Information on all ovarian cancer patients is available from the Surveillance, Epidemiology, and End Results (SEER) database (usig) SEER*Stat software (National Cancer Institute, Bethesda, MD, USA, version 8.4.1.2, Database: Incidence-SEER Research Plus Data, 18 Registries (excl AK), Nov 2020 Sub (2000–2018)). Our publicly available data from the SEER database do not require ethics committee approval.

On this basis, we identified a total of 88,653 ovarian cancer patients, and the inclusion criteria were as follows: (1) Patients were diagnosed clinically and/or histologically, (2) Variables: age, race, year of diagnosis, summary stage, survival months, chemotherapy recode, surgery, histological type. The exclusion criteria were: (1) Identified by autopsy or death certificate, (2) Unknown race, age, (3) Unknown summary stage, (4) Unknown chemotherapy, (5) No positive histology, (6) Unknown cause of death, (7) Unknown surgery, (8) Unknow type of reporting source.

Death from cardiovascular disease is the primary endpoint of our study. CVD is defined by the SEER database and includes the following six items: (1) Diseases of the heart, (2) Hypertension without heart disease, (3) Cerebrovascular diseases, (4) Atherosclerosis, (5) Aortic aneurysm and dissection, (6) Other diseases of the arteries, arterioles, and capillaries (Fig. S1).

We chose to include patients who were pathologically positively diagnosed with ovarian cancer from 2000 to 2017. Pathology was coded according to the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) and included the following: 8000, 8001–8005, 8010, 8011, 8012, 8013, 8014, 8015, 8020–8022, 8030–8034, 8041, 8043–8046, 8050–8053, 8060, 8070–8076, 8078, 8083, 8084, 8090–8095, 8097, 8140, 8240, 8245, 8246, 8255, 8260–8263, 8255, 8260–8263, 8310, 8313, 8320–8323, 8330–8333, 8336, 8337, 8340–8347, 8350, 8380–8384, 8440–8442, 8450, 8452, 8453, 8460–8463, 8470–8472, 8480–8482, 8490, 8500–8504, 8508, 8510, 8512–8514, 8520–8523, 8542, 8550, 8560, 8562, 8571–8575, 8590–8593, 8600–8602, 8610, 8620–8623, 8933, 8935, 8980, 8990, 9014, 9015, 9020, 9050–9052, 9071, 9080–9082, 9084, 9090, 9100–9102.

Detailed definitions and information about the variables are as follows: age at diagnosis (01–14, 15–29, 30–44, 45–59, 60–74, 75 + years), race (White, Black, American Indian/Alaska Native, Asian or Pacific Islander), year of diagnosis, histological type (adenocarcinoma, sarcoma, epithelial carcinoma, other types: bchondroblastic osteosarcoma, complex epithelial carcinoma, ependymoma, unclassification tumor, squamous cell carcinoma, special gonadal carcinoma, germ cell carcinoma), summary stage (unknown/unstaged, localized, regional, distant), surgery (no surgery, palliative surgery, cytoreductive surgery, other), chemotherapy recode, cause of death, and follow-up time.

The occurrence of CVM is our primary event of concern and therefore its corresponding competing events are the cause of death due to the primary tumor, other tumors and non-other tumors. Therefore, we have adopted the following statistical approach to illustrate.

All data were obtained from the SEER (Surveillance, Epidemiology, and End Results) database with SEER*Stat software (version 8.4.1.2, National Cancer Institute, Bethesda, MD, USA), Microsoft Excel 2021 (Microsoft, Redmond, 22,082,100, USA), and R studio software (version 4.1.2) were used to complete the data analysis. Independent risk factors related to prognosis were determined by univariate Cox analysis, and a nomogram was developed based on the identified independent risk factors. The ability to discriminate between observed and predicted outcome was evaluated by Harrell's concordance index (C-index) [9]. The higher the value, the better the effect of different variables on survival outcomes. At the same time, we further used the receiver operating characteristic (ROC) curve and area under the curve (AUC) values to evaluate the prediction efficiency of the model. The usefulness of decision curve analysis for evaluating nomograms has been detailed by Vickers et al. [10]. All tests were 2-sided, and a P-value < 0.05 signified statistical significance.