Preoperative Elevated International Normalized Ratio Cannot Increase Transfusion or Complication in Primary Total Hip Arthroplasty: A Retrospective Study of 552 Cases

Background: Verify if the elevated preoperative International Normalized Ratio (INR) can increase transfusion and complication rate in primary total hip arthroplasty. Methods: We retrospectively reviewed the database of adults who underwent primary total hip arthroplasty between 2014 to 2018 by the same surgeon. 552 cases were assigned into 3 groups by preoperative INR class: INR ≤ 0.9, 0.9 (cid:0) INR (cid:0) 1.0, and INR ≥ 1.0 eventually. We regarded the transfusion rate as the primary outcome. We also included perioperative blood loss, maximum Hb-drop, postoperative anemia needs medicine, length of stay (LOS), re-operation, the complication rate in 90 days and mortality as the secondary outcomes. Univariable analyses were utilized to compare baselines and outcomes between groups. Binary Logistic Regression was used to adjust differences of baselines among groups. Results: All the cases had an INR (cid:0) 1.5. Among all the cases, 93(16.8%) had INR ≤ 0.9, 268 (48.6%) had 0.9 (cid:0) INR (cid:0) 1.0, and 191 (34.6%) had INR ≥ 1.0, respectively. In the univariable analyses, with the INR elevated, The transfusion rates increased from 1.08% for INR ≤ 0.9, 1.12% for 0.9 (cid:0) INR (cid:0) 1.0 to 5.76% for INR ≥ 1.0 (p (cid:0) 0.05). The overall complication rate increased from 10.8% for INR ≤ 0.9, 16.4% for 0.9 (cid:0) INR (cid:0) 1.0 to 22.5% for INR ≥ 1.0 (p (cid:0) 0.05). When controlling for the demographics and comorbidities characteristics, there was no statistically signicant difference when evaluating the odds of transfusion nor overall complication rate between the groups (p (cid:0) 0.05). Conclusions: The transfusion and complication


Background
Total hip arthroplasty (THA) is considered one of the most successful surgeries in the history of modern medicine [1]. With the aging population and the prevalence of obesity [2], the utilization and cost of THA have grown annually [3]. THA is associated with excessive perioperative blood loss and a high need for transfusion [4,5], as well as THA related complications. Of all the reasons for THA revision, dislocation and mechanical loosening are the main indications [6].
INR refers to the ratio of prothrombin time to the normal prothrombin time [7] which were regarded as an ideal method for judging the anticoagulation effect [8]. It can overcome variability between different laboratories by thromboplastin sensitivity [9]. Among the warfarin used patients, INR was kept between 2.0 and 3.0 as the therapeutic range [10]. A preoperative international normalized ratio (INR) target < 1.5 is suggested by the guidelines [11]. INR was widely used in predicting mortality and bleeding in some diseases. Rudasill et al found that INR showed an important value in predicting mortality and bleeding in End-Stage Liver Disease [11]. Tan et al observed that patients who underwent endarterectomy with high preoperative INR (≥1.5) exhibited higher 90-day mortality than ones with low INR [12]. In another study about manage heart failure with nonvalvular atrial brillation, abnormal INR also showed higher independent risks of mortality [13].
Some previous studies agreed that high INR values have a strong association with the risk of bleeding [14,15]. A retrospective study found patients with INR≥1.3 had a higher transfusion rate in victims of abusive head trauma [16]. An elevated INR is also associated with increased complications after hand surgery [17].
Recently, a retrospective study of more than twenty thousand cases proved INR 1.25 was tightly associated with bleeding, infection, and mortality in total knee arthroplasty [18]. The same author conducted another similar retrospective study in THA and found elevated INR increased bleeding and mortality again [19].

