Hip fractures in older adults are associated with prolonged immobility, acute tissue injury, systemic inflammation and substantial perioperative morbidity. These factors activate all three components of Virchow’s triad and make venous thromboembolism an important complication before as well as after fracture fixation. A meta-analysis of hip-fracture cohorts identified advanced age, female sex, coronary heart disease, delayed surgery and raised inflammatory markers among the factors associated with preoperative DVT [1]. In a large cohort of older patients with intertrochanteric fractures, preoperative DVT was present in 12.3%, with prior venous thromboembolism, longer injury-to-screening time, body mass index, peripheral vascular disease, hypoalbuminaemia and elevated D-dimer contributing to risk [2].

Preoperative DVT is difficult to recognise clinically because pain, swelling and immobility caused by the fracture overlap with manifestations of thrombosis, while some thrombi remain asymptomatic. Duplex ultrasonography is a non-invasive reference investigation. A systematic review reported high sensitivity for proximal DVT and high specificity overall, although sensitivity was lower for distal thrombosis and estimates varied by ultrasound technique and study population [3]. Routine ultrasound screening of every patient can nevertheless strain personnel and delay surgery in resource-limited settings, creating a need for practical pretest risk stratification.

The Wells score combines clinical features and risk factors into a pretest probability estimate. Although it was developed outside the geriatric fracture setting, trauma data suggest that increasing Wells scores are associated with a greater likelihood of DVT [4]. In a prospective study of 92 hip-fracture patients, a Wells score ≥2 had 47% sensitivity and 81% specificity, whereas patients scoring 0–1 formed a lower-risk group [5]. Performance may differ in hip-fracture populations because immobilisation, limb swelling and tenderness are common even without thrombosis.

Laboratory markers may complement clinical assessment. D-dimer reflects fibrin turnover but is frequently elevated after fracture, which can limit specificity. The D-dimer-to-albumin ratio has been studied as a biomarker for preoperative DVT in geriatric hip-fracture patients [6]. More complex prediction models incorporating age, functional status, injury-to-admission time, fibrinogen and D-dimer have also shown promising discrimination, although they require external validation [7]. Asian data further indicate that venous thromboembolism may be under-recognised and that local validation of diagnostic pathways remains important [8].

This study therefore assessed the usefulness of Wells scoring in predicting preoperative DVT in geriatric patients with hip fractures. The primary objectives were to estimate sensitivity, specificity, PPV and NPV at a Wells score cut-off of ≥2 and to correlate Wells risk categories with Doppler ultrasonography. Secondary objectives were to assess DVT prevalence and explore the diagnostic contribution of D-dimer and other laboratory parameters.

Study design and setting

This prospective diagnostic cohort study was conducted in the Department of Orthopaedics at Bangalore Medical College and Research Institute and its affiliated hospitals from March 2024 to August 2025. Reporting was structured according to the STARD framework for diagnostic accuracy studies. Institutional ethics committee clearance and written informed consent were obtained before enrolment. The participant pathway is shown in Figure 1.

Participants

Patients of either sex were eligible if they were older than 60 years, had a hip fracture confirmed on radiography, had been immobilised for more than four days following injury and were willing to provide informed consent. Patients with multiple fractures, pathological fractures, open fractures, prior anticoagulant use or diagnosed peripheral vascular disease were excluded.

Index assessment and reference standard

Clinical variables were recorded on a structured case record form. The Wells score was calculated for every participant before surgery. A score of 0–1 was categorised as low risk, 2 as moderate risk and 3–4 as high risk. For the primary 2×2 analysis, a score ≥2 was considered test-positive.

All participants underwent routine blood investigations, D-dimer measurement and bilateral lower-limb arteriovenous Doppler ultrasonography. Doppler ultrasonography served as the reference standard for preoperative DVT [3]. The reported laboratory thresholds were D-dimer >0.5 µg/mL, INR >1.2 and AST >40 U/L. Combined strategies were evaluated by requiring both component tests to be positive.

Sample size

The planned sample size was 29. It was estimated with nMaster version 2.0 using an anticipated sensitivity of 85.3%, two-sided alpha of 0.05, 95% confidence and absolute precision of 13%, based on the study of D-dimer-to-albumin ratio by Yao and colleagues [6].

