Mortality and Morbidity Following Surgery for Primary Malignant Musculoskeletal Tumors in the Pelvis and Limbs : A Retrospective Analysis Using the Japanese Diagnosis Procedure Combination Database

Introduction: Resection of malignant pelvic tumors has long been considered to be associated with higher postoperative mortality and morbidity than resection of malignant limb tumors. We compared the postoperative adverse events of pelvic tumor surgery and limb tumor surgery using a national inpatient database. Methods: We identified patients who underwent surgery for primary musculoskeletal malignant tumors of the pelvis or limbs between July and December in 20072010 using the Japanese Diagnosis Procedure Combination inpatient database. We calculated the risk-adjusted odds ratio for the occurrence of postoperative complications following pelvic tumor surgery with reference to limb tumor surgery using a multivariable logistic regression analysis. Results: Of 3255 eligible patients, 3116 underwent limb tumor surgery and 139 underwent pelvic Corresponding author.


Introduction
Resection of malignant pelvic tumors has long been considered to be associated with poorer outcomes and higher morbidity than resection of malignant limb tumors [1]- [7].Although recent advances in perioperative management may have improved outcomes in pelvic tumor surgery, the availability of up-to-date data on mortality and morbidity is limited because these tumors are relatively rare.Data in previous studies were based on small sample sizes of fewer than 100 or obtained retrospectively over decades [2]- [8].In the present study, we used a national inpatient database to compare in-hospital mortality and postoperative complications between primary malignant pelvic tumor surgery and primary malignant limb tumor surgery.

Data Source
In this study, we utilized the Japanese Diagnosis Procedure Combination (DPC) database.Details of the database have been described elsewhere [9].Briefly, discharge abstract and administrative claims data are collected from participating hospitals between July 1 and December 31 each year.The numbers of inpatients in the DPC database were 2.99 million from 926 hospitals in 2007, 2.86 million from 855 hospitals in 2008, 2.57 million from 818 hospitals in 2009, and 3.19 million from 952 hospitals in 2010.The database includes the following data: patient age and sex; diagnoses, comorbidities at admission, and complications after admission recorded according to International Classification of Diseases, Tenth Revision (ICD-10) codes and text data in the Japanese language; procedures according to the original Japanese codes; drugs used; and in-hospital death.

Ethics
The anonymous nature of the data allowed the requirement for informed consent to be waived.This study was approved by the Institutional Review Board of The University of Tokyo.

Patient Background Characteristics
The following variables were abstracted from the DPC database: patient age and sex; tumor origin (bone and soft tissue); prosthetic joint replacement; free vascularized multi-tissue graft; distant metastasis (lung and brain); diabetes mellitus; use of hemodialysis; chemotherapy (cisplatin, doxorubicin hydrochloride, methotrexate, ifosfamide, dacarbazine, and etoposide); duration of anesthesia; volume of blood transfusion; and precise tumor site data.

Statistical Analysis
Univariable comparisons of the outcomes between the subgroups for individual patient characteristics were conducted using the chi-square test.A logistic regression analysis was performed to analyze the concurrent effects of various factors on the occurrence of postoperative complications, while adjusting for clustering of patients within hospitals using a generalized estimating equation [11].The threshold for significance was set at p < 0.05.All statistical analyses were conducted using IBM SPSS version 19.0 (IBM SPSS, Armonk, NY, USA).