Methods
We retrospectively identi ed a database of adults who underwent total hip arthroplasty between October 15, 2014, to October 14, 2018, by the same senior surgeon(B.S). We obtained relevant in-hospital data by checked medical records in our hospital system. We contacted the discharged patients to get their complication, re-operation, and mortality information after discharge. We brought adult cases (≥18 years of age) who underwent primary total hip arthroplasty into our study. We only included cases with enough relevant pre-operative laboratory data including INR, which was collected within three days preoperatively.
Exclusion criteria included bilateral THA during the same hospitalization, THA after acute infection, revision operation, multiple fractures and patients without adequate data.
We regarded postoperative transfusion rate during the same hospitalization as the primary outcome. We also included perioperative blood loss, maximum Hb-drop, postoperative anemia needs medicine, length of stay (LOS), re-operation, the complication in 90 days, mortality in 90 days and mortality in 12 months as the secondary outcomes. Postoperative complications contained urinary tract infection, renal failure, stroke, cardiovascular accident, pneumonia, stroke, septic shock, pulmonary embolism (PE), deep venous thrombosis (DVT), super cial infection, deep infection, dislocation, fracture, muscular vein thrombosis, subcutaneous ecchymosis, hematoma and wound-healing delay, etc. Perioperative blood loss was calculated by the formulas described by Gross with preoperative hematocrit and the lowest postoperative hematocrit during hospitalization [20][21][22]. All the complications, re-operation and complication data in the rst 90 days postoperatively were collected. Mortality in 12 months was brought into the study.
We compared the demographics and comorbidities characteristics of cases by the INR group. Continuous variables were exhibited as the mean with the standard deviation or as the median with the interquartile range. Categorical variables were exhibited as the number of cases (the percentage). We analyzed continuous variables by one-way ANOVA with post-hoc Tukey test, and categorical variables by the chisquare test or Fisher test. We utilized Binary Logistic Regressions to evaluate the predictive value of INR for transfusion and complication. We put the independent variables of p < 0.1 into the regression model for adjustment. Regressions were adjusted for contralateral THA, ASA class, asymptomatic bacteriuria, cardiac pacemaker, pre-operative hemoglobin, hematocrit, WBC count and serum albumin eventually. In all the comparisons, p<0.05 was considered statistically signi cant. We performed all the statistical analyses by SPSS (version 26; IBM).

Demographic and comorbidities characteristics
From October 15, 2014, to October 14, 2018, 589 cases were screened for eligibility. 21 cases were excluded: 14 failed to meet the inclusion criteria, 7 were lack of adequate preoperative data. We allocated remaining cases into three groups by the class of INR. 3, 8, 5 cases lost follow-up information in the three groups respectively. There was no signi cant difference in the distribution of lost-to-follow-up cases among groups. 552 cases were included in our study nally ( There were no statistical differences between INR groups in ages, genders, BMI, Comorbidities except cardiac pacemaker, pre-operative ESR, Platelet count, WBC count, or D-Dimer (Table 1).

Clinical outcomes
As presented in Table 2 ml for INR≥1.0 in mean, but there were no statistical differences detected (p=0.892, p=0.180, p=0.174, respectively). The maximum Hb drop increased from 25.7±10.5 g/L for INR≤0.9, 26.3±9.9 g/L for 0.9 INR 1.0 to 27.3±9.9 g/L for INR≥1.0 in mean, which were not statistically different among the three groups (p=0.857, p=0.406, p=0.553, respectively). The incidence of anemia needs medication also increased gradually among groups with no statistical signi cance. Length of stay (LOS) in group C was 5.7±2.2 days, which was statistically longer than ones in group A (4.7±1.6 days, p=0.000) and group B (5.1±2.0 days, p=0.007) (Fig.2).
Among all the cases, 10 (10.8%) in Group A, 44 (16.4%) in Group B and 43 (22.5%) in Group C had complications reported postoperatively. There was a statistical difference in complications between Group A and Group C (p=0.017). Continuous increasing trends were observed among groups about the incidence of cardiovascular accident, super cial infection, muscular vein thrombosis, subcutaneous ecchymosis, wound-healing delay and re-operation without statistical signi cance. We also found no statistical differences among groups about urinary tract infection, renal failure, stroke, pneumonia, stroke, septic shock, pulmonary embolism, DVT, deep infection, dislocation, fracture, hematoma, wound-healing delay, mortality in 90 days and mortality in 12 months (p 0.05) ( Table 3).
After adjusted for independent variables of p < 0.1 into the regression model, we found no statistical difference in transfusion (Table 4) and complication at 90 days (Table 5) among the three groups. We adjusted for contralateral THA, ASA class, asymptomatic bacteriuria, cardiac pacemaker, pre-operative hemoglobin, hematocrit, WBC count and serum albumin.

