1. Epidemiology and Global Health
  2. Microbiology and Infectious Disease
Download icon

Single-dose BNT162b2 vaccine protects against asymptomatic SARS-CoV-2 infection

  1. Nick K Jones
  2. Lucy Rivett
  3. Shaun Seaman
  4. Richard J Samworth
  5. Ben Warne
  6. Chris Workman
  7. Mark Ferris
  8. Jo Wright
  9. Natalie Quinnell
  10. Ashley Shaw
  11. Cambridge COVID-19 Collaboration
  12. Ian G Goodfellow
  13. Paul J Lehner
  14. Rob Howes
  15. Giles Wright
  16. Nicholas J Matheson
  17. Michael P Weekes  Is a corresponding author
  1. Cambridge University NHS Hospitals Foundation Trust, United Kingdom
  2. Clinical Microbiology & Public Health Laboratory, Public Health England, United Kingdom
  3. Medical Research Council Biostatistics Unit, University of Cambridge, United Kingdom
  4. Statistical Laboratory, Centre for Mathematical Sciences, University of Cambridge, United Kingdom
  5. Cambridge Institute of Therapeutic Immunology & Infectious Disease, University of Cambridge, United Kingdom
  6. Department of Medicine, University of Cambridge, United Kingdom
  7. Occupational Health and Wellbeing, Cambridge Biomedical Campus, United Kingdom
  8. Division of Virology, Department of Pathology, University of Cambridge, United Kingdom
  9. Cambridge COVID-19 Testing Centre and AstraZeneca, Anne McLaren Building, United Kingdom
  10. NHS Blood and Transplant, United Kingdom
  11. Cambridge Institute for Medical Research, University of Cambridge, United Kingdom
Research Advance
  • Cited 19
  • Views 4,642
  • Annotations
Cite this article as: eLife 2021;10:e68808 doi: 10.7554/eLife.68808

Abstract

The BNT162b2 mRNA COVID-19 vaccine (Pfizer-BioNTech) is being utilised internationally for mass COVID-19 vaccination. Evidence of single-dose protection against symptomatic disease has encouraged some countries to opt for delayed booster doses of BNT162b2, but the effect of this strategy on rates of asymptomatic SARS-CoV-2 infection remains unknown. We previously demonstrated frequent pauci- and asymptomatic SARS-CoV-2 infection amongst healthcare workers (HCWs) during the UK’s first wave of the COVID-19 pandemic, using a comprehensive PCR-based HCW screening programme (Rivett et al., 2020; Jones et al., 2020). Here, we evaluate the effect of first-dose BNT162b2 vaccination on test positivity rates and find a fourfold reduction in asymptomatic infection amongst HCWs ≥12 days post-vaccination. These data provide real-world evidence of short-term protection against asymptomatic SARS-CoV-2 infection following a single dose of BNT162b2 vaccine, suggesting that mass first-dose vaccination will reduce SARS-CoV-2 transmission, as well as the burden of COVID-19 disease.

Introduction

The UK has initiated mass COVID-19 immunisation, with healthcare workers (HCWs) given early priority because of the potential for workplace exposure and risk of onward transmission to patients. The UK’s Joint Committee on Vaccination and Immunisation has recommended maximising the number of people vaccinated with first doses at the expense of early booster vaccinations, based on single-dose efficacy against symptomatic COVID-19 disease (Department of Health and Social Care, 2021; Polack et al., 2020; Voysey et al., 2021).

At the time of writing, three COVID-19 vaccines have been granted emergency use authorisation in the UK, including the BNT162b2 mRNA COVID-19 vaccine (Pfizer-BioNTech). A vital outstanding question is whether this vaccine prevents asymptomatic as well as symptomatic SARS-CoV-2 infection or merely converts infections from symptomatic to asymptomatic. Sub-clinical infection following vaccination could continue to drive transmission. This is especially important because many UK HCWs have received this vaccine, and nosocomial COVID-19 infection has been a persistent problem.

