Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Women who present with ovarian cancer as an emergency have a very poor prognosis.Existing evidence includes only single-centre studies with a high risk of bias examining a wide range of potential risk factors, and population-based studies with a lower risk of bias that have mainly focused on stage and age.We found no population-based studies that assessed comprehensively a broader range of risk factors associated with emergency presentation in ovarian cancer.WHAT THIS STUDY ADDS This is the first study examining in depth the determinants of ovarian cancer diagnosis after an emergency admission.We found that two out of five women were diagnosed with ovarian cancer after an emergency admission.Younger and older women, women who were more frail, and those from deprived neighbourhoods were even more likely to be diagnosed as an emergency.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Identifying and understanding the risk factors associated with ovarian cancer are a first step towards the development of targeted interventions that can improve the timely diagnosis of ovarian cancer, reduce emergency presentations and improve outcomes.Our findings are relevant for many countries and therefore international concerted action is needed to develop strategies that can reduce the risk that women are only diagnosed when their symptoms are so severe that emergency admission is necessary.Introduction Ovarian cancer is the eighth most common cancer among women worldwide, accounting for more than 200 000 deaths each year. 1 Most women with ovarian cancer are diagnosed at an advanced stage (stage 3 or 4), which presents a therapeutic challenge and is associated with poor survival.2 This diagnostic challenge can be attributed to the wide range of non-specific symptoms that may be caused by ovarian cancer, including persistent abdominal distension, abdominal or pelvic pain, feeling full or loss of appetite, unexplained weight loss, fatigue, increased urinary frequency or urgency and changes in bowel habits. These symptoms are also common in patients without ovarian cancer, with one in two women presenting to primary care with one of these symptoms at least once a year.3Recent research has underlined the challenges faced by initiatives to diagnose ovarian cancer early.4 A multicentre prospective diagnostic accuracy study in England showed that four in five postmenopausal symptomatic women who were referred to hospital for suspected ovarian cancer did not have primary ovarian cancer.5 Also, general population screening of asymptomatic women was not associated with reduced ovarian and tubal cancer deaths in a UK based randomised controlled trial.6It has been shown that in England, women diagnosed with ovarian cancer after an emergency presentation were four times more likely to die within 2 months7 and five times more likely to die within 1 year,4 compared with women diagnosed through other clinical routes. The issue of ovarian cancer diagnosis following an emergency admission is not confined to England but also affects countries such as the USA, Australia, Denmark, Norway, Canada and New Zealand, where the rates range from about 20% to 50%.4 8This association between cancer diagnosis after an emergency admission and poor survival outcomes is not unique to ovarian cancer. Similar trends have been observed across other cancer types, including stomach, oesophageal, colorectal, liver, pancreatic and lung cancer.4 These patients are also less likely to receive cancer treatment, with patients with pancreatic cancer estimated to be 72% less likely to receive treatment.9Women diagnosed with ovarian cancer as an emergency tend to be more unwell, requiring urgent medical attention and hospital admission. Understanding their characteristics will facilitate improving the diagnostic pathway for ovarian cancer. Using national routinely collected data, this study aims to explore the patient and cancer characteristics associated with the risk of ovarian cancer being diagnosed within 28 days after an emergency admission. To our knowledge, this is the first population-based study to explores a range of patient risk factors associated with an emergency diagnosis of ovarian cancer. We also compared cancer characteristics between women with and without a diagnosis after an emergency admission.Methods Patient population In this national population-based study, we included all women diagnosed with ovarian cancer in England aged ≥18 years between 1 January 2017 and 31 December 2021. English cancer registration data were provided by the National Disease Registration Service. Women with any of the following International Classification of Diseases, 10th revision (ICD-10) codes in the cancer registration data were included: C48 (malignant neoplasm of retroperitoneum and peritoneum), C56 (malignant neoplasm of ovary), C57 (malignant neoplasm of other and unspecified female genital organs) and D39.1 (neoplasm of uncertain or unknown behaviour of female genital organs-ovary). Women with sarcomas and borderline tumours, identified using the second revision of ICD oncology (ICD-O2) (see National Ovarian Cancer Audit’s methodology supplement for more information) or those identified only on the basis of death certificate information were excluded from this study.Data sources The cancer registration data were linked to Hospital Episode Statistics (HES) records of hospital admissions. HES is a curated administrative dataset of care episodes provided by National Health Service (NHS) hospitals in England. 