopting standardized datasets for benchmarking and quality improvement. The Fragility Fracture Network Core 2022 Minimum Common Dataset includes fracture side as a core field,8 and laterality is also routinely captured in administrative databases for record linkage. This routine collection is biologically plausible because laterality has been shown to affect multiple aspects of musculoskeletal health and surgical practice. At the surgical level, failure of sliding hip screw fixation for femoral neck fractures has been reported to occur more frequently on the left side in departments staffed predominantly by a right-handed surgeon.9 At the skeletal level, studies measuring bilateral hip bone mineral density in post-menopausal females have reported modest but measurable right–left differences.10,11 At the behavioural level, population studies indicate that the majority of individuals are right-foot dominant,12 which may influence fall direction, balance-recovery stepping, and early rehabilitation tasks that depend on the injured side. Despite this plausibility and the near-universal availability of laterality data, prognostic analyses after hip fracture surgery have rarely evaluated laterality as an independent predictor, and most models implicitly treat fracture side as non-informative. Two practical knowledge gaps remain in large, population-based surgical cohorts. First, the laterality distribution of surgically treated hip fractures has not been well characterized in Asian populations. Second, it remains unclear whether fracture laterality independently predicts postoperative outcomes after accounting for key patient factors, fracture type, and comorbidity burden. To address these gaps, we analyzed a prefecture-wide Japanese administrative database based on the Japan’s Diagnosis Procedure Combination Per-Diem Payment System (DPC/PDPS), including more than 70,000 surgically treated hip fracture cases. Specifically, we aimed to quantify the distribution of left- versus right-sided fractures in this surgical cohort, and estimate the independent association of laterality with in-hospital mortality and discharge functional status, defined as impaired activities of daily living (ADL) at discharge (Barthel index (BI) < 60).13,14 Methods Data source This retrospective cohort study used a large-scale administrative claims database derived from Japan’s DPC/PDPS system. DPC/PDPS is a prospective reimbursement framework for acute-care hospitalizations, with participating facilities nationwide. Submission of DPC data is mandatory for participating hospitals and is linked to reimbursement, which provides an incentive for accurate coding of diagnoses and procedures. For the present analysis, we restricted the data source to a regional consortium of 172 acute-care academic and community hospitals in Aichi Prefecture, and extracted anonymized inpatient episode-level records. The DPC/PDPS database comprises several standardized files. For this study, we primarily used the FF1-file (patient-level discharge abstract) to ascertain hospital identifiers, patient demographic details (age, sex, and BMI at admission), smoking history, and discharge status, including in-hospital mortality and BI scores. We also used the E/F-file (fee-for-service claims for all billable items) to identify surgical procedures and medication use.15 Diagnoses are recorded using the International Classification of Diseases of the World Health Organization, 10th Revision (ICD-10),16 and surgical interventions are coded using Japanese K-codes.17 Importantly, the database contains fields for primary and secondary diagnoses (enabling identification of fracture type and laterality), complete procedure records (to define surgically treated cases), and discharge outcomes (including mortality and independence in ADL at discharge). The validity and reliability of key diagnostic and procedural information in DPC/PDPS have been demonstrated in prior research.18,19 All data were fully anonymized prior to the analysis. The study protocol was approved by the Institutional Review Board of Nagoya University (Ref. 2024-0075), which waived the requirement for individual patient informed consent due to the anonymized nature of the data. Study population We identified eligible admissions in the DPC/PDPS database between 1 January 2015 and 31 December 2023. We included