Samba II PCR testing for COVID-19 in an unselected cohort of pregnant women in the UK

OBJECTIVE: Asymptomatic carriage of COVID-19 in pregnant women has been reported and could lead to outbreaks in maternity units. We sought to ascertain the impact of rapid isothernal nucleic acid based testing for COVID-19 in an unselected cohort of pregnant women attending our maternity unit. We also assessed the correlation between community prevalence and asymptomatic carriage. DESIGN: review UK POPULATION: Pregnant women (with gestation age between 12-42 weeks) admitted to a single maternity unit over a 4-week period (07/05/2020-06/06/2020) METHODS: Data were collected using computerised hospital records. Literature searches were performed across multiple repositories. COVID-19 prevalence was extracted from online repositories. RESULTS: NP and OP swabs were obtained from 457/465 women during the study period (98%). The median turnaround time for results was 5.3 hours (interquartile range (IQR) 2.6-8.9 hours). 92% of results were returned within 24 hours. In our cohort, only one woman tested positive, giving a screen positive rate of 0.22% (1/457; 95% condence interval: 0.04-1.23%). One woman who tested negative developed a fever postnatally following discharge but was lost to follow-up. From our literature review, we did not nd any correlation between asymptomatic carriage in pregnant women and the reported regional prevalence of COVID-19. CONCLUSIONS: Testing using the SAMBA-II machine was acceptable to the vast majority of pregnant women requiring admission and had a low turnaround time. Asymptomatic carriage is low, but not correlated to community prevalence rates. Screening pregnant women on admission will remain an important component in order to minimise nosocomial infection.


Introduction
The World Health Organization (WHO) characterized Coronavirus disease 2019 (COVID-19) as a global pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in March 2020. [1] The rst case series was described by Huang et al [2] where 41 patients were admitted to Jin Yin-Tan Hospital in Wuhan, China. 98% of the 41 con rmed cases had fever and 76% had a dry cough, all developed pneumonia. Deep sequencing of lower respiratory tract samples identi ed the viral genome to be from the coronavirus family and this was named . Cellular access is gained through a spike protein that binds to the angiotensin-converting enzyme 2 receptor (ACE2), and viral uptake is promoted by the type 2 transmembrane serine protease (TMPRSS2). [3] Lauer reported a median 5.7 days to fever onset (CI 4.9-6.8 days), and 97.5% of cases had a fever within 12.5 days (CI 8.2-17.7 days of exposure). [4] Importantly, COVID-19 often takes a prolonged disease course with viral load in throat and nose declining after the rst week as antibody titres rise. [5] This can result in nucleic acid testing becoming negative after the rst week, though neutralisation and SARS-CoV-2 antibodies are readily detectable in nearly all cases [6]. Asymptomatic carriage of COVID-19 was estimated to be 15% (95% CI 10 to 22%) in general population [7] and Sutton et al [8] reported that positive rate of 13.7% (29/210) among asymptomatic group and 87.9% (29/33) of pregnant women infected with SARS-CoV-2 were asymptomatic at presentation.
Identi cation of asymptomatic pregnant women with COVID-19 is important for several reasons. Firstly, identi cation of asymptomatic carriers would allow changes in the pathway of care so as to prevent nosocomial infections, reducing the risk of asymptomatic transmission to other pregnant women and also to healthcare workers (HCW). Testing also enables early isolation and rationalisation of personal protective equipment (PPE). Thirdly, identi cation of viral carriage could allow closer monitoring both during and after delivery, consideration of low molecular heparin for prophylaxis of venous thromboembolism, and longer term follow-up.
SAMBA II is an isothermal point of care nucleic acid ampli cation based platform with a detection limit of around 250 genome copies/ml. [9] It has previously been clinically validated in parallel with standard reverse transcription polymerase chain reaction (RT-PCR) with sensitivity and speci city of 96.9 % (95% CI 0.838-0.999) and 99.1% (0.953-0.999) for COVID-19, respectively. [10] The median time to result has been signi cantly reduced from 26.4 hours (IQR 21.4 to 31.4) for the standard lab RT-PCR test to 2.6 hours (IQR 2.3 to 4.8) for SAMBA II SARS-CoV-2 test. [11] Aims The aims of this paper are two-fold: to ascertain the incidence of asymptomatic carriage in our maternity unit using a rapid diagnostic testing platform; and to assess if the rate of carriage of COVID-19 in asymptomatic pregnant women correlated with peak local prevalence.

Cohort study
Cambridge University Hospitals NHS Foundation Trust (CUH) covers a large geographical area with a total population of approximately 5 million people in the east of England. The Rosie Maternity Hospital is part of CUH, with an annual delivery rate of approximately 5500 women. All admissions to the Rosie Maternity Hospital were offered a nasopharyngeal (NP) and oropharyngeal (OP) swab for COVID-19. All staff members received training prior to obtaining swabs. Samples were processed using the SAMBA II machine (Diagnostics for the Real World, Chesterford, UK). [12,13] Data were extracted for population demographics and symptomatology. Symptoms were de ned as fever and or cough. Patients who were asymptomatic were de ned as women who lacked a fever and/or cough from the time of testing to discharge. Turnaround time was de ned as the time the swab was collected, to the time a result was uploaded onto the electronic hospital record. Where time entries were missing, electronic hospital notes were retrieved to check for a record of sample collection and availability of results. These were computed separately. Our study was registered as a service evaluation project and ethical approval was not required.