10Patient and cancer characteristics The cancer registration dataset was used to identify key data items, such as date of diagnosis, age, Index of Multiple Deprivation (IMD), ethnicity, FIGO (International Federation of Gynaecology and Obstetrics) stage, cancer grade and morphology. 11 12 The IMD was used as a neighbourhood level measure of socioeconomic deprivation to categorise women into five groups based on their residence address using quintiles of the national distribution of neighbourhoods.11 These neighbourhoods include an average population of 1500 residents or 650 households. Ethnicity data, based on information reported by patients themselves, were available in the cancer registration dataset. Ethnicity was categorised into five groups following the 2021 Census of England and Wales: white, Asian, black, mixed and other. If ethnicity was coded as not stated, it was considered missing.FIGO cancer stage information was available in cancer registration data to describe the spread of cancer.12 Tumour grading and morphology information were also derived from cancer registration dataset using the ICD-O2 classification.13 Morphology was further classified into eight categories using the Get Data Out lookup table: serous carcinoma, clear cell carcinoma, endometrioid carcinoma, mucinous carcinoma, other malignant epithelial, miscellaneous and unspecified, non-specific site and sex cord stromal and germ cell.14Comorbidity and frailty were measured using hospital admission records from the HES dataset. The Royal College of Surgeons (RCS) Charlson score was used to capture the number of comorbid conditions in hospital admission records up to 2 years before and 1 month after diagnosis.15 Frailty was assessed using the secondary care administrative records frailty (SCARF) index, which identifies 32 deficits in hospital admissions up to 2 years before and 1 month after the date of diagnosis.16 The RCS Charlson score and SCARF index were considered to be missing in women without a HES record within that period.Outcome variable Emergency admissions within 28 days before the date of the ovarian cancer diagnosis recorded in the cancer registration data were identified in HES dataset based on the admimeth field (with first character coded as 2). A 28 day period was considered clinically reasonable for histological diagnosis and is identical to that used in the English Routes to Diagnosis algorithm for public health surveillance. 17 It is also consistent with the timeframe (30 day cut-off) used in the international study by McPhail et al.4 The impact of using different time periods for defining emergency admissions is explored in the supplementary material.Statistical analysis Differences in the rate of emergency admissions according to sociodemographic and clinical risk factors were tested with the χ 2 test. Unadjusted risk ratios for the association between these risk factors and the risk of an ovarian cancer diagnosis within 28 days after emergency admission were estimated by univariable Poisson regression analysis with robust standard errors. Multivariable Poisson regression was used to estimate adjusted risk ratios (adjRRs).18 The adjusted regression model included the year of diagnosis, age, IMD, ethnicity, comorbidity according to the RCS Charlson score and frailty index according to the SCARF index. The generalised Wald test was used to examine whether the addition of each predictor improved the model fit. Missing data for ethnicity, RCS Charlson score, SCARF index, cancer stage and grade were handled using multiple imputation by chained equations. The imputation also included the outcome variable (emergency admission) and all other complete variables (year of diagnosis, age, morphology and IMD). We generated 34 imputed datasets, with the number chosen as 100 times the largest fraction of missing information. The distribution of variables with missing data before and after imputation is shown in the online supplemental material. Estimates from the final regression model were pooled across imputed datasets using Rubin’s rules.SP110.1136/bmjonc-2025-001053.supp1Supplementary