Through the implementation of a 24 hour turnaround PCR-based comprehensive HCW screening programme at Cambridge University Hospitals NHS Foundation Trust (CUHNFT), we previously demonstrated the frequent presence of pauci- and asymptomatic infection amongst HCWs during the UK’s first wave of the COVID-19 pandemic (Rivett et al., 2020). Here, we evaluate the effect of first-dose BNT162b2 vaccination on test positivity rates and cycle threshold (Ct) values in the asymptomatic arm of our programme, which now offers weekly screening to all staff.

Results and discussion

Vaccination of HCWs at CUHNFT began on 8 December 2020, with mass vaccination from 8 January 2021. Here, we analyse data from 2 weeks spanning 18–31 January 2021, during which (1) the prevalence of COVID-19 amongst HCWs remained approximately constant and (2) we screened comparable numbers of vaccinated and unvaccinated HCWs. During this period, 4408 (week 1) and 4411 (week 2) PCR tests were performed on individuals reporting well to work, from a weekly on-site HCW population of ~9000. We stratified HCWs <12 days or ≥12 days post-vaccination because this was the point at which protection against symptomatic infection began to appear in the phase III clinical trial (Polack et al., 2020). In the post-vaccination groups, the median number of days between vaccination and testing were 7 (interquartile range [IQR] 4–9; <12 day group) and 16 (14–18; ≥12 day group).

Twenty-six of 3252 (0.8%, Wilson’s interval 0.6–1.2%) tests from unvaccinated HCWs were positive (Ct < 36), compared to 13/3535 (0.4%, Wilson’s interval 0.2–0.6%) tests from HCWs <12 days post-vaccination and 4/1989 (0.2%, Wilson’s interval 0.1–0.5%) tests from HCWs ≥12 days post-vaccination (p=0.023 and p=0.004, respectively; Fisher’s exact test, Figure 1 and Table 1). This suggests a fourfold decrease in the risk of asymptomatic SARS-CoV-2 infection amongst HCWs ≥12 days post-vaccination, compared to unvaccinated HCWs, with an intermediate effect amongst HCWs <12 days post-vaccination.

Proportion of positive screening tests for SARS-CoV-2 amongst HCWs from the CUHNHFT asymptomatic screening programme (grey bars; week 1, 18–24 January 2021; week 2, 25–31 January 2021) and Ct values of positive tests (Ct < 36; blue dots; both weeks).

RT-PCR targeting the SARS-CoV-2 ORF1ab genes was conducted at the Cambridge COVID-19 Testing Centre (part of the UK Lighthouse Labs Network). For proportions of positive screening tests, p-values for pair-wise comparisons of unvaccinated HCWs with HCWs <12 days or ≥12 days post-vaccination are shown (Fisher’s exact test; both weeks). For Ct values, medians ± interquartile ranges are shown.

Figure 1—source data 1

Proportions of positive asymptomatic SARS-CoV-2 screens and distributions of Ct values.

https://cdn.elifesciences.org/articles/68808/elife-68808-fig1-data1-v2.xlsx
Table 1
Weekly numbers and proportions of positive SARS-CoV-2 test results spanning 6 weeks around the main study period (indicated in grey).
Unvaccinated<12 Days since vaccination≥12 Days since vaccination
Week startTotal testsPositive tests%Total testsPositive tests%Total testsPositive tests%
28 December 20202097160.8%800.0%600.0%
4 January 20214762430.9%9300.0%2200.0%
11 January 20213273270.8%97860.6%3000.0%
18 January 20212183170.8%171680.5%48310.2%
25 January 2021106990.8%181950.3%150630.2%
1 February 202169910.1%75810.1%282510.0%

A marked reduction in infections was also seen when analyses were repeated with (1) inclusion of HCWs testing positive through both the symptomatic and asymptomatic arms of the programme (56/3370 [1.7%, Wilson’s interval 1.3–2.2%] unvaccinated vs 8/2018 [0.4%, Wilson’s interval 0.2–0.8%] ≥12 days post-vaccination, 4.2-fold reduction, p<0.0001) and (2) inclusion of PCR tests that were positive at the limit of detection (Ct > 36, 42/3268 [1.3%, Wilson’s interval 1.0–1.7%] vs 15/2000 [0.7%, Wilson’s interval 0.5–1.2%], 1.7-fold reduction, p=0.07). In addition, the median Ct value of positive tests showed a non-significant trend towards increase between unvaccinated HCWs and HCWs ≥12 days post-vaccination (23.3 [IQR 13.5–33.0] to 30.3 [IQR 25.5–35.1], Figure 1), raising the possibility that vaccinated individuals who do go on to develop infection may have lower viral loads.