patients aged ≥ 65 years with a primary diagnosis of hip fracture (femoral neck fracture (ICD-10: S72.0), trochanteric fracture (ICD-10: S72.1), or sub-trochanteric fracture (ICD-10: S72.2)) and who underwent surgical treatment during the same admission. The analytic cohort was defined as index surgical admissions for a unilateral HF with reliable coding; thus, we excluded admissions with multiple fractures (n = 3,562), osteoarthritis (n = 67), or procedure-code inconsistency (n = 19). We also excluded cases in which laterality was coded as both-sided or was not identifiable (n = 2,568). After these exclusions, the final analytic cohort included 72,008 surgically treated hip fracture patients. For analyses of the functional outcome, we further restricted the cohort to patients with a recorded BI score at discharge (70,994/72,008); 1,014 patients had missing BI data (Figure 1). Patient selection flowchart. HF, hip fracture. Patient characteristics The mean age was 84 years (SD 8) in both groups, and approximately three-quarters of patients were female (left hip 28,227 (76%), right hip 26,893 (77%)). Study outcomes The primary outcome for this study was in-hospital mortality, which was ascertained directly from the DPC/PDPS discharge abstract (FF1-file) and defined as a discharge status recorded as ‘death’ occurring during the index surgical admission. The secondary outcome was impaired independence in ADL at discharge. ADL was assessed using the BI recorded in the DPC/PDPS dataset at the time of discharge. The BI is a standardized 100-point scale evaluating performance in ten basic ADLs, with lower scores indicating greater dependence.13 Impaired ADL was defined as a discharge BI score below 60. This threshold has been widely used in geriatric and rehabilitation research to operationalize clinically meaningful dependence, and marks the boundary between severe and moderate dependence.20 Other postoperative outcomes We assessed additional major postoperative clinical events captured in the DPC/PDPS data, including intensive care unit admission and in-hospital contralateral HF. To better capture clinically meaningful postoperative infectious complications, we also used medication claims as a surrogate indicator. Because administrative datasets may underestimate infections such as pneumonia, urinary tract infection, or cholecystitis when events are defined solely by ICD-10 diagnosis codes,19 we defined postoperative infection as the administration of systemic antibiotics. Specifically, we identified billable fee-for-service items (E/F-file) for systemic antibiotics administered during the postoperative period, excluding routine single-dose perioperative prophylaxis.21 Literature audit of laterality reporting We conducted a targeted PubMed audit to determine how often fracture laterality is reported or analyzed in hip fracture outcome studies. We searched for registry-based, multicentre retrospective, and administrative database studies published between January 2020 and December 2025 (full search strategy in the Supplementary Material). For each included study, we reviewed baseline characteristic tables and regression covariate lists to assess whether laterality was reported and/or included in prognostic models. Statistical analysis All analyses were conducted to evaluate the difference in incidence by laterality, and the association between fracture laterality and postoperative outcomes. Baseline characteristics were summarized by fracture laterality (left vs right). Continuous variables (e.g. age, BMI) were reported as mean (SD) or median (IQR), as appropriate, and categorical variables (e.g. sex, fracture type, comorbidities) were reported as counts and percentages (n, %). Between-group differences were assessed using the Mann–Whitney U test for continuous variables and the chi-squared test for categorical variables. The overall distribution of left- compared with right-sided fractures was evaluated using a chi-squared goodness-of-fit test against an expected 50:50 distribution. To address missing data, multiple imputations by chained equations were employed. The covariates included in our imputation models were baseline demographic details and outcomes.22 We performed chained equations with 20 imputations and then combined the results across the 20 imputed datasets using Rubin’s formula.23 Primary analysis We