Article Text

other Versions

Download PDFPDF

Provider education program on the Khorana score to promote venous thromboembolism chemoprophylaxis in patients with gynecologic cancer
  1. Annie Apple1,
  2. Lauren Prescott2,
  3. Marc Robinson2,
  4. Kendall Shultes3 and
  5. Alaina Brown2
  1. 1Obstetrics and Gynecology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
  2. 2Obstetrics and Gynecology, Division of Gynecologic Oncology, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  3. 3Pharmacy, VA Tennessee Valley Healthcare System Alvin C York Campus, Murfreesboro, Tennessee, USA
  1. Correspondence to Dr. Alaina Brown, Vanderbilt University Medical Center, Nashville, Tennessee, USA; a.j.brown{at}vumc.org

Abstract

Objective To evaluate the efficacy of a healthcare improvement initiative to improve provider compliance with the American Society of Clinical Oncology (ASCO) guidelines for venous thromboembolism chemoprophylaxis in patients with gynecologic cancer receiving chemotherapy.

Methods A healthcare improvement initiative was implemented at our institution to improve compliance with American Society of Clinical Oncology venous thromboembolism chemoprophylaxis guidelines in patients receiving chemotherapy with a Khorana score ≥2. Baseline Khorana score and venous thromboembolism data were retrospectively collected for chemotherapy-naïve patients with gynecologic cancer initiating chemotherapy between December 2018 and November 2019. Data for the post-intervention period from December 2019 to December 2020 were captured prospectively. Primary outcome was compliance with American Society of Clinical Oncology guidelines. Secondary outcomes were incidence of venous thromboembolism and complications surrounding venous thromboembolism chemoprophylaxis.

Results We identified 62 patients in the pre-implementation cohort. Approximately half had a Khorana score of ≥2 (52%). Median Khorana score was 2 (range 1–4). None of these patients received prophylactic chemoprophylaxis. Seven (11%) of these patients were diagnosed with venous thromboembolism. Multivariate logistic regression showed increasing Khorana score was associated with increased venous thromboembolism risk (OR 4.9, p=0.01). With cut-off Khorana score of 2, there was no significant increase in venous thromboembolism. However, with a cut-off Khorana score of 3, patients were 15 times more likely to have venous thromboembolism (OR 15.2, p=0.04). In the post-intervention cohort, 22 patients were eligible for chemoprophylaxis and 11 patients were given anticoagulation (50% compliance with guidelines), with no incidence of venous thromboembolism or adverse effects of therapy noted among those receiving chemoprophylaxis.

Conclusion Notifying providers of a patient’s Khorana score improves compliance with American Society of Clinical Oncology guidelines for venous thromboembolism chemoprophylaxis among chemotherapy patients.

  • venous thromboembolism

Data availability statement

Data are available upon reasonable request.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

HIGHLIGHTS

  • Provider education improves compliance with American Society of Clinical Oncology guidelines.

  • Khorana score of ≥3 was associated with increased risk of venous thromboembolism.

  • Khorana score of ≥3 may be a more appropriate cut-off point for venous thromboembolism prophylaxis.

INTRODUCTION

Venous thromboembolism, including deep venous thrombosis and pulmonary embolism, is a significant cause of morbidity and mortality in patients with gynecologic cancer. Patients with gynecologic cancers are at higher risk of venous thromboembolism in comparison with other malignancies due to compounding risk factors common in this patient population, including abdominopelvic surgery, advanced age, high body mass index, immobility, venous obstruction, and treatment modalities, including chemotherapy.1–3 Among patients who are initiating chemotherapy, up to 10% will experience venous thromboembolism within the first 6 months of treatment.4 5 Within the first 2 years after cancer diagnosis, as many as 40% of patients may experience venous thromboembolism. This can lead to poor outcomes, including delays in oncologic therapy, decreased quality of life, unplanned hospital admissions, and death.6