HCWs working in COVID-19 clinical areas were prioritised for vaccination, and a small number of clinically vulnerable HCWs were also given priority. Otherwise, vaccine allocation was arbitrary. Since asymptomatic infection was examined, the date of infection could have been earlier than the test date. These factors would all tend to lead to an underestimate of the vaccine’s effect (bias towards the null). Because of the rapid decline in the incidence of SARS-CoV-2 infection in the Cambridge community, this study could only examine the short-term impact of single-dose BNT162b2 vaccination. The frequency of prior SARS-CoV-2 infection (Cooper et al., 2020) was similar in all groups (seroprevalence 7.1%, unvaccinated; 5.6%, <12 days post-vaccination; 5.7%, ≥12 days post-vaccination), suggesting that this did not confound our observations.

Taken together, our findings provide real-world evidence of short-term protection against asymptomatic SARS-CoV-2 infection after a single dose of BNT162b2 vaccine, at a time when the UK COVID-19 variant of concern 202012/01 (lineage B.1.1.7) accounted for the great majority of infections (24/29 sequenced isolates from asymptomatic HCWs). A fourfold reduction from 0.8% to 0.2% in asymptomatic infection is likely to be crucial in controlling nosocomial SARS-CoV-2 transmission. Nonetheless, protection is incomplete, suggesting that continuing asymptomatic HCW screening, social distancing, mask-wearing, and strict hand hygiene remain vital.

Materials and methods

HCW screening programme

Request a detailed protocol

We previously described protocols for staff screening, sample collection, and results reporting in detail (Rivett et al., 2020; Jones et al., 2020). In general, these methods remained unchanged throughout this study period. Two parallel streams of entry into the testing programme included (1) HCW symptomatic and HCW symptomatic household contact screening arms and (2) an HCW asymptomatic screening arm. Since our prior description of the screening programme, weekly asymptomatic testing is now offered to all CUHNFT staff. Testing was performed (1) at temporary on-site ‘Pods’ and (2) via self-swabbing kits collected by HCWs. Individuals performed a self-swab of the oropharynx and anterior nasal cavity. Samples were subjected to RNA extraction and amplification using real-time RT-PCR, with all sample processing and analysis undertaken at the Cambridge COVID-19 Testing Centre (Lighthouse Laboratory).

Vaccination

Request a detailed protocol

HCW vaccination began at CUHNFT on 8 December 2020, with appointments made by invitation only for all high-risk HCWs working on-site. This was followed by self-booked appointments for HCWs working in designated COVID-19 clinical areas, from 8 January 2021 onwards. From 18 January 2021, vaccination was offered to all HCWs, with appointments made via a booking website and latterly using the hospital’s electronic patient record system ‘MyChart’. All vials of Pfizer-BioNTech COVID-19 Vaccine (BNT162b2) were stored at −74°C, before being transferred to storage at 2–8°C. From the moment the vials were removed from the freezer, they were given a 120 hr expiration date, of which 3 hr were dedicated to thawing the vaccines. All vaccine doses were administered intramuscularly by trained vaccinators, in accordance with the manufacturer’s instructions. Vaccination was undertaken exclusively at an on-site vaccination centre, with mandatory mask-wearing and social distancing in place. HCWs remained at the on-site vaccination centre for a minimum observation period of 15 min after vaccination.

Data extraction and analysis

Request a detailed protocol

Swab result, vaccination details, and serology data for HCWs were extracted directly from the hospital-laboratory interface software, Epic (Verona, WI). Data were collated using Microsoft Excel and the figure produced with GraphPad Prism (GraphPad Software, La Jolla, CA). Fisher’s exact test was used for the comparison of positive rates between groups, defined in the main text. Additionally, 95% confidence intervals were calculated using Wilson’s method.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Source data file has been provided for Figure 1.

