Prognostic value of the “modified Osteoporotic Fracture score” for treatment decision in osteoporotic vertebral fractures Authors: Teresa Kalantari, Alvaro Zamarron, Ana Royuela, Victor Bello, Mohamed M Aly, Raquel Gutiérrez-González Published: 2026 ABSTRACT AbstractBackgroundThe mOF score shows promise as a reliable tool to guide treatment decisions of osteoporotic vertebral fractures (OVFs). However, the prognostic value is lacking. This study aims to assess the prognostic value of the mOF score considering clinical and radiological outcomes in accordance with compliant or non-compliant management.MethodsA retrospective cohort study included all patients aged ≥50 years and diagnosed with acute thoracolumbar OVF between 2017 and 2021. Final treatment was compared with the recommendations according to the mOF score. Clinical and radiological outcome variables were pairwise compared in each group of treatment recommendation.ResultsFinal analysis included 482 patients. 81.95% received non-surgical management. Overall, the accuracy rate was 75.1%. The mOF score achieved a sensitivity of 24.7% and a specificity of 86.4% for surgical management prediction (AUC 0.55). In patients with surgical treatment recommendations, compliant management was associated with a significantly lower rate of fracture progression than non-compliant management (11% against 45%, p = 0.019), and lower decline in independent living status (0% compared with 10.6%, p = 0.311). In patients with non-surgical treatment recommendations, non-compliant management was associated with better results in radiological progression (p < 0.001), pain relief (p < 0.001), and the need for further therapy due to poor evolution (p = 0.088).ConclusionsThe mOF score demonstrated limited predictive capacity in discriminating the actual received treatment in our cohort. The surgical recommendation was meaningful, identifying better outcomes in patients compliant with the recommendation; for the non-surgical recommendations, vertebral augmentation still may provide a better outcome than conservative bracing and analgesics.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12891-026-10246-9.Keywords: Spine, Osteoporotic fractures, Clinical decision-making, Kyphoplasty, Vertebroplasty, Validation study Background The annual incidence of major osteoporotic fractures in the European Union ranged from 15 to 32 per 1000 person-years in 2017. Osteoporotic vertebral fractures (OVFs) account for 15.5% of them. The incidence is expected to grow by 23% between 2017 and 2030 [1]. OVFs have been associated with worse physical health-related quality of life, and increased mortality rate [2, 3]. The treatment of thoracolumbar OVFs includes non-surgical management (bracing, relative rest, analgesia), percutaneous vertebral augmentation (VA) techniques (vertebroplasty or balloon kyphoplasty), as well as spinal instrumentation in cases of clinical or radiological instability. Nonetheless, the criteria for endorsing one treatment over another are not distinctly defined, exhibiting significant differences in standard practice [4]. In 2015, the Spine Section of the German Society of Orthopedics and Trauma (DGOU) established a novel classification for thoracolumbar OVFs, termed the “osteoporotic fracture (OF) classification” [5, 6], which has just undergone validation [7]. The OF classification categorizes OVFs into 5 distinct morphological types (OF1 to OF5) of increasing severity [6]. Subsequently, the DGOU developed a scoring system, the “OF score”, to guide OVFs’ treatment decisions [8]. The OF score considers fracture morphology (OF1-5), but also other clinical variables including osteoporosis severity, deformity progression, pain intensity, neurological deficit, the degree of mobilization, and the patient’s health status (Table 1). The cumulative score guides treatment recommendation: <6 points, non-surgical treatment; >6 points, a surgical recommendation; and 6 points, inconclusive (either non-surgical or surgical treatment, according to the surgeon’s discretion) [8]. Ullrich et al. analyzed 518 patients and focused on short-term outcomes, with a mean follow-up of 5 months. Sensitivity and specificity were fair (60% and 68% for a cutoff point of 6.5 to predict surgical treatment), and the authors concluded that the tool was reliable and safe [8]. Another study with smaller sample size (N = 157 patients) and conducted in Thailand subsequently found that the OF score was a valuable tool for determining which patients may benefit from surgery. However, the authors highlighted the risk of over-recommending surgical treatment when relying solely on the OF score [9]. Table 1OF and mOF score ParameterOF SCOREModified OF SCOREGradeGradePointsMorphology (OF 1–5)1–51–52–10Severity of osteoporosisT-score < 3T-score < 3 or qCT HU ≤ 901Deformity progressionYes, NoYes, No1, -1Pain (under analgesia)VAS ≥ 4, <4VAS ≥ 5, <51, -1Neurological deficitYesYes2Mobilization (under analgesia)No, YesNo, Yes1, -1Health statusASA > 3, BMI < 20 kg/m2, nursing case, anticoagulationASA > 3, mFI > 2, anticoagulationEach − 1, Maximum − 2OF osteoporotic fracture, CT computed tomography, HU hounsfield units, VAS visual analogue scale, ASA American Society of Anesthesiologists, BMI body mass index, mFI modified Frailty Index OF and mOF score OF osteoporotic fracture, CT computed tomography, HU hounsfield units, VAS visual analogue scale, ASA American Society of Anesthesiologists, BMI body mass index, mFI modified Frailty Index In an attempt to improve the decision-making process, the DGOU provided a modified version of the OF score (mOF score) in 2020 (Table 1), with pain intensity threshold increased from 4 to 5, and assessment of the health status modified (case nursing and body mass index (BMI) disappear, and modified Frailty Index (mFI) is added) [10]. To date, there are only two studies that analyze the grade of agreement between the mOF score treatment recommendations and the actual therapy received in thoracolumbar OVFs [11, 12]. Both of them analyze small samples and determine a moderate degree of accuracy of the mOF score in identifying