Statistical analysis

Data were entered in Microsoft Excel 2016 and analysed with IBM SPSS Statistics for Windows, version 29.0 (IBM Corp., Armonk, NY, USA). Continuous variables were summarised with mean, median, standard deviation and range; categorical variables were described with frequency and percentage. Chi-square or Fisher’s exact tests were used for categorical associations. Laboratory measures across Wells score groups were compared using one-way analysis of variance with Welch’s correction. Pairwise relationships between laboratory variables were explored with Pearson correlation.

Diagnostic performance was calculated from 2×2 contingency tables as sensitivity, specificity, PPV, NPV and accuracy. Receiver operating characteristic analysis was used to calculate AUCs and optimal cut-offs. Multivariable logistic regression was used to explore independent predictors of Doppler-confirmed DVT. A two-sided p value <0.05 was considered statistically significant.

Cohort and prevalence of preoperative DVT

Twenty-nine eligible patients were enrolled and all underwent Wells scoring, laboratory assessment and bilateral lower-limb Doppler ultrasonography. Preoperative DVT was confirmed in 10 participants (34.5%), while 19 (65.5%) had no DVT. Descriptive laboratory values are presented in Table 1.

Wells score distribution and risk gradient

The most frequent Wells score was 3 (10/29, 34.5%), followed by 2 (8/29, 27.6%). Overall, 22 participants (75.9%) had a score ≥2. DVT prevalence increased across prespecified risk categories: 1/7 (14.3%) in the low-risk group, 2/8 (25.0%) in the moderate-risk group and 7/14 (50.0%) in the high-risk group (Table 2 and Figure 2).

Diagnostic performance

At a cut-off of ≥2, the Wells score classified 9 of 10 DVT-positive participants as positive but also classified 13 of 19 DVT-negative participants as positive. Sensitivity was 90.0% (95% CI 55.5–99.7%), specificity 31.6% (95% CI 12.6–56.6%), PPV 40.9%, NPV 85.7% and accuracy 51.7%. The AUC was 0.712 (95% CI 0.518–0.906; p=0.048).

D-dimer >0.5 µg/mL yielded 9 true positives, 8 false positives, 1 false negative and 11 true negatives, corresponding to 90.0% sensitivity, 57.9% specificity, 52.9% PPV, 91.7% NPV and 69.0% accuracy. INR >1.2 yielded 80.0% sensitivity, 68.4% specificity and 72.4% accuracy. Requiring both Wells score ≥2 and D-dimer >0.5 µg/mL to be positive produced 80.0% sensitivity, 73.7% specificity, 61.5% PPV, 87.5% NPV and 75.9% accuracy (Table 3 and Figure 3).

ROC analysis showed AUCs of 0.781 for D-dimer, 0.745 for INR and 0.691 for AST. The combined model incorporating Wells score, D-dimer and INR had the largest reported AUC, 0.834 (95% CI 0.683–0.985; p=0.003) (Table 4).

Laboratory associations and multivariable analysis

Across Wells score groups, AST (p=0.007) and INR (p=0.032) differed significantly. Significant pairwise correlations were reported between prothrombin time and D-dimer (r=0.455, p=0.013), AST and INR (r=0.460, p=0.012), creatinine and urea (r=0.437, p=0.018), and platelets and haemoglobin (r=0.371, p=0.047).

In multivariable logistic regression, Wells score ≥2 was associated with an adjusted odds ratio (OR) of 6.82 (95% CI 1.21–38.45; p=0.029), and D-dimer >0.5 µg/mL with an adjusted OR of 8.50 (95% CI 1.45–49.82; p=0.018). INR >1.2 approached statistical significance (OR 4.76, 95% CI 0.98–23.12; p=0.053), whereas AST >40 U/L was not significant (OR 2.66, 95% CI 0.61–11.58; p=0.189). The Hosmer-Lemeshow goodness-of-fit p value was 0.342 (Table 5 and Figure 4).

Table 1. Descriptive Statistics Of Laboratory Parameters (N=29).

SD, standard deviation; WBC, white blood cell count; AST, aspartate aminotransferase; ALP, alkaline phosphatase; INR, international normalised ratio.

Table 2. Wells Score Distribution And Preoperative DVT Prevalence.

Table 3. Diagnostic Performance Of Screening Strategies For Preoperative DVT.