evaluated the association between fracture laterality and postoperative outcomes using multivariable logistic regression. We fitted one model for in-hospital mortality in the full analytic cohort (n = 72,008) and one model for impaired ADL at discharge (BI < 60) among patients with BI recorded at discharge (n = 70,994). In both models, left-sided fracture was the reference category, and results were reported as adjusted odds ratios (aORs) with 95% CIs. Covariates were selected before analysis on the basis of clinical relevance as established prognostic factors for mortality and functional outcomes after hip fracture, and were retained in the final models regardless of their individual statistical significance; no stepwise or automated variable selection was performed. The models were specified to estimate the association between laterality and outcomes, not to develop a parsimonious prediction model. The models were adjusted for: age at admission (continuous), sex, BMI at admission (continuous), smoking history (current or past), fracture type (femoral neck, trochanteric, sub-trochanteric), BI at admission (0 to 100, continuous), diabetes mellitus, and Charlson Comorbidity Index (CCI; continuous). The CCI was calculated using a validated ICD-10 coding algorithm as implemented in the Stata user program, Charlson.24,25 To assess whether the association between laterality and outcomes differed acrss clinical sub-groups, we repeated the primary logistic regression analyses, stratifying the cohort by six prespecified patient sub-groups: sex (male vs female), age group (< 75 years vs ≥ 75 years), fracture type (femoral neck vs trochanteric vs sub-trochanteric), BMI category (< 18.5 kg/m2 vs 18.5 kg/m2 to < 25 kg/m2 vs ≥ 25 kg/m2), diabetes mellitus (DM; non-DM vs DM), and CCI category (0 vs 1 vs ≥ 2). For CCI sub-groups, the index was categorized as 0, 1, or ≥ 2. To assess the robustness of the primary findings to alternative modelling approaches, we fitted a double-selection lasso logistic model,26 with laterality specified as the variable of interest and the remaining covariates treated as candidate controls, for both mortality and impaired ADL. To quantify the sensitivity of our findings to potential unmeasured confounding, we calculated E-values for the primary association estimates.27 All statistical tests were two-sided, and p-values < 0.05 were considered statistically significant. All analyses were performed using Stata/MP (v. 19.0, StataCorp, USA). Results Patient characteristics and fracture incidence Among 72,008 surgically treated HF patients included in the analytic cohort, 37,121 (51.5%) sustained a left-sided fracture and 34,887 (48.5%) sustained a right-sided fracture. Fracture type differed by laterality: femoral neck fractures accounted for 49% of left-sided cases (n = 18,081) compared with 52% of right-sided cases (n = 18,164), whereas trochanteric fractures accounted for 49% (n = 18,284) compared with 46% (n = 15,976). Baseline patient and fracture characteristics are presented in Table I, with the admission Barthel Index presented separately in Table II. Table I.Demographic data.CharacteristicLeftRightp-valuePatients37,12134,887Mean age, yrs (SD)84 (8)84 (8)0.003* Sex, n (%) < 0.001†Male8,894 (23)7,904 (23)Female28,227 (76)26,893 (77)Mean BMI, kg/m2 (SD)21 (4)21 (4)0.792*Missing, n (%)2,380 (5)2,282 (5) Fracture type (ICD-10 code), n (%) Femoral neck fracture (S7200)18,081 (49)18,164 (52)< 0.001†Trochanteric fracture (S7210)18,284 (49)15,976 (46)Sub-trochanteric fracture (S7220)756 (2)747 (2)Missing, n (%)00 CCI total, n (%) < 0.001†017,336 (47)15,974 (46)111,380 (31)10,930 (31)> 28,303 (22)7,904 (23)Missing, n (%)102 (0.2)79 (0.2)Diabetes mellitus, n (%)6,939 (19)6418 (18)0.309†Missing, n (%)00Chronic kidney disease, n (%)1,467 (4)1,404 (4)0.632†Missing, n (%)102 (0.2)79 (0.2)Dialysis, n (%)171 (0)164 (0)0.851†Missing, n (%)00Chronic obstructive pulmonary disease, n (%)1,071 (3)978 (3)0.499†Missing, n (%)102 (0.2)79 (0.2)Smoking, n (%)4,022 (13)3,605 (12)0.028†Missing, n (%)5386 (14)5065 (14)*Mann–Whitney U test.