dataResults We included 28 204 women diagnosed with ovarian cancer in England between 1 January 2017 and 31 December 2021. Of these women, 11 377 (40.3%) were diagnosed within 28 days after an emergency admission. The highest percentage of women diagnosed after an emergency admission were those with severe frailty ( table 1 and figure 1). Of the 3372 women with a SCARF index of 4, 2313 (68.6%) were diagnosed after an emergency admission, compared with 2891 (29.2%) of the 9912 women with a SCARF index of 0 (adjRR 1.88, 95% CI 1.78 to 1.99).Table 1Patient characteristics and risk of an ovarian cancer diagnosis within 28 days after an emergency hospital admission in 28 204 women diagnosed between 1 January 2017 and 31 December 2021 in the English NHSNo of patients (%)No (%) of patients diagnosed within 28 days after an emergency admissionUnadjusted risk ratio95% CIP valueAdjusted risk ratio*95% CIP valueAll patients28 204 (100)11 377 (40.3)Year of diagnosis 20175643 (20.0)2204 (39.1)Reference0.0002Reference0.01 20185658 (20.1)2259 (39.9)1.020.98 to 1.071.020.97 to 1.06 20195822 (20.6)2268 (39.0)1.000.95 to 1.040.980.94 to 1.02 20205346 (19.0)2271 (42.5)1.091.04 to 1.141.061.01 to 1.10 20215735 (20.3)2375 (41.4)1.061.01 to 1.111.020.98 to 1.07Age (years) 18-29490 (1.7)209 (42.7)1.181.06 to 1.31<0.0011.361.22 to 1.52<0.001 30-39808 (2.9)255 (31.6)0.870.78 to 0.970.970.87 to 1.07 40-492109 (7.5)679 (32.2)0.890.83 to 0.950.950.88 to 1.02 50-594920 (17.4)1617 (32.9)0.910.86 to 0.960.950.90 to 1.00 60-696402 (22.7)2319 (36.2)ReferenceReference 70-798096 (28.7)3346 (41.3)1.141.09 to 1.191.061.02 to 1.11 >795379 (19.1)2952 (54.9)1.521.45 to 1.581.251.20 to 1.31Index of Multiple Deprivation (quintiles of national distribution) 1 (least deprived)6261 (22.2)2367 (37.8)Reference<0.001<0.001 26108 (21.7)2338 (38.3)1.010.97 to 1.061.000.96 to 1.05 35823 (20.7)2377 (40.8)1.081.03 to 1.131.061.01 to 1.11 45200 (18.4)2170 (41.7)1.101.06 to 1.151.081.03 to 1.13 5 (most deprived)4812 (17.1)2125 (44.2)1.171.12 to 1.221.111.06 to 1.16Ethnicity† White24 313 (91.1)9835 (40.5)Reference0.81Reference0.11 Asian/Asian British1231 (4.6)504 (40.9)1.000.94 to 1.081.050.98 to 1.12 Black/black British520 (2.0)222 (42.7)1.050.95 to 1.161.090.99 to 1.21 Mixed159 (0.6)59 (37.1)0.910.75 to 1.120.950.77 to 1.16 Other459 (1.7)186 (40.5)0.990.88 to 1.111.100.99 to 1.23 Missing1522 (5.4)571 (37.5)RCS Charlson score (No of comorbidities)† 015 643 (58.6)5440 (34.8)Reference<0.001Reference<0.001 16990 (26.2)3445 (49.3)1.411.37 to 1.461.131.09 to 1.17 22626 (9.8)1527 (58.2)1.661.59 to 1.731.131.07 to 1.18 ≥31427 (5.4)965 (67.6)1.921.84 to 2.011.161.09 to 1.22 Missing1518 (5.4)0 (0)SCARF index† Fit9912 (37.1)2891 (29.2)Reference<0.001Reference<0.001 Mild7616 (28.5)3200 (42.0)1.441.38 to 1.501.361.30 to 1.42 Moderate5786 (21.7)2973 (51.4)1.751.68 to 1.831.551.48 to 1.63 Severe3372 (12.6)2313 (68.6)2.322.23 to 2.421.881.78 to 1.99 Missing1518 (5.4)0 (0)*Adjusted for year of diagnosis, age, Index of Multiple Deprivation, ethnicity, RCS Charlson Comorbidity score and SCARF index, with missing values imputed using chained equations (see online supplemental information for distributions following imputation).†Percentage of women with available data, apart from percentage of women with missing data.RCS, Royal College of Surgeons; SCARF, secondary care administrative records frailty.Figure 1Graphical representation of risk ratios (results also presented in table 1) representing the association of patient characteristics and the risk of an ovarian cancer diagnosis within 28 days after an emergency admission, adjusted for all other patient characteristics (see Methods for further explanation). RCS, Royal College of Surgeons.The percentage of women diagnosed after an emergency admission was much higher in women with more comorbidities, but the associations were reduced with adjustment for other characteristics. For example, of the 1427 women with three or more recorded comorbidities, 965 (67.6%) were diagnosed after an emergency admission, compared with 5440 (34.8%) of the 15 643 women without comorbidities (adjRR 1.16, 95% CI 1.09 to 1.22).Furthermore, being diagnosed after an emergency admission was more common in younger women (18–29 years: 209 of 490 (42.7%), adjRR 1.36, 95% CI 1.22 to 1.52) and in older women (≥80 years: 2952 of 5379 (54.9%), adjRR 1.25, 1.20 to 1.31), compared with women aged 60–69 years (2319 of 6402 (36.2%)). Women from the most deprived quintile of neighbourhoods had a higher risk of being diagnosed after emergency admission (2377 of 5823 (44.2%)) than those from the least deprived quintile (2367 of 6261 (37.8%), adjRR 1.11, 1.06 to 1.16). No statistically significant differences were observed in the percentage of women diagnosed after an emergency admission according to ethnicity.The percentage of women diagnosed with ovarian