surgical candidates [11, 12]. However, the optimal cutoff point of the score differs. The OF and mOF scores show promise as useful and reliable tools to guide treatment decisions of OVFs. However, previous studies have framed the appraisal of these scores in terms of their “validity”, judged by the degree of agreement between the score-based recommendation and the treatment actually delivered in routine practice, which has implicitly been taken as the reference standard. It must be underlined that, in the absence of any consensus on the optimal treatment of OVFs, the real-world practice of a single institution cannot be regarded as a true gold standard. Accordingly, the agreement analysis presented here is intended as a descriptive, hypothesis-generating comparison rather than as a formal validation of the score, and the central aim of the study is to explore its prognostic value. Conversely, the prognostic value of the mOF score, understood as the ability of the treatment algorithm to predict the outcome of treatment with the operative or nonoperative management, is lacking. This study aims to assess the value of the mOF score in two ways: considering the grade of agreement between the mOF score-based treatment algorithm and the actual management, but also the clinical and radiological outcomes in accordance with compliant or non-compliant management. Methods Ethics statement The study was conducted in accordance with the 1964 Declaration of Helsinki and was approved by the local ethics committee (reference PI 64/25), which waived the requirement for patient consent. Because the study was retrospective and relied exclusively on routinely collected registry data, obtaining individual informed consent was impracticable: the cohort comprised a large number of elderly patients treated up to eight years before the analysis, many of whom had died or could no longer be contacted. In this context, and in accordance with Article 9(2)(j) and Recital 33 of the EU General Data Protection Regulation (Regulation (EU) 2016/679) and Spanish Organic Law 3/2018 on the Protection of Personal Data and Guarantee of Digital Rights, which permit the secondary processing of health data for scientific research subject to appropriate safeguards, the Ethics Committee considered the waiver justified. All data were fully anonymised before analysis, the dataset was handled solely by the investigators, and no individual patient can be identified from the reported aggregate results. Study setting and design The study employed a retrospective observational design that was single-center and was based on data from the hospital registry. Inclusion / exclusion The hospital’s institutional data repository was searched for all patients with thoracolumbar OVFs treated between January 1, 2017, and June 30, 2021, who met the following inclusion criteria: (1) acute symptomatic OVF of the thoracolumbar spine (from T1 to L5); (2) fragility fractures, including those caused by no or low-energy trauma (slips, trips, or falls from less than double the body height, and heavy lifting); (3) confirmation of the index fracture via computed tomography (CT) and magnetic resonance image (MRI) in case of doubt; (4) diagnosed with osteoporosis based on national guidelines [13]; (5) ≥ 50 years old; (6) single or multiple-level fractures, with the OF score calculated using only the most severe fracture level in multilevel fractures. The exclusion criteria included: (1) pathological fractures due to tumors or infections; (2) prior instrumentation or VA to the thoracic or lumbar spine; (3) insufficient clinical or radiological data to calculate mOF score, or lost follow-up; (4) high-energy trauma (fall from double the body height or higher, motor vehicle accident with > 100 km/h in cars with airbags, or motor vehicle accident > 50 km/h without airbags, polytrauma); (5) patients with concomitant fracture in the pelvis, upper extremities, and/or lower extremities, which could affect patient mobility and pain. Diagnosis, treatment and follow-up protocol When the chief complaint is back pain and an unknown OVF is present or suspected in plain X-ray, the patient routinely undergoes a CT scan to confirm the diagnosis. Subsequently, an MRI is obtained in case of doubt or needed help to decide the optimal treatment (in multiple fractures, or in case of unknown chronology). When the reason for admission to the hospital is not related to the OVF, but during hospitalization the suspicion arises, the first test is usually plain x-ray and/or MRI. If surgical treatment is intended, then a CT scan is usually obtained. In a minority of patients, the fracture is diagnosed in a CT scan of the abdomen or thorax obtained due to a different chief complaint. Non-surgical management consisted of analgesics and a back brace whenever the patient was incorporated -sitting or standing- until the clinician determined otherwise. The standard of care in our hospital involves bracing for all patients managed conservatively. Therapy assignment was ultimately determined by the individual surgeon, with individual preference being the principal factor; patient preference and comorbidity were considered in a minority of cases. In practice, candidate selection for VA at our institution was guided as much by exclusion criteria as by positive indications. Patients who were already bedridden, who had significant cognitive dysfunction, or who showed clear instability requiring instrumentation were generally not considered candidates for augmentation. Beyond these exclusions, surgeon judgement and patient preference played important roles in the final decision. The first follow-up visit was scheduled at 6–8 weeks after the initial presentation. A second follow-up visit was then scheduled within the 3-to-6-month window. The exact timing of this visit was not fixed a priori but reflected routine clinical practice and logistical availability rather than treatment allocation; in particular, surgically and conservatively treated patients did not follow systematically different schedules. To verify that the variable timing of the second visit did not bias