PPV, positive predictive value; NPV, negative predictive value; AUC, area under the receiver operating characteristic curve. D-dimer cut-offs are in µg/mL. Combined strategies required both component tests to be positive.

Table 4. Receiver Operating Characteristic Analysis.

Table 5. Multivariable Logistic Regression For Preoperative DVT (N=29).

OR, odds ratio; CI, confidence interval. Hosmer-Lemeshow goodness-of-fit p=0.342.

This prospective cohort found Doppler-confirmed preoperative DVT in approximately one-third of geriatric hip-fracture patients. The Wells score showed a clear risk gradient and high sensitivity at a cut-off of ≥2, but its low specificity produced many false-positive classifications. D-dimer had the same reported sensitivity with better specificity, and the combined Wells score and D-dimer strategy provided the best balance among the threshold-based screening approaches. The reported model incorporating Wells score, D-dimer and INR had the highest AUC.

The observed DVT prevalence of 34.5% was higher than the 16.3% reported in the prospective hip-fracture cohort of Luksameearunothai and colleagues [5] and the 12.3% reported by Wang and colleagues in elderly intertrochanteric fractures [2]. Differences in eligibility and case mix may explain part of this variation. The present study specifically required immobilisation for more than four days, which may have enriched the cohort for venous stasis. A sample of 29 participants also produces an imprecise prevalence estimate and makes comparison with larger cohorts sensitive to only a few events.

The primary strength of the Wells score in this cohort was sensitivity rather than specificity. Nine of ten Doppler-positive participants had scores ≥2, but 13 of 19 Doppler-negative participants also crossed this threshold. Fracture-related immobilisation, swelling and tenderness may inflate clinical scores even when thrombosis is absent. The resulting sensitivity of 90.0% and specificity of 31.6% differ from the 47% sensitivity and 81% specificity reported at the same threshold in a 92-patient hip-fracture cohort [5]. Trauma data nevertheless support a graded association between Wells score and DVT probability [4]. Together, these findings suggest that the score’s calibration is population-dependent and should not be transferred between ambulatory, general trauma and geriatric hip-fracture settings without local validation.

The stepwise rise in DVT prevalence—from 14.3% at scores 0–1 to 50.0% at scores 3–4—supports the clinical value of risk stratification. However, one of seven participants in the low-risk category still had DVT. An NPV of 85.7% is insufficient to treat a low score as a definitive exclusion test when the consequence of missing thrombosis may be severe. The Wells score may therefore be useful for prioritising urgent imaging, but the present data do not support replacing Doppler ultrasonography with clinical assessment alone.

D-dimer improved specificity and NPV compared with the Wells score while retaining 90.0% sensitivity. This accords with work showing that D-dimer-related measures carry predictive information in geriatric hip-fracture populations [6]. Nevertheless, tissue injury, inflammation and age can elevate D-dimer in the absence of DVT, so a positive result is not confirmatory. Requiring both Wells score and D-dimer to be positive reduced false positives and increased specificity to 73.7%, at the cost of lowering sensitivity to 80.0%. In practice, a sequential pathway would need to define whether the priority is to minimise missed DVT or unnecessary Doppler examinations.

The combined clinical-laboratory model achieved an AUC of 0.834, higher than the AUCs of Wells score, D-dimer, INR or AST alone. This direction is consistent with larger multivariable models in intertrochanteric fracture cohorts, where demographic, functional, timing and biomarker variables together yielded strong discrimination [7]. The reported adjusted associations for Wells score and D-dimer were large, but their confidence intervals were wide. With ten outcome events and four predictors, the regression should be considered exploratory and at risk of model instability or overfitting. It requires validation in a larger, independent cohort before clinical deployment.

These findings have practical relevance to Indian tertiary orthopaedic services, where access to immediate ultrasound may vary and venous thromboembolism in Asian populations may be under-recognised [8]. A simple admission assessment combining Wells score and D-dimer could help prioritise Doppler ultrasonography without delaying definitive fracture management. However, because the study was not designed to test a management algorithm or patient outcomes, such a pathway should be evaluated prospectively before implementation.

STRENGTHS AND LIMITATIONS

The prospective design, uniform application of Wells scoring and Doppler ultrasonography to all enrolled participants, and direct calculation of diagnostic operating characteristics are strengths. The study also evaluated clinically feasible combined strategies rather than relying on a single test.