In 2019, the American Society of Clinical Oncology (ASCO) published updated clinical practice guidelines recommending use of the Khorana score to guide shared decision-making regarding initiation of chemoprophylaxis against venous thromboembolism in high-risk outpatients with cancer.7 The Khorana score uses clinical and laboratory data to stratify risk of venous thromboembolism in patients initiating chemotherapy and has been validated in multiple cancer types.8 For those with Khorana score of ≥2 (high-risk), the American Society of Clinical Oncology recommends offering thromboprophylaxis in the absence of significant risk factors and drug interactions.7 Previous studies have criticized poor representation of patients with gynecologic cancer in studies used to develop and validate the Khorana score and have mixed findings regarding the utility of the Khorana score in gynecologic cancers.9 10 Therefore, both the utility of the Khorana score in gynecologic oncology and the role of American Society of Clinical Oncology guidelines in preventing venous thromboembolism in patients with gynecologic cancer require additional investigation.

Internal review of our practices within the Division of Gynecologic Oncology at our institution revealed 0% compliance with American Society of Clinical Oncology guidelines regarding use of chemoprophylaxis for chemotherapy patients with a Khorana score ≥2 during the period December 2018 through November 2019. We therefore designed and implemented a healthcare improvement initiative to improve compliance. Our primary aim was to improve compliance with American Society of Clinical Oncology venous thromboembolism guidelines by 50%, as measured by the percentage of eligible patients who were initiated on chemoprophylactic agents. Our secondary aims were to evaluate the utility of the Khorana score in our gynecologic oncology population as a risk stratification tool for predicting venous thromboembolism, to evaluate the incidence of venous thromboembolism in our population, to evaluate the severity of venous thromboembolism (venous thromboembolism requiring admission), and to characterize adverse outcomes associated with anticoagulation (bleeding).

METHODS

This study was approved by the Vanderbilt Institutional Review Board. This manuscript has been prepared in accordance with the Standards for Reporting Implementation Studies (StaRI) guidelines, and we used the Institute for Healthcare Improvement’s methodology of Plan, Do, Study, Act (PDSA) cycles for development and analysis of our quality improvement intervention.11 12

Context/Intervention

Our institution is a tertiary care academic medical center. A multidisciplinary team within the Division of Gynecologic Oncology was assembled including gynecologic oncologists, a clinical pharmacist, a research data analyst, and a medical student. The team developed the following intervention:

  1. Chemotherapy-naïve patients with a new diagnosis of gynecologic cancer initiating chemotherapy between December 2019 and December 2020 were screened for eligibility for venous thromboembolism chemoprophylaxis.

  2. Patients with a Khorana score ≥2 were eligible for intervention, and study staff sent information to providers via email.

  3. All gynecologic oncologists received emails and were asked to document their decision regarding initiation of chemoprophylaxis in the clinical note.

The healthcare improvement initiative was approved by the faculty in the Division of Gynecologic Oncology in December 2019, and notification of the intervention was provided to all gynecologic oncologists during a divisional meeting. Updates regarding compliance were discussed at quarterly divisional meetings as indicated. The intervention in Plan, Do, Study, Act Cycle 1 included screening patients based on review of pre-clinic notes/charting and sending emails to providers with a list of eligible patients, their Khorana scores, and a brief excerpt from the American Society of Clinical Oncology guidelines with a link to the publication. In Plan, Do, Study, Act Cycle 2, the intervention was modified to include recommended chemoprophylaxis options and dosing. In Plan, Do, Study, Act Cycle 3, the process for screening patients was modified to include screening for patients starting chemotherapy by the clinical pharmacist who had access to all new chemotherapy orders. These changes were made in response to questions, suggestions, and observations throughout the implementation process.