the comparisons, the distribution of follow-up duration was compared between the subgroups under each treatment recommendation. Then, another visit was scheduled at three months and/or six months. On each follow-up visit, the patient was clinically assessed and underwent a follow-up imaging study. Data collection The following variables were retrospectively collected from the patient’s medical record: Epidemiological: age, sex, BMI Epidemiological: age, sex, BMI Clinical: date of diagnosis, American Society of Anesthesiologists (ASA) classification, anticoagulant use, 5-item mFI, affected vertebra and spinal segment, number of acute vertebral fractures, mechanism of injury, neurological deficit on examination, ability to independently mobilize, use of anti-osteoporotic drugs, laboratory tests (including serum calcium, PTH and Vitamin D). Clinical: date of diagnosis, American Society of Anesthesiologists (ASA) classification, anticoagulant use, 5-item mFI, affected vertebra and spinal segment, number of acute vertebral fractures, mechanism of injury, neurological deficit on examination, ability to independently mobilize, use of anti-osteoporotic drugs, laboratory tests (including serum calcium, PTH and Vitamin D). Radiological: T-score (lumbar spine and hip), Hounsfield Units, vertebral height. Radiological: T-score (lumbar spine and hip), Hounsfield Units, vertebral height. Therapeutic: fracture treatment (non-surgical management), vertebral augmentation (VA) or spinal instrumentation, and length of hospitalization. Therapeutic: fracture treatment (non-surgical management), vertebral augmentation (VA) or spinal instrumentation, and length of hospitalization. Fracture classification and mOF score All fractures were classified according to the OF classification (OF1-5) by three neurosurgeons. The classification was performed retrospectively on the baseline imaging studies, and the three observers were blinded to the treatment actually received and to the clinical and radiological outcomes when assigning the OF type and computing the mOF score. In case of disagreement, consensus was obtained by the two senior neurosurgeons. Then, the mOF score was calculated [10]. Patients were divided into 3 groups depending on the mOF treatment recommendation (Group 1, score < 6: non-surgical management; Group 2, score = 6: inconclusive, conservative or surgery; Group 3, score > 6: surgery). Both Group 1 and Group 3 were divided into two subgroups depending on whether the treatment recommendation was concordant with the actual treatment received (A) or discordant (B). Also, Group 2 (inconclusive recommendation) was divided into two subgroups based on whether they received conservative treatment (A) or surgery (B). Outcome measures The following outcome measures were assessed at the time of fracture diagnosis as well as at the latest follow-up: The visual analogue scale (VAS) score of 0 to 10 with 0 representing no pain and 10 the worst imaginable pain. The number of weeks from the start of OVF treatment until pain relief was recorded. Pain relief was defined as the first documented time point at which the treating physician recorded a clinically meaningful reduction of back pain to a mild, non-disabling level that no longer limited mobilization and allowed tapering of major analgesia (VAS score < 3). -The degree of independent living was recorded according to Otsuka et al.’s methodology [14]. The visual analogue scale (VAS) score of 0 to 10 with 0 representing no pain and 10 the worst imaginable pain. The number of weeks from the start of OVF treatment until pain relief was recorded. Pain relief was defined as the first documented time point at which the treating physician recorded a clinically meaningful reduction of back pain to a mild, non-disabling level that no longer limited mobilization and allowed tapering of major analgesia (VAS score < 3). -The degree of independent living was recorded according to Otsuka et al.’s methodology [14]. Radiological fracture progression was defined as vertebral body height loss ≥ 15% or increased kyphotic angle ≥ 10º compared to pre-treatment imaging, determined by X-ray, CT, or MRI [15]. Radiological fracture progression was defined as vertebral body height loss ≥ 15% or increased kyphotic angle ≥ 10º compared to pre-treatment imaging, determined by X-ray, CT, or MRI [15]. Statistical analysis Statistical analysis was performed using Stata version 18 (StataCorp, 2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC) and Jamovi version 2.6.19.0 software. A significance level of p < 0.05 was determined. Descriptive statistics were reported using frequency distributions and percentages for qualitative variables, and measures of central tendency and dispersion for quantitative variables. Sensitivity, specificity, positive and negative likelihood ratios (LR + and LR-) of the mOF score recommendation regarding the therapy actually performed were calculated. For the dichotomous evaluation at the score-defined threshold (mOF score > 6, surgery; <6, non-surgical management), the 64 patients with an inconclusive score of exactly 6 were excluded from the calculation of sensitivity and specificity, because at this value the score does not issue a binary recommendation. These patients were, by contrast, fully included in the analyses based on the empirically derived optimal cut-off points where every patient is unambiguously assigned to one side of the threshold. The area under the receiving operating curve (AUC-ROC) and Youden’s Index were used to determine the optimal cut-off value, along with its sensitivity and specificity, establishing the mOF score threshold for recommending surgical management. Subgroup analysis considering both surgical techniques separately (VA and instrumentation) was then performed with the same methodology. Clinical and radiological outcomes were analyzed in all treatment groups. A pairwise comparison was performed between subgroups within the same treatment recommendation. Continuous variables were compared using the Mann-Whitney U or Kruskal-Wallis tests, while categorical variables were compared using