The principal limitation is the small, single-centre sample. Only ten DVT events occurred, leading to wide confidence intervals and limiting the reliability of multivariable modelling. Requiring more than four days of immobilisation may have selected a higher-risk group and reduced generalisability to patients who present or undergo surgery earlier. Details of blinding between the index assessment and Doppler interpretation were not stated. Follow-up outcomes such as pulmonary embolism, bleeding, surgical delay and mortality were not reported. The source thesis did not include a raw master chart in the manuscript folder, so participant-level recomputation and external verification of the reported values were not possible. The ethics approval number was also not stated.

In this cohort of 29 geriatric hip-fracture patients, the Wells score at a cut-off of ≥2 was sensitive but poorly specific for preoperative DVT. DVT prevalence increased across Wells risk categories, supporting the score’s value for stratification. D-dimer and a combined clinical-laboratory strategy improved diagnostic balance, but neither clinical scoring nor biomarkers should replace Doppler ultrasonography for definitive diagnosis. Larger multicentre studies should validate the reported thresholds and combined model before they are used to direct imaging or prophylaxis.

1. Wang T, Guo J, Long Y, Yin Y, Hou Z. Risk factors for preoperative deep venous thrombosis in hip fracture patients: a meta-analysis. J Orthop Traumatol. 2022 Apr 7;23(1):19. doi: 10.1186/s10195-022-00639-6. PMID: 35391566; PMCID: PMC8991371.
2. Wang X, Jiang Z, Li Y, Gao K, Gao Y, He X, Zhou H, Zheng W. Prevalence of preoperative Deep Venous Thrombosis (DVT) following elderly intertrochanteric fractures and development of a risk prediction model. BMC Musculoskelet Disord. 2022 May 4;23(1):417. doi: 10.1186/s12891-022-05381-y. PMID: 35509097; PMCID: PMC9065244.
3. Goodacre S, Sampson F, Thomas S, van Beek E, Sutton A. Systematic review and meta-analysis of the diagnostic accuracy of ultrasonography for deep vein thrombosis. BMC Med Imaging. 2005 Oct 3;5:6. doi: 10.1186/1471-2342-5-6. PMID: 16202135; PMCID: PMC1262723.
4. Modi S, Deisler R, Gozel K, Reicks P, Irwin E, Brunsvold M, Banton K, Beilman GJ. Wells criteria for DVT is a reliable clinical tool to assess the risk of deep venous thrombosis in trauma patients. World J Emerg Surg. 2016 Jun 8;11:24. doi: 10.1186/s13017-016-0078-1. PMID: 27279896; PMCID: PMC4898382.
5. Luksameearunothai K, Sa-Ngasoongsong P, Kulachote N, Thamyongkit S, Fuangfa P, Chanplakorn P, Woratanarat P, Suphachatwong C. Usefulness of clinical predictors for preoperative screening of deep vein thrombosis in hip fractures. BMC Musculoskelet Disord. 2017 May 22;18(1):208. doi: 10.1186/s12891-017-1582-5. PMID: 28532441; PMCID: PMC5440897.
6. Yao W, Zhang K, Lv Q, Deng Z, Ding W. D-dimer-albumin ratio (DAR) as a new biomarker for predicting preoperative deep vein thrombosis after geriatric hip fracture patients. J Orthop Surg Res. 2023 Aug 31;18(1):645. doi: 10.1186/s13018-023-04139-z. PMID: 37653556; PMCID: PMC10470167.
7. Xiang G, Dong X, Lin S, Cai L, Zhou F, Luo P, Zhu J. A nomogram for prediction of deep venous thrombosis risk in elderly femoral intertrochanteric fracture patients: A dual-center retrospective study. Front Surg. 2023 Jan 6;9:1028859. doi: 10.3389/fsurg.2022.1028859. PMID: 36684366; PMCID: PMC9852608.
8. Wang KL, Yap ES, Goto S, Zhang S, Siu CW, Chiang CE. The diagnosis and treatment of venous thromboembolism in asian patients. Thromb J. 2018 Jan 18;16:4. doi: 10.1186/s12959-017-0155-z. PMID: 29375274; PMCID: PMC5774147.