Outcome Measures and Analysis

Pre-implementation data were retrospectively collected on all chemotherapy-naive patients with a new diagnosis of gynecologic cancer who initiated chemotherapy from December 2018 through November 2019. Data for the post-implementation period from December 2019 to December 2020 were collected prospectively. We obtained demographic data, including age, race, oncologic history, and laboratory data. Khorana score was calculated for all patients using previously specified parameters of cancer type, pre-chemotherapy leukocyte count >11×109/L hemoglobin <10 g/dL or use of red blood cell factors, platelet count ≥350×109/L, and body mass index (BMI) ≥35 kg/m2.13 History of venous thromboembolism during treatment and inpatient admission for therapeutic intervention were also recorded.

Within the retrospective pre-implementation group, all patients with a new diagnosis of gynecologic cancer initiating chemotherapy were included, regardless of Khorana score. Given their gynecologic cancer diagnosis, all patients at minimum had a Khorana score of 1. All Khorana scores were allowed in this group as this cohort was not being used to assess the intervention but was used to provide baseline information about our study population. This broad inclusion criterion allowed us to better characterize the distribution of Khorana score within our gynecologic oncology population and to characterize the association between Khorana score and venous thromboembolism incidence, as part of our secondary aim. Patients within this group were evaluated for history of venous thromboembolism from time of diagnosis through the time of data collection, which was at least 18 months after starting chemotherapy. The intervention group included chemotherapy-naïve patients with a new diagnosis of a gynecologic cancer and a Khorana score ≥2 initiating chemotherapy (American Society of Clinical Oncology 2019 guidelines for initiating chemoprophylaxis). Patients who did not have gynecologic cancer or had previously received chemotherapy were excluded.

Our primary aim and process measure for the intervention group was compliance with American Society of Clinical Oncology 2019 guidelines. As previously described, providers were emailed with relevant information for their patient and asked to document discussion and/or consideration of venous thromboembolism chemoprophylaxis in their clinic note. Clinic notes were then reviewed for the presence or absence of Khorana score information and venous thromboembolism risk, which was a proxy for measuring if the educational intervention impacted the process of clinical decision-making regarding initiation of chemoprophylaxis. Our outcome measure was the number of patients prescribed anticoagulants, which was determined based on review of both the clinical note and the patient’s prescribed medication. Patients who were given chemoprophylaxis continued this medication based on the clinical discretion and recommendations of their primary oncologist with no predetermined endpoint, in accordance with American Society of Clinical Oncology guidelines. Secondary outcome measures included incidence of venous thromboembolism and admission for therapeutic anticoagulation, which were obtained via review of the electronic medical record. Balance measures included adverse events associated with anticoagulation, including bleeding.

Data were collected and managed using REDCap electronic data capture tools hosted at Vanderbilt University Medical Center. REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies.14 15 As described above, all data were obtained via review of the electronic medical record. No additional direct resources or costs were required for implementation of this quality improvement project. A sample size of 22 was needed to observe the desired improvement in compliance by 50%. This sample size was calculated using a pre-intervention rate of compliance with American Society of Clinical Oncology 2019 venous thromboembolism guidelines of 0% and a desired post-intervention rate of 50% with an alpha of 0.05, beta of 0.2, and power of 0.8.

Descriptive statistics and multivariate logistic regression were completed within the baseline cohort. The first regression used the outcome of venous thromboembolism (yes/no) with independent variables of age at chemotherapy start, advanced stage disease (stage 3 and 4), white (yes/no), and Khorana score. The second regression used the outcome of venous thromboembolism (yes/no) with independent variables of age at chemotherapy start, advanced stage disease (stage 3 and 4), white (yes/no), and Khorana score ≥3. Demographic information, clinical data, and rates of chemoprophylaxis initiation were compared between patients with Khorana score ≥2 in the pre-implementation and post-implementation cohorts using Mann-Whitney U, Χ2, and Fisher exact tests. IBM SPSS Statistics 27 was used for statistical analyses. In accordance with the journal’s guidelines, we will provide our date for the reproducibility of this study in other centers if such is requested.