†Chi-squared test.CCI, Charlson Comorbidity Index; ICD-10, International Classification of Diseases, 10th revision. Demographic data. Mann–Whitney U test. Chi-squared test. CCI, Charlson Comorbidity Index; ICD-10, International Classification of Diseases, 10th revision. Table II.Bartel index on admission.Characteristic, n (%)LeftRightp-value* Feeding 0.258014,433 (39)13,763 (40)513,608 (37)12,616 (36)108,937 (24)8,390 (24) Transfer 0.042027,374 (74)25,545 (73)52,620 (7)2,461 (7)102,972 (8)2,999 (9)154,012 (11)3,764 (11) Grooming 0.451030,846 (83)29,076 (84)56,132 (17)5,693 (16) Toilet use 0.033028,713 (78)26,830 (77)53,730 (10)3,714 (11)104,535 (12)4,225 (12) Bathing 0.205032,861 (89)31,001 (89)54,117 (11)3,768 (11) Mobility 0.113030,331 (82)28,389 (82)51,078 (3)1,070 (3)101,909 (5)1,912 (5)153,660 (10)3,398 (10) Stairs 0.043031,526 (85)29,578 (85)51,998 (5)2,020 (6)103,454 (9)3,171 (9) Dressing 0.452028,178 (76)26,418 (76)54,529 (12)4,365 (13)104,271 (12)3,986 (11) Bowel control 0.549025,303 (68)23,775 (68)54,115 (11)3,951 (11)107,560 (20)7,043 (20) Bladder control 0.909025,579 (69)24,072 (69)54,068 (11)3,845 (11)107,331 (20)6,852 (20)*Chi-squared test. Bartel index on admission. Chi-squared test. The mean age was 84 years (SD 8) in both groups, and approximately three-quarters of patients were female (left hip 76% (n = 28,227), right hip 77% (n = 26,893)). Clinical outcomes (unadjusted) Unadjusted postoperative clinical outcomes are summarized in Table III. In-hospital mortality was 2% in both the left- and right-sided fracture groups (p = 0.093). Intensive care unit admission was similarly uncommon (1% vs 1%; p = 0.060), and the rates of in-hospital contralateral HF (3% vs 3%; p = 0.821) and systemic antibiotic use (37% vs 38%; p = 0.787) were comparable between groups. Table III.Postoperative clinical outcomes and complications.OutcomeTotal (n = 72,008)Left (n = 37,121)Right (n = 34,887)p-value*In-hospital mortality, n (%)1,476 (2)729 (2)747 (2)0.093Intensive care unit admission, n (%)820 (1)396 (1)424 (1)0.060Contralateral hip fracture (in-hospital), n (%)1,813 (3)930 (3)883 (3)0.826Antibiotic use (systemic), n (%)27,013 (38)13,908 (37)13,105 (38)0.787*Chi-squared test. Postoperative clinical outcomes and complications. Chi-squared test. ADL outcomes at discharge (unadjusted) BI item-level distributions at discharge are shown in Table IV. Across multiple ADL domains, patients with right-sided fractures consistently showed a higher proportion of complete dependence (score 0) than those with left-sided fractures, suggesting worse independence in ADL at discharge. This pattern was most evident for core mobility-related activities (e.g. transfers, mobility, and stair climbing) and for self-care items such as grooming, toileting, and bathing (Table IV). Table IV.Barthel index at discharge.Barthel index itemsLeft(n = 37,121)Right(n = 34,887)p-value*Missing, n (%)821 (2.2)753 (2.2) Feeding 04,497 (12)4,347 (13)0.011514,436 (40)13,841 (41)1017,367 (48)15,946 (47) Transfer 06,261 (17)5,994 (18)<0.00154,293 (12)3,988 (12)1015,455 (43)14,949 (44)1510,291 (28)9,203 (27) Grooming 022,677 (62)21,800 (64)0.0001513,623 (38)12,334 (36) Toilet use 012,094 (33)11,527 (34)< 0.001512,861 (35)12,510 (37)1011,345 (31)10,097 (30) Bathing 029,007 (80)27,767 (81)< 0.00157,293 (20)6,367 (19) Mobility 015,217 (42)14,618 (43)< 0.00155,094 (14)4,840 (14)107,897 (22)7,488 (22)158,092 (22)7,188 (21) Stairs 022,973 (63)21,949 (64)0.00257,528 (21)7,038 (21)105,799 (16)5,147 (15) Dressing 013,130 (36)12,715 (37)< 0.001513,907 (38)13,176 (39)109,263 (26)8,243 (24) Bowel control 011,934 (33)11,460 (34)< 0.00158,259 (23)8,193 (24)1016,107 (44)14,481 (42) Bladder control 012,134 (33)11,692 (34)< 0.00158,553 (24)8,397 (25)1015,613 (43)14,045 (41)*Chi-squared test. Barthel index at discharge. Chi-squared test. Multivariable analysis The results of the multivariable logistic regression analyses for both primary and secondary outcomes are presented in Figure 2. Sub-group analyses for in-hospital mortality and impaired activities of daily living (ADL). a) In-hospital mortality (n = 72,008). b) Impaired ADL (n = 70,994). CCI, Charlson Comorbidity Index; DM, diabetes mellitus; OR, odds ratio. For the primary outcome of in-hospital mortality (Figure 2a), right-sided fracture was not significantly associated with higher odds of mortality relative to left-sided fracture, in either the unadjusted comparison (2.0% vs 2.0%, p = 0.093) or the adjusted analysis (n = 72,008; aOR 1.10; 95% CI 0.99 to 1.22; p = 0.075). Other covariates independently associated with higher in-hospital mortality included older age (aOR 1.06), male sex (aOR 2.83), trochanteric and sub-trochanteric fracture types (aOR 1.34 and 2.34, respectively, vs femoral neck fracture), and higher comorbidity burden (aOR 1.39). Conversely, a higher BMI (aOR 0.89) and higher BI at admission (aOR 0.99) were associated with lower odds of mortality. These findings are summarized in Table V. Table V.Multivariable logistic regression analysis for all-cause death.VariableOdds ratio95% CIp-valueRight (vs left)1.0993870.990347 to 1.2204320.075Age at admission, yrs1.0646461.056119 to 1.073241< 0.001Sex: male (vs female)2.8260942.515761 to 3.174709< 0.001BMI at admission0.8925770.876699 to 0.908743< 0.001Smoking (current or past)1.0192070.866478 to 1.1988570.818 Fracture type Femoral neck fracture (S7200)ReferenceTrochanteric fracture (S7210)1.3374231.197674 to 1.493478< 0.001Sub-trochanteric fracture (S7220)2.3381691.732146 to 3.156221< 0.001Barthel index (at admission, 0 to 100)0.9974350.995549 to 0.9993240.008Diabetes mellitus0.7336310.633268 to 0.8499< 0.001Charlson Comorbidity Index1.3862491.336341 to 1.438021< 0.001 Multivariable logistic regression analysis for all-cause death. For the secondary outcome of impaired ADL at discharge (BI < 60) (Figure 2b), right-sided fracture was independently associated with higher odds of impaired ADL relative to left-sided fracture (n = 70,994; aOR 1.09; 95% CI 1.05 to 1.13; p < 0.001). Other covariates associated with impaired ADL, including older age (aOR 1.09), male sex (aOR 1.39), trochanteric and sub-trochanteric fracture types (aOR 1.45 and 1.61, respectively, vs femoral neck fracture), and higher comorbidity burden (as assessed by the CCI; aOR 1.24), are summarized in Table VI. Table VI.Multivariable logistic regression analysis for functional non-independence (Barthel index < 60) at discharge.VariableOdds ratio95% CIp-valueRight (vs left)1.0913511.054355 to 1.129645< 0.001Age at admission, yrs1.0863411.083656 to 1.089033< 0.001Sex: male (vs female)1.347421.288524 to 1.409008< 0.001BMI at admission0.9578790.95306 to 0.962722< 0.001Smoking (current or past)0.8759670.825702 to 0.929293< 0.001 Fracture type Femoral neck fracture (S7200)ReferenceTrochanteric fracture (S7210)1.4346771.384257 to 1.486934< 0.001Sub-trochanteric fracture (S7220)1.562991.377727 to 1.773166< 0.001Barthel index (at admission, 0 to 100)0.9809160.98037 to 0.981463< 0.001Diabetes mellitus0.7740420.73644 to 0.813565< 0.001Charlson Comorbidity Index1.2398671.216866 to 1.263303< 0.001 Multivariable logistic regression analysis for functional non-independence (Barthel index < 60) at discharge. Sub-group analyses are presented in Figure 2 and Supplementary Tables i to ii. For in-hospital mortality (Figure 2a; Supplementary Table i), the association between laterality and mortality differed by comorbidity burden (p for interaction = 0.022). Among patients with CCI ≥ 2, right-sided fracture was associated with higher in-hospital mortality relative to left-sided fracture (aOR 1.29; 95% CI 1.09 to 1.53). No clear association was observed in patients with CCI 0 or 1. For impaired ADL at discharge (BI < 60) (Figure 2b; Supplementary Table ii), effect estimates for right-sided fracture were generally consistent across prespecified sub-groups. Estimates differed by age group (< 75 vs ≥ 75 years; p for interaction = 0.023), with a larger association in patients < 75 years of age. No meaningful heterogeneity was observed by sex, fracture type, BMI category, diabetes mellitus, or CCI category. During sensitivity analyses, double-selection lasso logistic regression produced an aOR for laterality of 1.10 (95% CI 0.99 to 1.22; p = 0.079) for mortality and 1.09 (95% CI 1.05 to 1.13; p < 0.001) for impaired ADL, materially unchanged from the primary logistic regression. E-values for the association between right-sided fracture and impaired ADL were 1.26 (point estimate) and 1.18 (lower confidence limit); for mortality in the CCI ≥ 2 sub-group, 1.90 and 1.37, respectively Literature audit Using the prespecified PubMed strategy, we identified 65 registry-based, multicentre retrospective, and database studies. Only two studies reported fracture laterality in baseline characteristics, and only one study included laterality in prognostic analyses (Supplementary Table iii). Discussion In this large administrative cohort of 72,008 surgically treated HF patients, fracture laterality