Results
The patient background characteristics are shown in Table 1.We identified 3255 eligible patients (1740 men and 1515 women; mean age (±standard deviation), 59.5 ± 18.8 years), consisting of 3116 patients with primary malignant limb tumor surgery and 139 patients with primary malignant pelvic tumor surgery.Sixty-three patients were excluded from the logistic regression analysis owing to lack of data about duration of anesthesia.Patients with pelvic tumors were more likely to receive blood transfusion (54.7%) than those with limb tumors (13.8%) (p < 0.001).More patients undergoing pelvic tumor surgery required longer duration of anesthesia (>480 min) than patients undergoing limb tumor surgery (34.5% vs 9.1%).The demographic patterns of the tumor sites and surgical procedures are shown in Table 2.
The in-hospital mortality and postoperative complications for each category are shown in Table 3. In-hospital mortality for pelvic tumor surgery was comparable to that for limb tumor surgery (0.6% vs 0.7%, p = 0.830).The incidence of postoperative complications following pelvic tumor surgery was two-fold higher than that following limb tumor surgery.Duration of anesthesia, use of blood transfusion, and volume of blood transfusion were associated with higher in-hospital mortality and postoperative complication rate.Of note, patients who required blood transfusion of more than 2500 ml were more likely to have postoperative complications than those who did not (44.6%vs 6.5%, p < 0.001).Similarly, patients with duration of anesthesia over 480 min were more likely to have postoperative complications than those who did not (25.0%vs 6.6%, p < 0.001).
Table 4 shows the results of logistic regression analyses for perioperative complications.Pelvic tumor surgery showed no significant differences for postoperative complications compared with primary limb tumor surgery (odds ratio, 0.96; 95% confidence interval, 0.60 -1.55; p = 0.869).A higher complication rate was significantly associated with higher volume of blood transfusion, distant metastasis, and longer duration of anesthesia.In particular, patients with blood transfusion volumes greater than 2,500 ml and those with duration of anesthesia longer than 480 min showed high odds ratios for postoperative complications (≥2500 ml: 3.69; ≥480 min: 6.11).

Discussion
In this study, we used a Japanese nationwide inpatient database to compare the in-hospital mortality and postoperative complications of patients who underwent surgery for primary malignant musculoskeletal tumors of the pelvis and limbs.Our results indicate that increased risks in pelvic tumor surgery was largely attributable to major intraoperative bleeding requiring blood transfusion and long operation time requiring long duration of anesthesia.
The occurrence of postoperative complications following pelvic tumor surgery was twice as high as that following limb tumor surgery.Of note, patients who underwent pelvic tumor surgery were six-fold more likely to receive a blood transfusion volume greater than 2500 ml that those who underwent limb surgery.After adjusting for confounding variables, including the volume of blood transfusion, we found that the risk of perioperative complications did not differ between pelvic and limb tumor surgery.Our results indicate that the increased risk for pelvic tumor surgery was largely attributable to major intraoperative bleeding requiring blood transfusion.This finding is consistent with those in previous reports [3] [4], although the referred papers reported only external hemipelvectomy.Our data indicated similar tendency in postoperative complications both in external and Other part 807 ( internal hemipelvectomy.
Our results indicate that another increased risk in pelvic tumor surgery was largely attributable to long operation time requiring long duration of anesthesia.There are clear associations between longer duration of anesthesia or operation time and postoperative complications in various medical settings [12]- [15].Shortening of the operation time was also important for musculoskeletal tumor resection.Therefore, to reduce complications after musculoskeletal malignant tumor resection, we should strive to decrease the operation time.
Our study has several limitations inherent to all administrative database studies.First, the DPC database does not provide important clinical data, such as pathological data of each case, tumor volume, individual chemotherapy regimens and dosage of each agent, and details of surgical procedures such as type of pelvic resection or instrumentation used for limb salvage.Second, the DPC database is restricted to information on in-hospital and major complications only and does not provide any information pertaining to those before admission and after discharge.Third, the DPC database provides epidemiological and broad data.Thus, precise and extensive analysis is difficult for the DPC database study.

Conclusion
Our data demonstrated that a larger blood transfusion volume and longer operative time were significantly associated with worse outcomes.The higher morbidity rate after pelvic tumor resection could result from the larger blood transfusion volume and longer anesthesia duration.We need to decide carefully whether to carry out surgical treatment with consideration of the risk and benefit of each candidate treatment plan, in cases with expectations of larger volume of blood transfusion and longer operative time.

Table 2 .
Demographic patterns of tumor sites and surgical procedures.

Table 3 .
In-hospital mortality and postoperative complications.

Table 4 .
Logistic regression analyses for postoperative complications.