RESULTS

Pre-implementation Cohort

Study Population

Within our pre-implementation cohort, 62 patients met the inclusion criteria. Median age was 64 years (range 28–88). The majority of patients were white (n=51, 82%). Other races included black (n=8, 12.9%), Asian (n=1, 1.6%), and other (n=2, 3.2%). The most common cancer type was ovarian (n=26, 42%), followed by uterine (n=23, 37%), cervical (n=7, 11%), and other (n=6, 10%). The majority of patients (56%) had advanced stage disease with stage III (n=20, 32%) or stage IV (n=15, 24%, Table 1).

Table 1

Descriptive statistics for pre-implementation and post-implementation study populations

Khorana Score

The median Khorana score for our pre-implementation study population was 2. Khorana score of 1 occurred in 30 (48%) patients, Khorana score of 2 in 19 (31%) patients, Khorana score of 3 in 12 (19%) patients and Khorana score of 4 in 1 (2%) patient (Figure 1). Within the specific components of the Khorana score, 15 patients (24%) had platelets ≥350×109/L, 14 patients (23%) had hemoglobin <10 g/dL or were using red blood cell factors, 10 patients (16%) had a leukocyte count >11×109/L, and 16 (26%) patients had BMI ≥35 kg/m2.

Figure 1

Distribution of Khorana score for all patients in pre-implementation cohort.

Venous Thromboembolism Incidence and Utility of the Khorana Score

Within our pre-implementation study population, a total of seven patients (11%) were diagnosed with venous thromboembolism during treatment. Venous thromboembolism occurred during neoadjuvant chemotherapy in two patients, within 28 days post-operatively in two patients, during adjuvant chemotherapy in one patient, during radiation therapy (and prior to chemotherapy) in one patient, and during chemoradiation therapy in one patient. Among those with venous thromboembolism, uterine cancer (n=4) was the most common cancer type. Two patients had advanced disease, defined as stage III or stage IV. Three patients were admitted to the hospital and underwent treatment for deep venous thromboembolism, including therapeutic anticoagulation and inferior vena cava filter placement. Khorana score of 3 was the most common score within this population (Figure 2). Increasing Khorana score was noted to be associated with increased risk of venous thromboembolism (OR 4.9, p=0.01). When using a cut-off Khorana score ≥2, no significant increase in venous thromboembolism was noted. However, when using a cut-off Khorana score ≥3, patients were 15 times more likely to have venous thromboembolism (OR 15.2, p=0.04).

Figure 2

Distribution of Khorana score in patients with venous thromboembolism (venous thromboembolism) in pre-implementation cohort. KS, Khorana score.

Post-Implementation Cohort

Study Population

Within the post-implementation cohort, 22 patients were eligible for the intervention with a Khorana score ≥2 and no prior diagnosis of venous thromboembolism. Among this group, median age was 61 years (range 21–79). The majority of patients were white (n=17, 77%). Other races included black (n=2, 9%) or other (n=3, 14%). Cancer types included ovarian (n=11, 50%), uterine (n=6, 27%), cervical (n=2, 9%), and other (n=3, 14%). Most patients had advanced stage disease defined as stage III (n=7, 32%) and stage IV (n=9, 41%, Table 1).

Khorana Score

The median Khorana score among all who were enrolled in the intervention was 3. Nine (41%) patients had Khorana score of 2, seven (32%) patients had Khorana score of 3, five (23%) patients had Khorana score of 4, and one (5%) patient had Khorana score of 5. Sixteen (73%) of patients had pre-chemotherapy platelet count ≥350×109/L. Eight patients (36%) had pre-chemotherapy hemoglobin <10 g/dL or use of red blood cell growth factors. Six patients (27%) had pre-chemotherapy leukocyte count >11×109/L. Ten patients (45%)had BMI ≥35 kg/m2.