Literature review
We conducted a literature search from PubMed, Cochrane COVID-19 trials for published studies, MedRxiv for pre-prints and an uno cial online repository on 15/05/2020 and updated on 07/08/2020. Where datasets from the same institution were replicated in more than one publication, but with a larger sample size or longer duration, the publication with the largest sample size and/or duration was selected. Data were extracted for case de nition, diagnostic test employed, duration of follow up, gestational age at the time of testing and turnaround time. In order to assess if the prevalence of COVID-19 positive symptomatic and asymptomatic pregnant women correlated with local prevalence rates, we extracted local population COVID-19 prevalence rates for each of the included studies from open source repositories (Supplementary information). Data were analysed using GraphPad Prism version 5.00 for Windows, GraphPad Software, San Diego California USA, www.graphpad.com. Data were extracted by two reviewers (RX and HC), and disagreements were resolved by discussion with a 3 rd reviewer (TP).

SAMBA-II results
During the period 07/05 -06/06 (4 weeks) 465 women attended the Rosie Maternity Hospital for obstetric indications. NP and OP swabs were obtained from 457/465 women upon admission (98%). 18 (3.84%) women had more than one swab taken either due to prolonged inpatient admission beyond 7 days, or reattendance within the 4 week period. Self-reported ethnicity was available for 99.5% of women. The majority of our population consisted of White British, Irish or European ancestery (407/465, 87.5%) and this is re ective of the local population in Cambridgeshire. The median gestational age at the time of the admission swab was 39 weeks (Interquartile range (IQR) 37-40 weeks). Six women (1.3%) attended in the postnatal period due to obstetric concerns. The median duration of inpatient stay for all patients was two days (IQR 1-3). The median duration of follow-up in the post-natal period was 5 days (IQR 3-9).
None of the women had a cough on admission. 37/465 women developed a fever either during labour or following delivery (7.9%). All 37 women had a negative result on admission, and the fever was attributed to an obstetric cause. Two women were admitted with breathlessness on a background of cardiac disease. One woman developed symptoms of fever or sore throat four days after delivery. Advice from Public Health England at the time was to contact a national helpline and she was lost to follow up.
Only one woman in our cohort tested positive over this four-week interval. This woman had neither cough nor fever on admission, but had symptoms of altered taste and smell four weeks prior to her admission. To ensure that this was a true positive, swabs were repeated (but not included in our current analyses for TAT) until she tested negative. As this patient remained symptom-free during admission, she would be the only true asymptomatic COVID-19, giving an asymptomatic carriage rate of 0.22% (1/457; 95% con dence interval: 0.04-1.23%). Knowledge of her results enabled HCW to convert from using standard surgical masks, plastic aprons and gloves to full personal protective equipment when caring for this patient.
Result turnaround times (TAT) were veri ed and available for 432/457 (94.5%) samples. The median TAT was 5.3 hours (IQR 2.6-8.9) (Figure 1). 424/457 (92.7%) were returned within 24 hours. Where sample turnaround times were absent, these were due to an underestimation of turnaround times rather than a delay in sample processing. This information was gleaned from verbatim entries in the patient records regarding the sample obtained, and the results entered. As no speci c record of time was entered, these were excluded from further analyses. A small percentage (16/465, 3.4%) of women declined to be swabbed on admission. 5/16 (31.2%) women were in discomfort from labour and declined testing. No reasons were given for the remaining 11 women who declined testing.
We found a correlation between the test positivity with the regional background prevalence rates of COVID-19. There was a direct correlation between the number of con rmed cases at the peak (R 2 =0.41, p=0.0053) ( Figure 2B) and end of the study period with those who tested positive (R 2 =0.48, p=0.002) ( Figure 2C), but not with the number of con rmed cases at the start of the study period (R 2 =0.004, p=0.82) (Figure 2A). There was no correlation between the proportions of asymptomatic pregnant carriers with background infection rates (Figure 3).
Only three papers reported turnaround time. Breslin et al [31] reported an average of 8 hours for 43 women screened [Conventional PCR, (New York, New York)] and LaCourse et al [25] reported a median turnaround time of 2.5 hours for rapid testing [DiaSorin Simplexa (MDX Liaison) EUA assay, n=82 women] and 7.1 hours for routine PCR (Seattle, Washington). London et al [22] reported an average turnaround time of 5 hours using GeneXpert PCR in a cohort of 75 women (Brooklyn, New York).