demonstrated two distinct patterns. First, HFs occurred more frequently on the left side. Second, right-sided fractures were consistently associated with worse independence in ADL at discharge, as evidenced by both item-level BI distributions and adjusted analyses of impaired ADL (BI < 60). Fracture laterality was not associated with in-hospital mortality in the overall cohort; however, among patients with a higher comorbidity burden (CCI ≥ 2), right-sided fractures were associated with increased mortality. Despite being routinely captured in HF datasets, fracture laterality has rarely been examined as an exposure of prognostic interest. In a structured audit of registry-based, multicentre retrospective, and administrative database studies published in seven high-impact journals since 2020, only two studies reported laterality in baseline characteristics, and only one incorporated laterality into prognostic analyses (Supplementary Table iii; Supplementary Methods). This pattern holds even in state-of-the-art prospective cohorts: for example, the WHiTE cohort analysis of 24,523 hip fracture patients evaluated multiple prespecified risk factors for postoperative complications using Cox regression, but fracture laterality was not among the covariates considered.26 This underutilization represents a notable gap in the relevant literature and limits current understanding of whether fracture side conveys clinically meaningful information beyond descriptive reporting. The left-sided predominance of HFs observed in our cohort can be interpreted using at least two non-mutually exclusive mechanistic hypotheses. First, contralateral asymmetry in proximal femoral strength may bias fracture occurrence. In a CT-derived finite element study of post-menopausal females, the right proximal femur was, on average, slightly stronger than the left under simulated sideways-fall loading, which is directionally consistent with higher susceptibility to left-sided fracture.8 In contrast, a dual-energy X-ray absorptiometry-based study of post-menopausal females reported no clear left–right difference at the femoral neck or trochanter, with only small discrepancies at the total femur level; importantly, the mean age was 62 years and the cohort profile differed from our older surgical population.9 Motor laterality may also influence fall mechanics and the side exposed to injurious impact. In older adults, leg preference has been linked to protective stepping responses during laterally directed perturbations, suggesting that lower-limb preference can shape recovery-step strategy and lateral balance loss.28 Because right-foot dominance predominates in the general population, population-level asymmetry in limb preference could contribute to a systematic laterality pattern in HF occurrence.6 Right-sided HF was independently associated with impaired ADL at discharge. Although the effect size was small, the direction was consistent across multiple ADL domains, and was concentrated in lower-limb-dependent items such as transfers, mobility, and stair climbing. A plausible explanation is preferential involvement of the dominant lower limb at the population level, which could disproportionately constrain early mobility processes that are repeatedly required during inpatient recovery. A particularly relevant early-mobility task is gait initiation (GI), the transition from quiet standing to the first step. GI requires anticipatory postural adjustments, defined as preparatory postural shifts that precede step onset and coordinate weight transfer. Experimental work shows that initiating gait with the non-preferential limb alters the characteristics of anticipatory postural adjustments, indicating that limb preference can influence the mechanics of GI.29 If injury more often involves the dominant limb, patients may be forced to initiate gait using non-dominant control strategies, potentially increasing the adaptive and balance demands of early stepping and weight transfer. This mechanistic account presupposes that lower-limb dominance reflects functional specialization in locomotor control rather than a nominal side preference. A functional asymmetry framework supports this premise by positing that the two limbs can contribute differently to locomotor tasks and that the observable asymmetry depends on task demands, including walking speed.30,31 