Shared Decision-Making, chemoprophylaxis initiation, and venous thromboembolism incidence

Among those eligible for the quality improvement intervention (Khorana score ≥2, n=22), eleven patients (50%) had documented shared decision-making and were given chemoprophylactic agents, specifically the direct oral anticoagulant rivaroxaban, after discussion with their primary oncologist. This represented an improvement in compliance with American Society of Clinical Oncology 2019 guidelines by 50% from our baseline of 0% over the course of the Plan, Do, Study, Act Cycle 3. None of the patients who received chemoprophylaxis developed venous thromboembolism within 6 months of starting chemotherapy or had any adverse effects associated with treatment. Two patients who were eligible but did not initiate chemoprophylaxis ultimately developed venous thromboembolism. Their Khorana scores were 3 and 4, respectively. One patient required admission for therapeutic anticoagulation. Among all patients who did not start chemoprophylaxis, reported barriers for two patients respectively included patient-specific factors like co-morbidities and cost. Reasons why treating physicians did not prescribe chemoprophylaxis for eligible patients were not captured in this study.

Pre-Implementation versus Post-Implementation Differences

Among patients with Khorana score ≥2, there were no significant differences between the pre-implementation and post-implementation cohorts with regards to age, race, stage of disease, or cancer type. The frequency of documented shared decision-making and initiation of chemoprophylaxis was significantly increased in the post-intervention cohort in comparison with the baseline cohort (50 vs 0%, respectively, p<0.01). No significant differences in frequency of venous thromboembolism (n=5 (16%) vs n=29%)) or admission for therapy of venous thromboembolism [(n=2 (6%) vs n=1 (5%))] were observed between baseline and intervention groups (p=0.695, p=1.00, Table 2).

Table 2

Pre-implementation and ost-implementation (Khorana score ≥2) outcome comparisons

DISCUSSION

Summary of Main Results

Our intervention successfully increased shared decision-making for those who were at high risk according to the Khorana score (Khorana score ≥2), from 0% to 50%. Eleven patients were given prophylactic anticoagulation with a direct oral anticoagulant for primary prophylaxis in ambulatory patients with cancer, which has previously been found to reduce the incidence of venous thromboembolism and does not significantly increase major bleeding risk.16 Patients reported increased ease and decreased pain with apixaban in comparison with enoxaparin in patients for venous thromboembolism prophylaxis in patients having surgery for gynecologic malignant neoplasm, which may improve adherence to these medications.17 None of the patients in our study who were given chemoprophylaxis were diagnosed with venous thromboembolism, nor did they experience medical side effects associated with the medication. Notably, two patients who were eligible but did not start chemoprophylactic therapy developed venous thromboembolism, one of whom required inpatient admission for therapeutic anticoagulation.

Although there was a decrease in the incidence of venous thromboembolism in the post-intervention cohort in comparison with the pre-intervention group with a Khorana score ≥2, this difference was not statistically significant. This could be because our study was underpowered to detect a difference in venous thromboembolism or could be associated with barriers to initiation of chemoprophylactic agents. All barriers to initiation of chemoprophylaxis were not captured completely in this study and may be a target of future research, but cost was identified as a barrier for one patient. While direct oral anticoagulants are notably more expensive than other prophylactic agents, including vitamin K antagonists and low molecular weight heparin, apixaban has been shown to be more cost effective for deep venous thrombosis prophylaxis than enoxaparin in the post-operative gynecologic surgical setting.18 Because most venous thromboembolism in patients with gynecologic cancer occurs outside the post-operative setting, further research is required to determine if direct oral anticoagulants are similarly cost effective beyond the post-operative period.19

Results in the Context of Published Literature

Although the Khorana score has been validated across multiple cancer types, it has proved to be less informative for both lung cancer and hematologic malignancies than other cancer types.13 Previous studies focusing on a gynecologic oncology population have found mixed results, showing that a high-risk Khorana score was not associated with risk of venous thromboembolism in ovarian or cervical cancer, but this association may exist in uterine cancer.10 Others have advocated the development of a new risk stratification tool altogether because of poor representation of patients with gynecologic cancer in development of the Khorana score.9 In our pre-implementation gynecologic oncology population, increasing Khorana score was significantly associated with increased risk of venous thromboembolism; however, a cut-off Khorana score of 2, as used in the American Society of Clinical Oncology guidelines, was not associated with increased venous thromboembolism risk. The most notable increase was found in those with a Khorana score ≥3. Of note, all patients with a gynecologic malignancy automatically have a Khorana score of 1 at minimum. These results suggest that a Khorana score of 3 may be a more appropriate cut-off point to determine risk for venous thromboembolism and guide clinical management regarding venous thromboembolism chemoprophylaxis in patients with gynecologic cancer.