cancer after an emergency admission was slightly higher in 2020 (2271 of 5346 (42.5%)) than in other years (adjRR 1.06, 95% CI 1.01 to 1.10), and compared with women diagnosed in 2017 (2204 of 5643 (39.1%)). Of the 8438 women diagnosed after an emergency admission with available stage information, 1155 (13.7%) had early stage cancer (stage 1 or 2) compared with 5442 (39.2%) of 13 865 not diagnosed after an emergency admission (table 2). Women diagnosed after an emergency admission were also less likely to have low or moderate grade tumours (958 of 2224 (14.4%)) than women not diagnosed after an emergency (2892 of 11 946 (24.2%)). The differences in cancer morphology between those diagnosed and not diagnosed after an emergency admission were relatively small.Table 2Distribution of cancer characteristics according to whether or not cancer was diagnosed within 28 days after an emergency hospital admission in 28 204 women diagnosed between 1 January 2017 and 31 December 2021 in the English NHSDiagnosed within 28 days after an emergency admissionAll patientsP valueYesNoAll11 377 (100)16 827 (100)28 204 (100)Cancer stage*<0.001 1878 (10.4)4175 (30.1)5053 (22.7) 2277 (3.3)1267 (9.1)1544 (6.9) 33651 (43.3)5518 (39.8)9169 (41.1) 43632 (43.0)2905 (21.0)6537 (29.3) Missing2939 (25.8)2962 (17.6)5901 (20.9)Grade*<0.001 Low552 (8.3)1954 (16.4)2506 (13.5) Moderate406 (6.1)938 (7.9)1344 (7.2) High5706 (85.6)9054 (75.8)14 760 (79.3) Missing4713 (41.4)4881 (29.0)9594 (34.0)Morphology†<0.001 Serous carcinoma6226 (54.7)9256 (55.0)15 482 (54.9) Clear cell carcinoma240 (2.1)958 (5.7)1198 (4.3) Endometrioid carcinoma340 (3.0)1278 (7.6)1618 (5.7) Mucinous carcinoma426 (3.7)1179 (7.0)1605 (5.7) Other malignant epithelial1939 (17.0)1813 (10.8)3752 (13.3) Miscellaneous and unspecified1563 (13.7)974 (5.8)2537 (9.0) Non-specific site342 (3.0)268 (1.6)610 (2.2) Sex cord stromal and germ cell301 (2.7)1101 (6.5)1402 (5.0)Data are number (%).*Percentage of women with available data, apart from percentage of women with missing data.†Grouped using the Get Data Out categories.Different time windows (ranging from 7 to 56 days) before diagnosis were used to define an emergency admission in sensitivity analyses. Using a 7-day cut-off value, 8329 patients (29.5%) were identified as having been diagnosed following an emergency admission. This proportion increased to 10 791 patients (38.3%) with a 21-day cut-off and to 11 377 patients (40.3%) with the 28 day cut-off used in our primary analysis. Using a 35-day cut-off increased the value to 11 810 patients (41.9%), and a 56-day cut-off further increased it to 12 626 patients (44.8%). More information can be found in the online supplemental material.Discussion This national study, using English cancer registration data linked to administrative records of admission to NHS hospitals, is the largest to date examining the patient risk factors associated with an ovarian cancer diagnosis after an emergency hospital admission. About two in five women were found to be diagnosed with ovarian cancer after an emergency admission.A key strength of this study is its use of national datasets that allowed the inclusion of all women diagnosed with ovarian cancer in England between 2017 and 2021. Fallopian tube and primary peritoneal cancers were also included in the analysis as they share a common embryological origin as well as clinical characteristics and treatment options with ovarian cancer. Borderline tumours were excluded as their malignant potential is much lower than invasive ovarian cancers and they have significantly better survival outcomes.Younger and older age, socioeconomic deprivation, and frailty were found to be increasing risk factors for ovarian cancer being diagnosed after an emergency admission. Also, women who were diagnosed after an emergency admission were less likely to have early stage disease and low or moderate grade tumours. These findings are broadly in line with other population-based studies that were carried out in England in populations diagnosed with cancer at least a decade earlier.17 19–21An important finding was that the risk of being diagnosed with ovarian cancer after an emergency admission was higher in younger women, despite having higher rates of early stage low grade sex cord and germ cell cancers, factors associated with lower rates of ovarian cancer diagnosis following emergency admission.22 The increased risk was greater following adjustment for other factors, such as frailty and comorbidity, that are less common in younger women. The finding of higher risks in both the youngest and older age groups has been reported in other cancers, such as chronic lymphocytic leukaemia, liver, colon, acute myeloid leukaemia and acute lymphoblastic leukaemia.20 In common with ovarian cancer, patients diagnosed with some of these other cancers tend to present with