Discussion
Preoperative elevated INR had been con rmed to increase bleeding, complication and mortality in some diseases [11][12][13][14][15][16][17]. Measuring INR before joint surgery was a standard pattern in our institution. But few studies explored the relationship between elevated INR and postoperative outcomes in total joint replacement, which was related to the risk of transfusion closely. In this retrospective study of 552 cases, we found the transfusion and complication rate cannot increase along with the INR elevated in primary THA.
In the group of INR≥1.0, more cases had ASA class 3 and 4 with statistical signi cance (p=0.005), which was relatively high risk. Knol et al also found the proportion of high ASA class in the patients with INR≥1.8 was higher than ones with INR 0.8(69% and 48.7%, respectively, p=0.036) [23]. A larger proportion of cases (29.0%) in the INR≤0.9 group had contralateral THA than other groups with statistical signi cance. Some other studies showed contralateral surgery increases the risk of recurring complications in both THA and TKA [24,25]. We did not nd this trend in our study. Demographic data showed cases in high INR values had statistically lower preoperative hematocrit and serum albumin, Rudasill et al detected similar baseline data in THA and TKA [18,19]. There were no statistical differences in other demographic and comorbidities characteristics among groups. This is the rst study, as far as we know, found the transfusion and complication rate cannot increase along with the INR elevated in primary THA. Our ndings were different from a recent study of Rudasill SE [19], who retrospectively analyzed 17 the improvement of arthroplasty protocol and use of tranexamic acid, our institute has greatly reduced transfusion, complication, and mortality in total joint arthroplasty [26][27][28][29]. Since 2012, almost all the patients who underwent TKA or THA surgery in our institution have received tranexamic acid during the perioperative period. In our institution, ESR and CRP must be controlled in not more than two times the normal range before the operation for most patients, and not more than three times for in ammatory disease patients such as rheumatoid arthritis and SLE. Asymptomatic bacteriuria must be cured before arthroplasty surgery. Thus, our average transfusion rate of all cases was 2.7%, which was largely lower than Rudasill SE (15.5%). The mortality in 90 days of Rudasill SE's study is 0.8%, while we have no patient died in three months after surgery. A ten-year national database from the UK also showed postoperative complications reduced year by year in THA, although the levels of comorbidity elevating [30]. Second, some THA-related complications such as dislocation, fracture, wound-healing delay, and vein thrombosis were not included by Rudasill SE. Also, we followed up the mortality rate in 12 months and found no statistical difference among the groups. The debate about INR security thresholds has been going on for a long time. Some previous studies agreed that high INR values have a strong association with the risk of bleeding in heart valve replacement and head trauma [14][15][16]. But some others observed that an elevated INR did not increase bleeding risk in chest tube placement and hand surgery [31,32].
With the INR elevated by the class step, the length of hospital stay increased gradually compared with INR≤0.9 with statistical signi cance. Similar results can be found in other studies [18,19,33]. We included some minor complications in our study such as muscular vein thrombosis and subcutaneous ecchymosis, which were common in other studies [34][35][36][37][38]. In the aspect of speci c complications, an increasing trend can be detected about the incidence of cardiovascular accident, super cial infection, muscular vein thrombosis, subcutaneous ecchymosis, and wound-healing delay. But there was no statistical difference in this growing trend. We detected no statistical difference in re-operation, mortality in 90 days or mortality in 12 months.
There are some limitations to the current study. First, because this was a retrospective study, there may be some nature bias about the data. Second, sixteen cases lost follow-up among them all, but there was no signi cant difference in the distribution of lost-to-follow-up cases among groups (3.1%, 2.9%, 2.6%, respectively). Thus we believed our lost-follow-up rate was acceptable for these parameters. Third, some patients failed to report adequate information because of forgetting, which may lower the complication rate. But we get most postoperative outcomes data during hospitalization. Other postoperative outcomes were mainly included major complications, re-operation, and mortality, which were unlikely to be forgotten by patients and their families. Therefore, we did not think those limitations would affect the results severely.

Conclusions
In conclusion, the transfusion and complication rate cannot increase along with the INR elevated in primary THA. With the improvement of arthroplasty protocol and use of tranexamic acid, the INR 1.

Declarations Ethics approval and consent to participate
The study was approved by the Ethical Committee of West China Hospital of Sichuan University. And the written informed consent was obtained from each patients included in the study.

Consent for publication
Not applicable.

Availability of data and material
All data and materials are contained within the manuscript.

Competing interests
The authors Linbo Peng, Junfeng Zeng, Yi Zeng, Yuangang Wu, Jing Yang, and Bin Shen declare that they have no con icts of interest.

Funding
This study was not supported by any funding sources.