Discussion
Main ndings: Our work summarises existing literature on universal screening of asymptomatic carriage of COVID-19 in pregnant women with speci c reference to the turnaround time of results, and local prevalence rates. The key ndings of our study are threefold. We captured data on all attendances in a large UK maternity hospital and ours is the largest, to have reported turnaround times of less than 24 hours in over 90% of samples returned. We were also able to demonstrate a high acceptability of COVID-19 screening amongst our maternity population with only 3% declining to be tested. Thirdly, ours is the rst study to demonstrate the utility of point of care testing in an unselected cohort of pregnant women attending a large UK maternity unit.
Whilst the data are re ective of our maternity service, laboratory services were also processing samples from patients attending the accident and emergency department, and medical and surgical wards within a national health service. We were still able to achieve a short turnaround time of ve hours. This allowed for changes to be made within postnatal care pathway, and escalation of the use of personal protective equipment. Testing of newborns could have also been performed had it been required.
The low positive screening rate in our cohort may be partly explained by the low rates of infection in the east of England (cumulative infection rate 0.42%), in comparison to 1.27% in California and 2.12% in New York. As the pandemic evolved, advice and guidance were issued to pregnant women in the UK, who may have adopted behavioural changes that minimised interaction with the general public. Additionally, home working and minimising the commute to work could have protected them further. Interestingly, the proportion of asymptomatic carriers did not correlate with the regional infection rate. Many hospitals introduced visiting restrictions during the lockdown, and ours was no different. Partners were not allowed to attend the antenatal ward nor ultrasound scan appointments. Obstetric and non-obstetric face-to-face clinic appointments were changed to phone appointments where possible, thus reducing footfall within the hospital premises. Staff working in clinical areas were required to wear personal protective equipment in the form of a surgical mask, gloves and aprons as a minimum, and FFP3 masks for aerosol generating procedures. Thus, pregnant women attending hospital should also be reassured, not only that asymptomatic carriage of COVID-19 is low, infection within the inpatient setting is low. Whether or not this is maintained against a backdrop of rising COVID-19 infections rates is unknown.

Strengths And Limitations
We did not compare the test accuracy of the Samba-II machine in our population. However, this device has been tested previously in a cohort of over 1000 individuals and found to have 97% accuracy to conventional RT-PCR testing. There is no reason to believe why it should perform differently in pregnant women. In contrast, GeneXpert PCR testing has been previously been validated on a much smaller cohort of less than 50 women. [32] We did not perform radiological investigations to look for manifestations of COVID-19 pneumonia, as some others have clinical diagnosis criteria as well as laboratory diagnosis. [33,34] Women could therefore have a negative NP or OP result for COVID-19, but have radiological changes. However, the likelihood of this is low. Although we had a high acceptability rate, we did not explore women's and staff views on screening for COVID-19, thus further research is required to evaluate this. For example, there may be a need to develop tests which are acceptable to women in labour, or training for hospital staff in contact tracing where women simply declined to be tested. In the interest of patient safety, we would suggest a conservative approach to patient pathways for women who decline testing so as not to result in contagion within the hospital setting.

Conclusion
The Royal College for Obstetricians and Gynaecologists in the UK have recently published guidelines on testing for asymptomatic pregnant women attending maternity units. [35] These are broadly in line with our current practice but for the type of swab being offered (conventional lab based RT-PCR as opposed to rapid testing with the SAMBA II). Owing to the variable rates of asymptomatic carriage of COVID-19 in pregnant women, and the rapid turnaround time using the SAMBA II machine, we propose the introduction of SAMBA II or other point of care testing platforms in maternity units as an accurate and acceptable test to pregnant women requiring admission. This work was registered as service evaluation project (ID 3079). All women booked for maternity care at the Rosie Maternity Hospital are asked verbally whether they give their consent for data to be used for research and this is recorded in their computerised hospital records. We can therefore con rm that this is a formally approved method of consenting for use of data in research at the Rosie Maternity Unit. Identi able data were removed from cases to ensure anonymity. Identi able data were removed from cases to ensure anonymity. Additionally, in accordance with the United Kingdom National Health Service National Research Ethics Service guidance, neither individual informed consent nor formal research ethics committee review was required, because the study was undertaken by the direct clinical team using information collected in the course of routine care.

Consent for publication
Not applicable Availability of data and materials The datasets generated for the cohort study are not publically available due to current UK regulations with regards to data protection, but are available from the corresponding author on reasonable request. The datasets generated for the peak coronavirus prevalence rates can be found from international repositories listed in the references.

Competing interests
All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no nancial relationships with any organisations that might have an interest in the submitted work in the previous three years, no other relationships or activities that could appear to have in uenced the submitted work. Completed disclosure of interest forms are available to view online as supporting information.
Contribution to authorship HC conceived the study. HML, RG, RH and HC designed the study. RX and HC screened titles and abstracts for inclusion and literature review. HC, RX and TP extracted and analysed data. HML helped interpret the study ndings from a clinical viewpoint. HC and RX wrote the rst draft, which all authors revised for critical content. All authors approved the nal manuscript. HC is the guarantor. The guarantor had full access to all the data in the study, take responsibility for the integrity of the data and the accuracy of the data analysis, and had nal responsibility for the decision to submit for publication. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding: None