Strengths and Weaknesses

Strengths of our study include the quality improvement methodology used for design and execution of this initiative. Limitations of this study include the retrospective nature of our pre-implementation data and the resultant relatively limited data for each patient. Other clinical features and co-morbidities beyond those included in the Khorana score parameters and post-surgical chemoprophylaxis data that may affect the risk of venous thromboembolism were not captured. Furthermore, patients who were enrolled in the prospective healthcare improvement initiative have been followed up for varying amounts of time dependent on their chemotherapy start date. As a result, incidence of venous thromboembolism is not completely known in this population. Our study was powered to detect a difference in implementation but not for clinical metrics such as venous thromboembolism, and therefore our study may be underpowered to conclude that there is not a difference in venous thromboembolism rates or that a Khorana score of 3 is the optimal value for recommendations for prophylaxis.

Contextual Changes and Adaptations

During the study period, the COVID-19 pandemic may have affected rates of diagnosis and chemotherapy initiation at our institution in comparison with the previous year. Adaptations made to the intervention throughout the study period in accordance with Plan, Do, Study, Act cycles included improvement of the screening methodology to include the clinical pharmacist and refinement of education to include potential chemoprophylaxis agents and dosing, as described previously.

Implications for Practice and Future Research

The results of this healthcare improvement initiative contribute novel findings to the field of gynecologic oncology that may advance strategies for prevention of venous thromboembolism in patients initiating chemotherapy. Our findings indicate that a relatively straightforward intervention of notifying providers of a patient’s Khorana score can improve compliance with American Society of Clinical Oncology 2019 venous thromboembolism chemoprophylaxis guidelines. Additionally, we noted that an increasing Khorana score is associated with increased risk of venous thromboembolism, which supports the utility of the Khorana score as a useful tool within the context of the overall clinical picture that may help guide clinical management and prophylaxis recommendations. While American Society of Clinical Oncology guidelines have recommended shared decision-making for initiation of chemoprophylaxis with Khorana score ≥2 in all patients starting chemotherapy, our data provide rationale for future studies evaluating the relationship between Khorana score and incidence of venous thromboembolism in patients with gynecologic cancer. A Khorana score of 3 may be a more appropriate cut-off point for patients with gynecologic cancer, but additional research is needed. Additional future directions for this work include using an implementation science approach to better characterize barriers to starting chemoprophylaxis for both patients and providers in gynecologic oncology.

CONCLUSIONS

Screening patients for risk of venous thromboembolism during the pre-chemotherapy visit with the Khorana score followed by shared decision-making regarding initiation of a direct oral anticoagulant is recommended to reduce venous thromboembolism risk. Provider education on use of the Khorana score and chemoprophylactic options is effective in promoting the integration of American Society of Clinical Oncology guidelines with clinical practice. Further evaluation of these clinical guidelines in the unique gynecologic oncology population is important to improve risk reduction for venous thromboembolism and ultimately, provide comprehensive care for women with gynecologic malignancies.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

References

Footnotes

  • Contributors AA and AB: Involved in conception and design of the project, collected and reviewed data; completed statistical analysis; drafted and revised the manuscript; approved the final manuscript. LP: Involved in conception and design of the project; drafted and revised the manuscript; approved the final manuscript. MR and KS: Collected and reviewed data; drafted and revised the manuscript; approved the final manuscript. KS: collected and reviewed data; drafted and revised the manuscript; approved the final manuscript. AB is the guarantor.

  • Funding This study was funded by NCATS/NIH (UL1 TR000445).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.