Therefore, dominant-limb injury would be expected to produce diffuse, across-domain limitations that extend beyond a single ADL component, consistent with the broad functional disadvantage observed at discharge. Cognitive load may further constrain early mobility. Dual-task paradigms modify postural control and anticipatory postural adjustments in a dominance-dependent manner, with larger effects reported when the non-dominant limb is engaged in laboratory balance tasks.32 If dominant-limb injury forces greater reliance on non-dominant control strategies during early mobilization, the adaptive and cognitive demands of stepping, weight transfer, and GI may increase, which is compatible with the diffuse, across-domain ADL disadvantage observed at discharge. This effect would be expected to be most clinically consequential in patients with a limited physiological reserve. In patients with CCI ≥ 2, physiological reserve is limited, and even modest delays in mobilization or early functional decline may increase susceptibility to complications such as pneumonia, delirium, and thromboembolic events, thereby increasing in-hospital mortality risk. This pattern is consistent with laterality having limited prognostic relevance in the overall surgical HF population but becoming detectable in patients with a lower reserve, in whom small differences in early mobility may carry greater downstream consequences. Direct evaluation of this pathway will require cohorts with measures of early mobilization, in-hospital complications, and rehabilitation dose, which are not captured in our administrative dataset. Laterality is routinely recorded in HF registries and administrative datasets, yet our literature audit suggests that it is rarely reported and seldom incorporated into prognostic analyses. Our findings indicate that fracture side is not merely descriptive: right-sided fracture was associated with impaired ADL at discharge, and laterality may therefore serve as a simple marker of early functional vulnerability available at presentation. The mortality signal confined to patients with high comorbidity burden further suggests that any laterality-related mobility disadvantage may be most consequential when the physiological reserve is limited. These observations support including laterality alongside established risk factors in registry-based prognostic analyses, and warrant further investigation of its role in planning early mobilization and discharge-oriented rehabilitation. This study had several limitations. First, as an analysis of administrative data, our results are subject to the inherent constraints of routinely collected records: variables are limited to those captured for billing and quality reporting, and residual confounding from unmeasured factors such as frailty, sarcopenia, prefracture mobility, and social support cannot be excluded. E-values suggest that an unmeasured confounder would need to be associated with both exposure and outcome by an odds ratio of approximately 1.26 (impaired ADL) or 1.90 (mortality in the CCI ≥ 2 sub-group) to fully explain the observed associations. Second, fracture laterality (right vs left) is an imperfect proxy for the functionally relevant constructs of limb dominance and the standing or jumping leg, neither of which is captured in administrative datasets. Right- compared with left-sided fracture should therefore be interpreted as a marker for these unmeasured factors rather than as an exposure of intrinsic causal interest, and prospective cohorts that record limb dominance directly will be needed to confirm our interpretation. Third, the DPC/PDPS lacks clinical granularity relevant to recovery, including fall direction, and the timing and dose of rehabilitation; therefore, we could not directly evaluate the hypothesized mediating pathways linking laterality to early mobility and downstream complications. Fourth, ADL at discharge was assessed using the BI, which reflects early in-hospital recovery and may not capture longer-term post-discharge trajectories; discharge destination and postacute rehabilitation services were not fully observable. Fifth, although the item-level distribution of discharge BI showed differences concentrated in lower-limb-dependent activities — consistent with a hip fracture-specific mechanism, we cannot exclude the possibility that the observed laterality, BI