References

    1. Voysey M
    2. Clemens SAC
    3. Madhi SA
    4. Weckx LY
    5. Folegatti PM
    6. Aley PK
    7. Angus B
    8. Baillie VL
    9. Barnabas SL
    10. Bhorat QE
    11. Bibi S
    12. Briner C
    13. Cicconi P
    14. Collins AM
    15. Colin-Jones R
    16. Cutland CL
    17. Darton TC
    18. Dheda K
    19. Duncan CJA
    20. Emary KRW
    21. Ewer KJ
    22. Fairlie L
    23. Faust SN
    24. Feng S
    25. Ferreira DM
    26. Finn A
    27. Goodman AL
    28. Green CM
    29. Green CA
    30. Heath PT
    31. Hill C
    32. Hill H
    33. Hirsch I
    34. Hodgson SHC
    35. Izu A
    36. Jackson S
    37. Jenkin D
    38. Joe CCD
    39. Kerridge S
    40. Koen A
    41. Kwatra G
    42. Lazarus R
    43. Lawrie AM
    44. Lelliott A
    45. Libri V
    46. Lillie PJ
    47. Mallory R
    48. Mendes AVA
    49. Milan EP
    50. Minassian AM
    51. McGregor A
    52. Morrison H
    53. Mujadidi YF
    54. Nana A
    55. O’Reilly PJ
    56. Padayachee SD
    57. Pittella A
    58. Plested E
    59. Pollock KM
    60. Ramasamy MN
    61. Rhead S
    62. Schwarzbold AV
    63. Singh N
    64. Smith A
    65. Song R
    66. Snape MD
    67. Sprinz E
    68. Sutherland RK
    69. Tarrant R
    70. Thomson EC
    71. Török ME
    72. Toshner M
    73. Turner DPJ
    74. Vekemans J
    75. Villafana TL
    76. Watson MEE
    77. Williams CJ
    78. Douglas AD
    79. Hill AVS
    80. Lambe T
    81. Gilbert SC
    82. Pollard AJ
    83. Aban M
    84. Abayomi F
    85. Abeyskera K
    86. Aboagye J
    87. Adam M
    88. Adams K
    89. Adamson J
    90. Adelaja YA
    91. Adewetan G
    92. Adlou S
    93. Ahmed K
    94. Akhalwaya Y
    95. Akhalwaya S
    96. Alcock A
    97. Ali A
    98. Allen ER
    99. Allen L
    100. Almeida T
    101. Alves MPS
    102. Amorim F
    103. Andritsou F
    104. Anslow R
    105. Appleby M
    106. Arbe-Barnes EH
    107. Ariaans MP
    108. Arns B
    109. Arruda L
    110. Azi P
    111. Azi L
    112. Babbage G
    113. Bailey C
    114. Baker KF
    115. Baker M
    116. Baker N
    117. Baker P
    118. Baldwin L
    119. Baleanu I
    120. Bandeira D
    121. Bara A
    122. Barbosa MAS
    123. Barker D
    124. Barlow GD
    125. Barnes E
    126. Barr AS
    127. Barrett JR
    128. Barrett J
    129. Bates L
    130. Batten A
    131. Beadon K
    132. Beales E
    133. Beckley R
    134. Belij-Rammerstorfer S
    135. Bell J
    136. Bellamy D
    137. Bellei N
    138. Belton S
    139. Berg A
    140. Bermejo L
    141. Berrie E
    142. Berry L
    143. Berzenyi D
    144. Beveridge A
    145. Bewley KR
    146. Bexhell H
    147. Bhikha S
    148. Bhorat AE
    149. Bhorat ZE
    150. Bijker E
    151. Birch G
    152. Birch S
    153. Bird A
    154. Bird O
    155. Bisnauthsing K
    156. Bittaye M
    157. Blackstone K
    158. Blackwell L
    159. Bletchly H
    160. Blundell CL
    161. Blundell SR
    162. Bodalia P
    163. Boettger BC
    164. Bolam E
    165. Boland E
    166. Bormans D
    167. Borthwick N
    168. Bowring F
    169. Boyd A
    170. Bradley P
    171. Brenner T
    172. Brown P
    173. Brown C