the Pearson’s chi-square test or Fisher’s exact test. This study was made following the recommendations of the STROBE statement [16]. Results Patients and injury characteristics A total of 631 fractures were initially identified (Fig. 1). Seventy-one patients were excluded due to high-energy trauma, 16 had undergone previous spinal surgery, and information was incomplete in 62 patients. Then, 482 patients were included for final analysis (355 female, 73.7%). Mean age was 75.5 years (SD 9.67) and mean mFI was 1.06 (SD 0.95). The mechanism was a low-energy trauma in 64.4% of the fractures, and most of them were located in the thoracolumbar junction (T11-L2) (68.0%). Multiple acute fractures were present at diagnosis in 23.7% of the patients. Only 26.3% of all patients had undertaken antiosteoporosis treatment prior to the OVF, and 55.7% received it after diagnosis. Patient flow and treatment allocation. The diagram shows the inclusion and exclusion process and the classification of patients into the three mOF groups (Group 1, mOF score < 6, n = 353; Group 2, mOF score = 6, n = 64; Group 3, mOF score > 6, n = 65). The pie charts summarize the treatment actually delivered within each group (conservative management, vertebral augmentation, or instrumentation), with the corresponding numbers and percentages No patient presented with neurological symptoms, and only 8.7% were severely limited for active mobilization. Deformity progression was only demonstrated in 29 patients (6.0%) before treatment. A bone density scan was available in 154 patients (31.9%). Interobserver reliability for OF classification was substantial (0.79; 95% CI 0.75–0.83). Most fractures were classified as OF 2 (43.2%), followed in frequency by OF 4 (29.9%) and OF 3 (25.1%). Treatment A total of 395 patients (82%) were managed non surgically, and only 87 (18%) underwent surgery, including VA (N = 61, 70%) and spinal instrumentation (N = 26, 30%). Regarding OF score-based treatment recommendation, most patients fell in the non-surgical Group 1 (N = 353, 73.2%), while 65 fell in the surgical Group 3 (13.5%), and 64 (13.3%) in Group 2 (inconclusive recommendation; Table 2). Then, in 86.7% of patients, the management recommendation was clear according to the mOF score. Most cases classified in the inconclusive Group 2 underwent non-surgical management (78.1%. Group 2 A), and only 21.9% were selected for surgery (Group 2B). The patient flow and the distribution of the treatment actually delivered within each mOF group are summarised in Fig. 1. Table 2Description of the received therapy and mOF-based recommendationOF score recommendationNon-surgicalIndifferentSurgeryTotalPerformed therapySurgical VA Short fusion Long fusion 55 (15.6%) 46 6 3 14 (21.9%) 8 2 4 18 (27.7%) 7 6 5 87 61 14 12 Non-surgical298 (84.4%)50 (78.1%)47 (72.3%)395Total353 (100.0%)64 (100.0%)65 (100.0%)482OF osteoporotic fracture, VA vertebral augmentation Description of the received therapy and mOF-based recommendation Surgical VA Short fusion Long fusion 55 (15.6%) 46 6 3 14 (21.9%) 8 2 4 18 (27.7%) 7 6 5 87 61 14 12 OF osteoporotic fracture, VA vertebral augmentation Of the 61 VA procedures performed, 46 (75.4%) corresponded to patients with a non-surgical management recommendation. Conversely, 34.6% of instrumentations were performed in patients with non-surgical management recommendations. Overall, 75.6% cases were in concordance when comparing the actual therapy received and the recommendation based on the mOF score calculated at diagnosis (Table 2). Validity of the mOF score for management prediction The accuracy of the mOF score in predicting surgical management was limited in our experience, with a specificity of 86.4% but a sensitivity of 24.7% (Table 3). AUC was even lower (0.544) when a cut-off point of 6.5 was selected. Empirical optimal cut-off point was calculated with Youden’s index and settled at 4.5 (Table 3, Suppl Material 1). Table 3mOF score validity for treatment predictionsNon-surgical / SurgicalNon-surgical / VANon-surgical / InstrumentationCut-off pointCut-off pointCut-off point6*Optimal (4.5)6*Optimal (4.5)6*Optimal (5.5)Accuracy75.662.476.161.284.074.8Sensitivity24.751.713.237.755.065.4Specificity86.464.886.464.885.675.4LR+1.811.470.971.074.042.67LR-0.870.751.010.960.520.46AUC0.560.580.500.510.710.71*excluding mOF score of 6 mOF score validity for treatment predictions *excluding mOF score of 6 A subsequent analysis focused on differentiating between non-surgical treatment and VA was performed. Test sensitivity was only 13.2% and specificity rosed to 85.6%. These values were 37.7% and 64.8%, respectively, when considering an optimal cutoff point of 4.5 to predict the need for VA (Table 3). When the analysis focused on predicting instrumentation vs. non-surgical treatment, the test sensitivity was fair (55.0%) and the specificity was high (85.6%). The optimal cutoff point for this purpose was settled at 5.5 (Table 3). Clinical outcomes according to treatment received and recommended Median follow-up was 169 (97.25; 259) days. Baseline characteristics of patients in each treatment group are summarized in Table 4. Patients with the same treatment recommendation were homogeneous regardless of treatment received. There was only one exception, in the nonsurgical recommendation group (Group 1), observing a higher percentage of lumbar fractures (L3-L5) in the compliant subgroup 1 A when compared with the non-compliant subgroup 1B (p = 0.038). Table 4Baseline characteristics according to treatment group combining recommended and received therapies. Pairwise comparisons were made between subgroups of each treatment recommendation groupVARIABLENON SURGICAL RECOMMENDATIONINCONCLUSIVESURGICAL RECOMMENDATION GROUP 1A ( n = 298) GROUP 1B ( n = 55) p-value GROUP 2A ( n = 50) GROUP 2B ( n = 14) p-value GROUP 3A ( n = 18) GROUP 3B ( n = 47) p-valueAGE74.7 (9.64)73.4 (8.98)0.93679.8 (10.1)74.6 (8.30)0.37174.4 (9.58)79.4 (8.55)0.467GENDER (female)223 (74.8)35 (63.6)0.09835 (70.0)12 (85.7)0.31913 (72.2)37 (78.7)0.734BMI26.4 (4.8)27.4 (5.1)0.74126.3 (4.28)25.5 (4.37)0.99224.6 (5.04)24.8 (4.27)0.988LABORATORY