association reflects a more general property of the BI instrument rather than a phenomenon unique to hip fracture. Confirmation in independent cohorts, including other lower-limb fracture populations, would strengthen this interpretation. Sixth, missing data were handled using multiple imputation, which relies on assumptions about the missingness mechanism; departures from these assumptions could bias estimates. Finally, generalizability may be limited to surgically treated HF patients within this regional DPC/PDPS consortium and may not extend to nonoperative patients or other health systems. In conclusion, in this large surgical HF cohort, left-sided fractures were more common, and right-sided fractures were independently associated with impaired ADL at discharge. Right-sided fractures were also associated with higher in-hospital mortality among patients with substantial comorbidity (CCI ≥ 2), although no overall mortality difference was observed. These associations persisted across sensitivity analyses but, given the observational design and the inability of administrative data to capture limb dominance, require confirmation in independent cohorts. Fracture laterality, which is a variable routinely captured at presentation but seldom analyzed, nonetheless merits consistent reporting and consideration in registry-based prognostic research. Take home message - Fracture laterality is routinely available at presentation and may provide additional prognostic information without requiring further testing. - Right-sided hip fractures were associated with a modestly higher risk of impaired activities of daily living at discharge, particularly among patients with substantial comorbidity. - Laterality may therefore be considered as an adjunct to established risk factors when planning early rehabilitation and discharge support. Author contributions Y. Takegami: Conceptualization, Project administration, Writing – original draft, Writing – review & editing T. Imaizumi: Data curation, Formal analysis, Software, Validation, Writing – review & editing K. Sato: Resources Y. Osawa: Writing – review & editing H. Funahashi: Writing – review & editing S. Imagama: Supervision Funding statement The authors received no financial or material support for the research, authorship, and/or publication of this article. Data sharing The datasets generated and/or analyzed during the current study are not publicly available due to data protection regulations. Access to the data is restricted to researchers who have obtained the necessary permissions. Further inquiries may be directed to the corresponding author. Acknowledgements The authors used ChatGPT (OpenAI, USA) to assist with English-language editing and improve the clarity of the manuscript. The authors reviewed and revised all AI-generated text and take full responsibility for the final content. Ethical review statement The study protocol was approved by the Institutional Review Board of Nagoya University, Nagoya, Japan (Approval No. 2024-0075). The requirement for individual informed consent was waived because all data were anonymized before analysis. Open access funding The open access publication charge was self-funded by the corresponding author. Supplementary material Detailed methods for the literature audit of hip fracture prognostic studies, and supplementary tables of subgroup analyses as well as a complete list of studies included in the literature audit. © 2026 Takegami et al. This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives (CC BY-NC-ND 4.0) licence, which permits the copying and redistribution of the work only, and provided the original author and source are credited. See https://creativecommons.org/licenses/by-nc-nd/4.0/ Data Availability The datasets generated and/or analyzed during the current study are not publicly available due to data protection regulations. Access to the data is restricted to researchers who have obtained the necessary permissions. Further inquiries may be directed to the corresponding author. References Associated Data Data Availability Statement The datasets generated and/or analyzed during the current study are not publicly available due to data protection regulations. Access to the data is restricted to researchers who have obtained the necessary permissions. Further inquiries may be directed to the corresponding author.