    174. Brown-O'Sullivan C
    175. Bruce S
    176. Brunt E
    177. Buchan R
    178. Budd W
    179. Bulbulia YA
    180. Bull M
    181. Burbage J
    182. Burhan H
    183. Burn A
    184. Buttigieg KR
    185. Byard N
    186. Cabera Puig I
    187. Calderon G
    188. Calvert A
    189. Camara S
    190. Cao M
    191. Cappuccini F
    192. Cardoso JR
    193. Carr M
    194. Carroll MW
    195. Carson-Stevens A
    196. Carvalho YdeM
    197. Carvalho JAM
    198. Casey HR
    199. Cashen P
    200. Castro T
    201. Castro LC
    202. Cathie K
    203. Cavey A
    204. Cerbino-Neto J
    205. Chadwick J
    206. Chapman D
    207. Charlton S
    208. Chelysheva I
    209. Chester O
    210. Chita S
    211. Cho J-S
    212. Cifuentes L
    213. Clark E
    214. Clark M
    215. Clarke A
    216. Clutterbuck EA
    217. Collins SLK
    218. Conlon CP
    219. Connarty S
    220. Coombes N
    221. Cooper C
    222. Cooper R
    223. Cornelissen L
    224. Corrah T
    225. Cosgrove C
    226. Cox T
    227. Crocker WEM
    228. Crosbie S
    229. Cullen L
    230. Cullen D
    231. Cunha DRMF
    232. Cunningham C
    233. Cuthbertson FC
    234. Da Guarda SNF
    235. da Silva LP
    236. Damratoski BE
    237. Danos Z
    238. Dantas MTDC
    239. Darroch P
    240. Datoo MS
    241. Datta C
    242. Davids M
    243. Davies SL
    244. Davies H
    245. Davis E
    246. Davis J
    247. Davis J
    248. De Nobrega MMD
    249. De Oliveira Kalid LM
    250. Dearlove D
    251. Demissie T
    252. Desai A
    253. Di Marco S
    254. Di Maso C
    255. Dinelli MIS
    256. Dinesh T
    257. Docksey C
    258. Dold C
    259. Dong T
    260. Donnellan FR
    261. Dos Santos T
    262. dos Santos TG
    263. Dos Santos EP
    264. Douglas N
    265. Downing C
    266. Drake J
    267. Drake-Brockman R
    268. Driver K
    269. Drury R
    270. Dunachie SJ
    271. Durham BS
    272. Dutra L
    273. Easom NJW
    274. van Eck S
    275. Edwards M
    276. Edwards NJ
    277. El Muhanna OM
    278. Elias SC
    279. Elmore M
    280. English M
    281. Esmail A
    282. Essack YM
    283. Farmer E
    284. Farooq M
    285. Farrar M
    286. Farrugia L
    287. Faulkner B
    288. Fedosyuk S
    289. Felle S
    290. Feng S
    291. Ferreira Da Silva C
    292. Field S
    293. Fisher R
    294. Flaxman A
    295. Fletcher J
    296. Fofie H
    297. Fok H
    298. Ford KJ
    299. Fowler J
    300. Fraiman PHA
    301. Francis E
    302. Franco MM
    303. Frater J
    304. Freire MSM
    305. Fry SH
    306. Fudge S
    307. Furze J
    308. Fuskova M
    309. Galian-Rubio P
    310. Galiza E
    311. Garlant H
    312. Gavrila M
    313. Geddes A
    314. Gibbons KA
    315. Gilbride C
    316. Gill H
    317. Glynn S
    318. Godwin K
    319. Gokani K
    320. Goldoni UC
    321. Goncalves M
    322. Gonzalez IGS
    323. Goodwin J
    324. Goondiwala A
    325. Gordon-Quayle K
    326. Gorini G
    327. Grab J
    328. Gracie L
    329. Greenland M
    330. Greenwood N
    331. Greffrath J
    332. Groenewald MM
    333. Grossi L
    334. Gupta G
    335. Hackett M