non-specific symptoms. This might suggest greater difficulty in suspecting these cancer diagnoses in younger patients who present to primary care, preventing them from following an elective pathway.23 However, these speculations require further exploration before definitive conclusions can be drawn.Women living in the most deprived quintile of neighbourhoods were found to have a 10% higher risk of being diagnosed with ovarian cancer after an emergency admission compared with those living in the least deprived quintile, even with adjustment for age, comorbidity and frailty.24 A similar pattern of results has been reported for cancers, including colorectal, oral, oropharyngeal and anal cancer.19 20 However, it is important to note that women from different ethnic backgrounds do not appear to be at higher risk of an emergency diagnosis. The association between comorbidity and risk of diagnosis after an emergency admission remained but was much reduced once the analysis was adjusted for variables closely associated with comorbidity, such as frailty and age.Other studies looking at emergency presentation before an ovarian cancer diagnosis have reported contradictory results.25–28 These studies have several methodological limitations, especially small study sizes. However, all studies but one25 reported an association between a previous emergency admission and more advanced stage, which suggests that for many women with an ovarian cancer diagnosis after an emergency admission, the cancer was detected at a stage that it already causes symptoms so severe that acute hospital care was deemed to be necessary.26–28A limitation of our study was the lack of complete data on comorbidities and frailty for a proportion of the cohort (about 5%), as well as missing cancer stage (21%) and tumour grade (34%) information. This limitation was minimised through the use of multiple imputation. Also, relying on secondary administrative hospital data means that we could not capture primary care interactions before the emergency admission. However, the major strength of this analysis was its scale and representativeness, using linked national datasets to provide a comprehensive, population-based picture of emergency ovarian cancer diagnosis.The results in this paper showed the urgency of a detailed investigation examining the interplay between the patient, cancer and health system factors that increase the risk of late diagnosis and emergency presentation.29 This investigation should include patient awareness of alarm symptoms and healthcare seeking behaviour, initiation of testing as well as the performance of current diagnostic testing in primary care, limitations in service capacity in primary and secondary care, lack of adequate prioritisation of waiting lists and, perhaps most challenging of all, the difficulty of developing efficient pathways for patients with non-specific symptoms.5 30 It is important that this type of exploration takes into account our findings that the risk of an emergency diagnosis is higher in younger and in older women, in frailer women, and in those from deprived neighbourhoods.In England and Wales, the National Ovarian Cancer Audit launched a national quality improvement initiative in November 2025 with the aim of improving outcomes for women diagnosed with ovarian cancer following an emergency admission by increasing the receipt of treatment for these women.31 Gynaecological cancer systems across England and Wales were able to participate in this national effort to identify each system’s pressure points.Our findings are also relevant to other countries. This implies that an international perspective could benefit the development of effective and efficient targeted interventions. Moreover, this work, aiming to improve the timeliness of the diagnostic pathway, may not need to be limited to ovarian cancer as other solid cancer types follow similar patterns.4Conclusions Two in five women are diagnosed with ovarian cancer only after an emergency admission, an issue that affects many high-income countries. 4 8 We found that younger and older women, frailer women, and those living in more deprived neighbourhoods were at an even higher risk of being diagnosed only after the symptoms of their cancer were so severe that an emergency admission was deemed necessary. Interventions need to be developed and implemented to identify the women who are most at risk, in secondary care settings as well as in the community.Concerted action, where possible with support from international collaborations, is needed to improve referral and diagnostic pathways with a focus on increasing patient awareness, improving early recognition of alarm symptoms, handling the prioritisation of waiting lists and developing efficient diagnostic pathways that can provide a timely service to the many women with non-specific symptoms.