Cancer related maternal mortality and delay in diagnosis and treatment: a case series on 26 cases

Patients registered between 2001 and 2012 by the MMC who died during pregnancy or within 1 year postpartum were screened. We selected the patients that died because of (recurrent) cancer. Patients with cancer related symptoms during pregnancy but diagnosed after delivery were included. If patients died from cancer but only showed symptoms after delivery, they were not eligible. Missing data in the reports of the MMC were completed by review of the medical files. Ten medical specialists from different disciplines, including medical oncologists, gynaecological oncologists and obstetricians, were provided detailed case descriptions, with information on diagnosis and therapy. They scored anonymously whether delay had occurred and whether this was doctor- and/or patient-related delay. Delay was defined as an extension of the period between symptoms/presentation and diagnosis, and diagnosis and treatment, compared to optimal care based on expert opinions. For the patients with recurrent disease during pregnancy, delay was assessed from the moment of the new onset of symptoms caused by recurrent disease. Delay due to other primary health care providers like midwifes were considered to be doctor-related delay.

Because of the observational nature of this study, statistical analysis was restricted to descriptive statistics and the evaluation of inter-observer agreement with a kappa score. Because our data was scored by a fixed number of observers using a numeric scorings system the Fleiss kappa score was used.

Between 2001 and 2012, 32 cancer related maternal deaths were reported to the MMC. Four of these cases were excluded: two patients were treated with chemotherapeutic agents for other indications than malignant disease, one patient had a benign brain tumour and one patient became symptomatic after delivery. Two additional anonymous cases were reported by the CBS, which were included but could only be used to calculate the incidence of cancer related maternal mortality. The 26 remaining cases were available for analysis of delay. With an average number of 188,604 live births per year (range 202,603 in 2001 to 175,959 in 2012) in the Netherlands, the Dutch cancer related MMR between 2001 and 2012 was 1.23 per 100,000 live births [11].

Brain tumours, gastro-intestinal tumours and melanomas were the most common types of cancer causing maternal mortality. Patient characteristics are shown in Table 1. The median maternal age at diagnosis was 34 years (range 27–45 years). Of the 26 patients, 73% was diagnosed during pregnancy; one patient (4%) was diagnosed in the first trimester, ten patients (39%) in the second trimester and eight patients (31%) in the third trimester. Seven patients (27%) were diagnosed after delivery, but experienced cancer related symptoms during pregnancy. Median survival after diagnosis was 101 days with a range of 3–352 days. Stage IV disease was present at diagnosis in 90% of the patients; the other 10% had stage III disease. Four patients had recurrent disease during pregnancy and were diagnosed with advanced stage disease at recurrence while pregnant; one melanoma and three astrocytomas.

Standard curative therapy was applied in nine cases. Four patients received adjusted treatment because of pregnancy or complications. In six cases, the patient decided to postpone treatment until a higher gestational age (GA). Four of these patients were still induced preterm. Palliative care was the only option for six patients. For one patient, data on therapeutic decision-making was not available.

Obstetrical outcome was available for all 26 patients with detailed case information (Fig. 1). Five pregnancies were complicated by intra-uterine fetal death: 1) at a GA of 17 weeks, when mother died of an astrocytoma, 2) at a GA of 18 weeks, when mother died of a glioblastoma multiforme, 3) at a GA of 22 weeks after suboptimal intubation during surgery, 4) at a GA of 27 weeks, when mother was admitted with a status epilepticus due to progressive glioblastoma multiforme, and 5) at a GA of 29 weeks, when the mother died suddenly at home due to metastatic melanoma. One patient experienced an immature delivery at a GA of 18 weeks after a bilateral laparotomic oophorectomy.

Of the 20 ongoing pregnancies ending in a live birth, one was a twin pregnancy resulting in 21 live new-borns. However, one of the children died 4 days after caesarean section at a GA of 26 weeks due to complications of extreme preterm delivery. Caesarean section was performed because of presumed HELLP syndrome. After delivery, the complaints progressed and stage IV cancer of unknown origin with liver metastases was diagnosed.

Median GA at delivery of all 20 ongoing pregnancies was 35 weeks (range 26–40 weeks). Thirteen (65%) of these ended preterm, of which 11 deliveries (85%) were vaginally induced (n = 3) or terminated by caesarean section (n = 8). Reasons for iatrogenic preterm delivery were maternal deterioration (n = 4), therapy planning (n = 4) or obstetrical complications (n = 3).

Ten medical specialists dealing with pregnant cancer patients from six different medical centres independently scored all 26 case descriptions on delay in diagnosis and treatment. The score was considered conclusive if six or more specialists shared the same opinion.

See Table 2 for an overview of the types of delay. In 65% of patients (n = 17) delay was found to be present, 50% (n = 13) or all patients were found to have delay in diagnosis and 27% (n = 7) in starting therapy. Three patients were considered to have both types of delay. Regarding the other nine cases, six showed no signs of delay, two had not enough information available for adequate evaluation and one score was inconclusive. Even in patients with recurrent disease, delay in diagnosis occurred in 50%. In one of these patients with recurrent astrocytoma, delay was caused by doctors in a non-academic centre where symptoms like increasing blood pressure, headache and general seizure were interpreted as eclampsia, for which delivery was induced. In the other patients with delay in diagnosis with recurrent melanoma, progressive back pain and pain on her ribcage at a GA of 35 weeks, was treated with pain medication until 2 weeks after delivery. Diagnostic investigations showed wide spread melanoma metastasis including bones and lungs.

Inter-observer agreement on the presence of delay in diagnosis was found to be substantial, with a Fleiss Kappa score of 0.71. The agreement for the delay in therapy was less, with a score of 0.33, which is a fair agreement.

In the cases where delay in diagnosis was caused, the origin was evaluated to assess whether the delay was caused by the involved health care provider or by the patient. In eight of 13 patients with delay in diagnosis, doctor-related delay was considered to have influenced the time between presentation of symptoms and diagnosis. In two additional cases, both the health care provider and the patient caused delay. For the ten patients with doctor-related delay, delay was caused by the midwife (n = 5), general practitioner (n = 2) and by doctors from a non-academic centres (n = 2) and in two patients, multiple non-academic health care workers caused delay. In nine cases, symptoms were described as pregnancy related complaints, and in one patient, a wrong diagnosis was made based on additional diagnostic tests. In one patient, delay was caused by a long period between referral to a hospital and actual diagnostic evaluation. Patient-related delay in diagnosis was present in only two cases, and in one case enough information was not available to evaluate the cause of delay.

Doctor-related delay in therapy was found in four of the seven cases. In the other three cases, patients refused the advised treatment during pregnancy and wished to delay until after delivery.