    336. Hallis B
    337. Hamaluba M
    338. Hamilton E
    339. Hamlyn J
    340. Hammersley D
    341. Hanrath AT
    342. Hanumunthadu B
    343. Harris SA
    344. Harris C
    345. Harris T
    346. Harrison TD
    347. Harrison D
    348. Hart TC
    349. Hartnell B
    350. Hassan S
    351. Haughney J
    352. Hawkins S
    353. Hay J
    354. Head I
    355. Henry J
    356. Hermosin Herrera M
    357. Hettle DB
    358. Hill J
    359. Hodges G
    360. Horne E
    361. Hou MM
    362. Houlihan C
    363. Howe E
    364. Howell N
    365. Humphreys J
    366. Humphries HE
    367. Hurley K
    368. Huson C
    369. Hyder-Wright A
    370. Hyams C
    371. Ikram S
    372. Ishwarbhai A
    373. Ivan M
    374. Iveson P
    375. Iyer V
    376. Jackson F
    377. De Jager J
    378. Jaumdally S
    379. Jeffers H
    380. Jesudason N
    381. Jones B
    382. Jones K
    383. Jones E
    384. Jones C
    385. Jorge MR
    386. Jose A
    387. Joshi A
    388. Júnior EAMS
    389. Kadziola J
    390. Kailath R
    391. Kana F
    392. Karampatsas K
    393. Kasanyinga M
    394. Keen J
    395. Kelly EJ
    396. Kelly DM
    397. Kelly D
    398. Kelly S
    399. Kerr D
    400. Kfouri RA
    401. Khan L
    402. Khozoee B
    403. Kidd S
    404. Killen A
    405. Kinch J
    406. Kinch P
    407. King LDW
    408. King TB
    409. Kingham L
    410. Klenerman P
    411. Knapper F
    412. Knight JC
    413. Knott D
    414. Koleva S
    415. Lang M
    416. Lang G
    417. Larkworthy CW
    418. Larwood JPJ
    419. Law R
    420. Lazarus EM
    421. Leach A
    422. Lees EA
    423. Lemm N-M
    424. Lessa A
    425. Leung S
    426. Li Y
    427. Lias AM
    428. Liatsikos K
    429. Linder A
    430. Lipworth S
    431. Liu S
    432. Liu X
    433. Lloyd A
    434. Lloyd S
    435. Loew L
    436. Lopez Ramon R
    437. Lora L
    438. Lowthorpe V
    439. Luz K
    440. MacDonald JC
    441. MacGregor G
    442. Madhavan M
    443. Mainwaring DO
    444. Makambwa E
    445. Makinson R
    446. Malahleha M
    447. Malamatsho R
    448. Mallett G
    449. Mansatta K
    450. Maoko T
    451. Mapetla K
    452. Marchevsky NG
    453. Marinou S
    454. Marlow E
    455. Marques GN
    456. Marriott P
    457. Marshall RP
    458. Marshall JL
    459. Martins FJ
    460. Masenya M
    461. Masilela M
    462. Masters SK
    463. Mathew M
    464. Matlebjane H
    465. Matshidiso K
    466. Mazur O
    467. Mazzella A
    468. McCaughan H
    469. McEwan J
    470. McGlashan J
    471. McInroy L
    472. McIntyre Z
    473. McLenaghan D
    474. McRobert N
    475. McSwiggan S
    476. Megson C
    477. Mehdipour S
    478. Meijs W
    479. Mendonça RNA
    480. Mentzer AJ
    481. Mirtorabi N
    482. Mitton C
    483. Mnyakeni S
    484. Moghaddas F
    485. Molapo K
    486. Moloi M
    487. Moore M
    488. Moraes-Pinto MI
    489. Moran M
    490. Morey E
    491. Morgans R
    492. Morris S
    493. Morris S
    494. Morris HC
    495. Morselli F
    496. Morshead G
    497. Morter R
    498. Mottal L
    499. Moultrie A
    500. Moya N
    501. Mpelembue M
    502. Msomi S