Calcium 9.21 (0.7)9.16 (0.6)0.9418.91 (0.6)8.83 (0.5)0.8608.98 (0.6)9.10 (0.7)1.000 OH-Vit D 53.8 (28.5)44.0 (27.7)0.56750.5 (28.2)39.2 (17.5)0.97343.5 (31.5)50.1 (21.4)0.991 PTHi 60.8 (33.2)65.9 (45.5)0.99464.5 (45.0)70.9 (36.0)0.86945.5 (17.3)66.2 (31.8)0.658ASA1.0000.9010.906 I 32 (10.7)5 (9.1)2 (4.0)0 (0)0 (0)0 (0) II 124 (41.6)23 (41.8)15 (30.0)7 (50.0)8 (44.4)13 (27.7) III 139 (46.6)26 (47.3)31 (62.0)7 (50.0)9 (50.0)33 (70.2) IV 3 (1.0)1 (1.8)2 (4.0)0 (0)0 (0)0 (0) mFI1.04 (0.9)0.91 (0.8)0.9821.24 (0.9)1.07 (1.1)0.8731.17 (0.6)1.13 (0.9)0.998INDEPENDENT LIVING0.957-0.709 A 270 (90.6)50 (90.9)36 (72.0)14 (100.0)18 (100.0)37 (78.7) B 11 (3.7)4 (7.3)7 (14.0)0 (0)0 (0)2 (4.3) C 16 (5.4)0 (0)6 (12.0)0 (0)0 (0)8 (17.0) D 1 (0.3)1 (1.8)1 (2.0)0 (0)0 (0)0 (0)MECHANISM No-trauma 64 (21.5)9 (16.4)0.47116 (32.0)3 (21.4)0.5266 (33.3)19 (40.0)0.777 Overextertion 31 (10.4)6 (10.9)0.8153 (6.0)2 (14.3)0.2993 (16.7)7 (14.9)0.979 Low-energy 203 (68.1)40 (72.7)0.53131 (62.0)9 (64.3)0.9949 (50.0)21 (44.7)0.784MORPHOLOGY0.6550.6650.998 OF 1 3 (1.0)1 (1.8)0 (0)0 (0)0 (0)0 (0) OF 2 180 (60.4)26 (47.3)2 (4.0)0 (0)0 (0)0 (0) OF 3 80 (26.8)20 (36.4)14 (28.0)2 (14.3)1 (5.6)4 (8.5) OF 4 35 (11.7)8 (14.5)33 (66.0)11 (78.6)16 (88.9)41 (87.2) OF 5 0 (0)0 (0)1 (2.0)1 (7.1)1 (5.6)2 (4.3)SEGMENT Dorsal 49 (16.4)9 (16.4)1.00014 (28.0)4 (28.6)1.0003 (16.7)7 (14.9)0.987 Lumbar 50 (16.8)3 (5.5)0.0385 (10.0)0 (0)0.5773 (16.7)7 (14.9)0.987 Junction (T11-L2) 199 (66.8)43 (78.2)0.11431 (62.0)10 (71.4)0.75412 (66.7)33 (70.2)0.773MULTIPLE62 (20.8)13 (23.6)0.72016 (32.0)4 (28.6)0.9816 (33.3)13 (27.7)0.762LUMBAR T-score-2.2 (1.3)-2.0 (1.6)0.710-2.9 (1.5)-3.6 (1.7)0.939-3.3 (1.4)-2.7 (1.5)0.928HU80.3 (28.4)83.3 (28.3)0.94863.6 (18.8)71.2 (24.0)0.71664.8 (23.7)68.6 (29.8)0.986PRIOR OP THERAPY82 (28.1)9 (16.4)0.09412 (25.5)4 (28.6)0.9825 (27.8)10 (21.7)0.744POSTERIOR OP THERAPY153 (52.4)30 (54.5)0.88324 (50.0)10 (71.4)0.22410 (55.6)37 (78.7)0.073VAS (median)4 (2; 4)4 (2; 4)0.8614 (2; 5)3.5 (2; 4)0.9927 (5.25; 7.75)5 (4; 7)0.365mOF score2.78 (1.51)3.11 (1.56)0.6766 (0)6 (0)-8.5 (1.38)7.74 (1.01)0.168TREATMENT VA 46 (83.6)8 (57.1)7 (38.9) Instrumentation 9 (16.4)6 (42.9)11 (61.1)FOLLOW-UP (days)169 (100; 244)162 (87.5; 344)0.996119 (86.25; 194)198 (62.25; 5800.924324 (143; 400)179 (134; 220)0.483BMI body mass index, ASA American Society of Anesthesiologists, mFI modified Frailty Index, OF osteoporotic fracture, HU Hounsfield Units, VAS visual analogue scale, mOF modified Osteoporotic Fracture, VA vertebral augmentation Baseline characteristics according to treatment group combining recommended and received therapies. Pairwise comparisons were made between subgroups of each treatment recommendation group BMI body mass index, ASA American Society of Anesthesiologists, mFI modified Frailty Index, OF osteoporotic fracture, HU Hounsfield Units, VAS visual analogue scale, mOF modified Osteoporotic Fracture, VA vertebral augmentation Table 5 and Fig. 2 summarize outcome results. Considering patients with surgical treatment recommendations (Group 3), compliant management (3 A) was associated with a significantly lower rate of radiological progression of the fracture than non-compliant management (11% against 45%, p = 0.019), as well as a lower decline in independent living status (0% compared with 10.6%, p = 0.311). However, the median number of weeks until pain relief was similar in compliant and non-compliant patients (15.5 vs. 14, respectively, p = 1.000). Table 5Clinical and radiological outcomes are presented according to treatment group, combining recommended and received therapies. Pairwise comparisons were made between subgroups of each treatment recommendation groupOUTCOMENON SURGICAL RECOMMENDATIONINCONCLUSIVESURGICAL RECOMMENDATION GROUP 1 A ( n = 298) GROUP 1B ( n = 55) p-value GROUP 2 A ( n = 50) GROUP 2B ( n = 14) p-value GROUP 3 A ( n = 18) GROUP 3B ( n = 47) p-valueHOSPITAL STAY5.2 (6.5)12 (4.8) < 0.001 8.7 (12)14 (8.1) 0.001 18 (9.8)6.5 (6.1) < 0.001 PROGRESSION118 (39.6)2 (3.6) < 0.001 18 (36.0)0 (0) 0.007 2 (11.1)21 (44.7) 0.019 NEW FRACTURE32 (10.7)7 (12.7)0.6425 (10.0)2 (14.3)0.6423 (16.7)11 (23.4)0.740N WEEKS TO PAIN RELIEF11 (8; 15)7 (1; 11) < 0.001 9 (7; 13.5)5 (1; 10) 0.020 15.5 (9.75; 21.3)14 (10; 25.3)0.782CONVERSION THERAPY9 (3.0)1 (1.8)0.0881 (2.0)1 (7.1)0.3921 (5.6)1 (2.1)0.480INDEPENDENT LIVING WORSENING4 (1.4)0 (0)1.0000 (0)0 (0)-0 (0)5 (10.6)0.311In bold, p < 0.05N number Clinical and radiological outcomes are presented according to treatment group, combining recommended and received therapies. Pairwise comparisons were made between subgroups of each treatment recommendation group In bold, p < 0.05 N number Clinical and radiological outcomes by mOF recommendation group. Each panel compares the subgroups within one recommendation group: (A) Group 1 (mOF score < 6, non-surgical recommended), (B) Group 2 (mOF score = 6, inconclusive), and (C) Group 3 (mOF score > 6, surgical recommended). The outcomes displayed are hospital stay (days), radiological progression (%), new fracture (%), weeks to pain relief, conversion therapy (%), and worsening of independent living (%). A consistent color scheme is used for conservative and surgical treatment across all panels to facilitate the comparison between compliant and non-compliant management When focusing on patients with non-surgical treatment recommendations (Group 1), compliant management (1 A) was associated with significantly lower hospital stay compared with non-compliant surgical treatment (5 against 12 days, p < 0.001). However, other outcomes such as radiological progression (p < 0.001), pain relief (p < 0.001), and the need for further therapy due to poor evolution (p = 0.088) showed better results in the non-compliant group (1B). Finally, for the inconclusive recommendations (Group 2), hospital stay was longer for surgical patients (p = 0.001), but progression rate was significantly lower when surgical treatment was performed compared with non-surgical management (p = 0.007). Pain relief was achieved earlier in the surgical group (p = 0.020). However, further therapy due to poor evolution was more frequent in this group when compared with non-surgical patients (p = 0.392). Discussion Key findings and interpretations The modified score developed by the DGOU to facilitate OVFs’ decision-making demonstrated limited predictive capacity in discriminating the actual received treatment in our cohort. The sub-analysis to differentiate between VA and instrumentation in the surgical treatment group showed better results in predicting instrumentation and a poor ability to identify patients who may benefit from VA. Taken together, these results suggest that the score helped to rule out the need for surgical intervention. Still, it showed a notable deficiency in accurately identifying surgical candidates, particularly those who can benefit from VA, as it frequently excluded a significant proportion of individuals deemed eligible in our experience. Regarding the prognostic value of the score, the study, which is the first one to add this analysis, showed that the surgical recommendation of the mOF score was meaningful, identifying better outcomes in patients compliant with the surgical recommendation compared with non-compliant patients; for the non-surgical recommendations, VA still may provide a better outcome than conservative bracing and analgesics; for the inconclusive group, surgery may also be slightly superior, but further studies are needed due to the small sample size in this group. A central conceptual point deserves emphasis. Most fracture classifications and treatment algorithms, including the OF and mOF systems, are built on face validity, that is, on expert consensus regarding morphological and clinical markers of severity. Demonstrating that these markers actually predict meaningful patient outcomes (prognostic validation) is considerably more demanding and, consequently, far less frequently undertaken than reliability or agreement studies [17, 18]. This distinction matters because a severity score captures only a snapshot of the patient’s condition at a single time point, whereas the eventual outcome is shaped not only by the initial fracture severity but also by the subsequent healing process and by multiple patient-related factors, including the severity of osteoporosis, sagittal alignment, body weight, comorbidity, and rehabilitation [19]. The OF classification itself illustrates this gap: it has so far been validated mainly in terms of its reliability and reproducibility, most recently through the internal validation performed by the AO Spine Knowledge Forum Trauma [7], whereas its prognostic value with respect to clinical and radiological outcomes had not been formally examined until the present study. The main novelty of the present work is therefore its emphasis on prognostic validation rather than on mere agreement with treatment decisions. Within this framework, the lack of correlation between the mOF recommendation and our institutional practice is best interpreted as an expression of the clinical equipoise that surrounds the management of OVFs: in the absence of universally accepted guidelines, treatment is shaped by surgeon training and experience, reimbursement systems, institutional philosophy, resource availability, and regional practice culture, which generate genuine practice variation rather than a reference standard against which the score should be judged [19]. Previous literature The findings of the present study differ from those of previous reports on the agreement of both the original and modified OF scores with real-world practice. Ulrich et al. [8] conducted a prospective, multicenter European study involving more than 500 patients. They found that the original tool helped guide initial treatment decisions (AUC 0.684), with a cutoff point for recommending surgical treatment set at an OF score of 6.5. An Asian study with a smaller sample size also described the acceptable performance of the OF score in discriminating the need for surgery, establishing the cutoff point for recommending surgery at 6.5 (Table 6) [9]. Table 6Comparison of epidemiological, therapeutic, prognostic and validity variables between the studies that validate OF and mOF scoresVARIABLESUllrich et al. [8]*(N = 518)Mekariya et al. [9]*(N = 157)Mitani et al. [11]**(N = 237)Tokunga et al. [12]**(N = 118)Present study**(N = 482)Geographic areaCentral EuropeThailandJapanJapanSpainMean age (SD)75 (10)74.2 (10.5)77.3–78.377.175.5 (9.7)Female (%)75.384.771.383.973.7BMI (SD)-23.7 (3.6)21 (3.8)20.2226.3 (4.8)History of trauma (%)-54.838.4-64.4OF classificationOF3 (42%); OF4 (27%); OF2 (26%)OF4 (49%); OF3 (41%); OF2 (9%)OF4 (38%); OF3 (23.6%); OF2 (21%)OF4 (35%); OF3 (25%); OF2 (23%)OF2 (43%); OF4 (30%); OF3 (25%)Joint location (%)6583.462.9-68.0Prior mobility77% full; 22% walking aid---88.2% full; 5% walking aidMean preop VAS (surgery/conservative)-5.7–6.4 / - (instrumentation-VA) / -6.8 / 5.86.5–6.5 / 5.84.25 / 3.63OF score < 6 / 6 / >6 (%)36.9 / 14.5 / 46.617.8 / 7 / 75.2-34.7 / 11 / 54.273.2 / 13.3 / 13.5Surgical treatment (%)66.073.946.4***57.118.0VA/instrumentation (%)-46 / 5467.8 / 32.278.7 / 21.3****70.1 / 29.9Conversion therapy3% (7% group 1, 7% group 3B, 5% group 2 A)---2.7%Progression-VA > instrumentation--33.4 (Conservative > VA > instrumentation)New fracture8% (surgery) / 4% conservative> 20% (surgery)16.4% (surgery) / 0% (conservative)-12.4Accuracy71%82.9%-84.7%75.6%Cutoff point (Se; Sp)6.5 (60; 68)6.5 (87.9; 61)5.5 (88.9; 56.7)6.5 (82; 75)5.5 (36.8; 75.4)AUC0.6840.770.786-0.561BMI body mass index, OF osteoporotic fracture, VAS visual analogue scale, VA vertebral augmentation, Se sensitivity, Sp specificity, AUC area under the curve*original OF score** mOF score*** includes laminectomy, interbody fusion, lateral fusion**** 1 only decompression / 18 open surgeries Comparison of epidemiological, therapeutic, prognostic and validity variables between the studies that validate OF and mOF scores BMI body mass index, OF osteoporotic fracture, VAS visual analogue scale, VA vertebral augmentation, Se sensitivity, Sp specificity, AUC area under the curve *original OF score ** mOF score *** includes laminectomy, interbody fusion, lateral fusion **** 1 only decompression / 18 open surgeries The subsequent modification of the OF score aimed to promote a more conservative management and raised the threshold for pain intensity to VAS 5 [10], identifying this parameter as the most important in therapy decisions [16]. To date, there have been only two Japanese studies, including 237 and 118 patients, respectively, that have analyzed the capacity of this mOF score in predicting management. The first study reported high sensitivity (88.9%) and fair specificity (56.7%) in predicting surgical or non-surgical treatment when the cutoff point was 5.5 [11]. The results were comparable (84% sensitivity and 75% specificity) in the other study, but in that case considering a 6.5 cutoff point (Table 6) [12]. Mitani et al. [11] differentiated between non-surgical management and VA, finding sensitivity and specificity values similar to previously reported when the cutoff point was 4.5. However, they found high sensitivity (86.2%) and moderate specificity (80.3%), with an AUC of 0.929, when the cutoff mOF score was set at 8.5 to differentiate between conservative management and open surgery [11]. The present cohort, which was substantial in size and showed a clear predominance of conservative treatment (superior to the proportion observed in other studies) [8, 9, 11, 12] did not replicate these findings. When dividing the analysis by surgical technique, the predictive value of the test was good for differentiating conservative treatment from instrumentation. Still, the optimal cutoff point was 5.5, three points lower than the value found by Mitani et al. [11]. However, no good discrimination was obtained between conservative and VA therapies (the optimal cutoff point was set at 4.5). The observed discrepancies can be attributed to several factors. First, epidemiological differences: the patients in this study showed a higher BMI and different proportions of each OF classification type, including a lower proportion of severe (OF4-OF5) fractures, Second, local clinical practice: conservative treatment is the preferred strategy at our center, with surgery offered far more restrictively than in previous series (18% of patients, versus 46–74%), which likely contributed to the low sensitivity observed. Finally, pain intensity is an important parameter included in the score [20]. However, this is a subjective concept, whose weighting may be influenced by cultural and other factors; mean pain intensity was more than 2 points out of 10 lower when comparing with that reported by Mitani et al. [11] or Tokunaga et al. [12] (Table 6). These factors converge on a single methodological point: the empirically optimal cut-off in a given cohort is not an intrinsic property of the score but a data-driven value that maximizes discrimination against the local distribution of treatments. A markedly lower mean baseline pain score, a lower proportion of severe fractures, and the clear predominance of conservative management act together to displace the Youden-derived cut-off toward lower values, which explains why our optimal thresholds (4.5–5.5) lie substantially below the 6.5–8.5 values reported in cohorts with a more interventionist profile. This reinforces the notion that a single, universally transferable numerical cut-off is unlikely to exist, and that any threshold must be interpreted considering the local treatment culture and case-mix. When analyzing the prognostic value of the mOF score, patients with clear surgical recommendations but non-compliant conservative management showed a non-significant worsening of their functional status and higher progression rate, compared with patients in the compliant surgical management group. Therefore, the need to accurately treat patients with surgical recommendations must be underscored. Indeed, when the outcomes are examined across the three recommendation groups (Fig. 2), surgical treatment appears to reduce radiological fracture progression in all cohorts; however, this advantage seems to translate into a clear benefit in preserved independent living mainly in patients with an mOF score above 6, which supports reserving the strongest surgical recommendation for this subgroup while still acknowledging a meaningful, predominantly pain- and mobilization-related benefit of augmentation at lower scores. For the non-surgical recommendations, patients who received compliant treatment showed reasonable outcomes, but those who underwent surgical management showed earlier pain relief and lower progression rate. Most of the latter (83.6%) had undergone VA, whereas instrumentation was chosen for the remaining 16.4%. Notably, only 3% of patients compliant with the non-surgical recommendation required subsequent surgical therapy despite the high rate of fracture progression, a proportion comparable to that found in other studies [8]. Then, the key, in this regard, is identifying which patients with lower mOF scores may benefit from invasive management, particularly VA, which has been demonstrated to provide benefits not only as biomechanical support but also in alleviating pain [21]. Consistent with this, recent studies report that kyphoplasty and vertebroplasty outperform conservative treatment in improving pain, mobility, and quality of life, with significantly faster recovery and shorter brace use, while the incidence of new fractures does not differ significantly between augmented and conservatively treated groups [22, 23]. Because these external findings parallel the outcomes observed in our surgically treated subgroup, they support offering VA to appropriately selected patients with lower scores and caution against withholding it solely on the basis of a sub-threshold mOF value, particularly when refractory pain is the dominant clinical feature. In our experience, patients with an mOF score below 7, with OF 2, OF 3, and selected OF 4 fractures, and a fragile clinical or functional baseline situation may benefit from a minimally invasive surgical procedure, which does not need general anesthesia, and, in many cases, is outpatient. The findings hereby reported reinforce the notion that the mOF score should not be the exclusive criterion for decision-making in OVFs. Instead, it should be regarded as a complement to the comprehensive clinical, radiological, and functional assessment of the patient. The “Health status” parameter assesses the medical situation through the following metrics: anticoagulation, ASA, and mFI. However, it does not consider independence in activities of daily living. As an example, patients with less functional independence but a mOF score of 6 or higher were assigned to conservative management in our cohort (Group 3B). Strengths and limitations This study provides the largest sample analyzed to date and evaluates the reliability and prognostic value of the mOF score, with notable strengths in the homogeneity of the cohort and the availability of comprehensive clinical and radiological data. Nonetheless, three limitations deserve particular emphasis. First, because treatment was ultimately surgeon-dependent and no standardized institutional protocol existed, individual surgeon preference and unmeasured indication factors may have influenced both treatment allocation and outcome. The comparisons between compliant and non-compliant subgroups should therefore be read as associations rather than as proof of the causal superiority of one strategy, and residual confounding by indication cannot be excluded despite the comparable baseline characteristics observed across subgroups. Second, the median follow-up of approximately six months, although consistent with previous validation studies, is insufficient to capture long-term fracture progression, delayed vertebral collapse, adjacent-level fractures, and progressive functional decline, all of which may evolve well beyond this interval. The present results on radiological progression and independent living should accordingly be interpreted as short- to mid-term outcomes, and longer prospective follow-up is required to confirm them. Third, the agreement between the score-based recommendation and the treatment actually delivered was benchmarked against single-institution practice, which does not constitute a true gold standard and therefore does not allow a definitive judgment on the validity of the mOF score. This part of the analysis should be regarded as descriptive and hypothesis-generating, and a prospective evaluation of the mOF score is therefore mandatory. Conclusions A comparison of the mOF score with our clinical practice suggests that it does not correlate with the actual treatment in its current format. This lack of correlation reflects the divergence between the score and the conservative-leaning practice of a single center and does not, in itself, constitute an assessment of the validity of the mOF score, since single-institution practice cannot be regarded as a gold standard; the main contribution of the present study lies instead in the analysis of the prognostic value of the score. Patients with mOF score over 6 benefited from surgical treatment compared with those with the same score that were managed conservatively. However, for the conservative treatment recommendation based on a mOF score below 6, the tool did not adequately identify a significant number of patients who ultimately underwent and benefited from surgery, particularly VA. Further studies are needed to validate the score in diverse clinical contexts and populations, and to determine the role of VA in patients with a mOF score below 6. The inclusion of other variables into prediction models, such as functional living independence and patient preferences, should be explored to enhance therapeutic decision-making in this highly prevalent condition. Supplementary Information Supplementary Material 1: Receiver operating characteristic curve of mOF score for predicting surgical versus non-surgical treatment. The cut-off point of 4.5 demonstrated a sensitivity of 51.7% and a specificity of 64.8%. Acknowledgements Not applicable. Abbreviations American Society of Anesthesiologists Area under the receiving operating curve German Society of Orthopedics and Trauma Body mass index Computed tomography Positive likelihood ratio Negative likelihood ratio Modified Frailty Index Modified OF score Magnetic resonance imaging Osteoporotic Fracture Osteoporotic Vertebral Fracture Standard deviation Vertebral augmentation Visual analogue scale Authors’ contributions TK has designed the study; acquired the data, drafted the manuscript, and approved the final version. AZ has acquired the data for the work, reviewed the manuscript for important intellectual content and approved the final version. AR has analyzed the data, reviewed the manuscript for important intellectual content and approved the final version. VB has analyzed the data, reviewed the manuscript for important intellectual content and approved the final version. MMA has designed the study; reviewed the manuscript for important intellectual content and approved the final version. RGG has designed the study; acquired the data, drafted the manuscript, and approved the final version. Funding The author(s) received no financial support for the research, authorship, and/or publication of this article. Data availability The dataset generated and analyzed during the current study is available in the Zenodo repository, https://doi.org/10.5281/zenodo.17726708. Declarations Ethics approval and consent to participate This study was performed in accordance with the principles of the Declaration of Helsinki. It was approved by the Ethics Committee of Puerta de Hierro University Hospital (reference 64/25). Informed consent was not obtained since the study was retrospective (the need for consent was waived by the Ethics Committee of Puerta de Hierro University Hospital). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1: Receiver operating characteristic curve of mOF score for predicting surgical versus non-surgical treatment. The cut-off point of 4.5 demonstrated a sensitivity of 51.7% and a specificity of 64.8%. Data Availability Statement The dataset generated and analyzed during the current study is available in the Zenodo repository, https://doi.org/10.5281/zenodo.17726708.