    503. Mugodi Y
    504. Mukhopadhyay E
    505. Muller J
    506. Munro A
    507. Munro C
    508. Murphy S
    509. Mweu P
    510. Myasaki CH
    511. Naik G
    512. Naker K
    513. Nastouli E
    514. Nazir A
    515. Ndlovu B
    516. Neffa F
    517. Njenga C
    518. Noal H
    519. Noé A
    520. Novaes G
    521. Nugent FL
    522. Nunes G
    523. O'Brien K
    524. O'Connor D
    525. Odam M
    526. Oelofse S
    527. Oguti B
    528. Olchawski V
    529. Oldfield NJ
    530. Oliveira MG
    531. Oliveira C
    532. Oosthuizen A
    533. O'Reilly P
    534. Osborne P
    535. Owen DRJ
    536. Owen L
    537. Owens D
    538. Owino N
    539. Pacurar M
    540. Paiva BVB
    541. Palhares EMF
    542. Palmer S
    543. Parkinson S
    544. Parracho HMRT
    545. Parsons K
    546. Patel D
    547. Patel B
    548. Patel F
    549. Patel K
    550. Patrick-Smith M
    551. Payne RO
    552. Peng Y
    553. Penn EJ
    554. Pennington A
    555. Peralta Alvarez MP
    556. Perring J
    557. Perry N
    558. Perumal R
    559. Petkar S
    560. Philip T
    561. Phillips DJ
    562. Phillips J
    563. Phohu MK
    564. Pickup L
    565. Pieterse S
    566. Piper J
    567. Pipini D
    568. Plank M
    569. Du Plessis J
    570. Pollard S
    571. Pooley J
    572. Pooran A
    573. Poulton I
    574. Powers C
    575. Presa FB
    576. Price DA
    577. Price V
    578. Primeira M
    579. Proud PC
    580. Provstgaard-Morys S
    581. Pueschel S
    582. Pulido D
    583. Quaid S
    584. Rabara R
    585. Radford A
    586. Radia K
    587. Rajapaska D
    588. Rajeswaran T
    589. Ramos ASF
    590. Ramos Lopez F
    591. Rampling T
    592. Rand J
    593. Ratcliffe H
    594. Rawlinson T
    595. Rea D
    596. Rees B
    597. Reiné J
    598. Resuello-Dauti M
    599. Reyes Pabon E
    600. Ribiero CM
    601. Ricamara M
    602. Richter A
    603. Ritchie N
    604. Ritchie AJ
    605. Robbins AJ
    606. Roberts H
    607. Robinson RE
    608. Robinson H
    609. Rocchetti TT
    610. Rocha BP
    611. Roche S
    612. Rollier C
    613. Rose L
    614. Ross Russell AL
    615. Rossouw L
    616. Royal S
    617. Rudiansyah I
    618. Ruiz S
    619. Saich S
    620. Sala C
    621. Sale J
    622. Salman AM
    623. Salvador N
    624. Salvador S
    625. Sampaio M
    626. Samson AD
    627. Sanchez-Gonzalez A
    628. Sanders H
    629. Sanders K
    630. Santos E
    631. Santos Guerra MFS
    632. Satti I
    633. Saunders JE
    634. Saunders C
    635. Sayed A
    636. Schim van der Loeff I
    637. Schmid AB
    638. Schofield E
    639. Screaton G
    640. Seddiqi S
    641. Segireddy RR
    642. Senger R
    643. Serrano S
    644. Shah R
    645. Shaik I
    646. Sharpe HE
    647. Sharrocks K
    648. Shaw R
    649. Shea A
    650. Shepherd A
    651. Shepherd JG
    652. Shiham F
    653. Sidhom E
    654. Silk SE
    655. da Silva Moraes AC
    656. Silva-Junior G
    657. Silva-Reyes L
    658. Silveira AD
    659. Silveira MBV
    660. Sinha J
    661. Skelly DT
    662. Smith DC
    663. Smith N
    664. Smith HE
    665. Smith DJ
    666. Smith CC
    667. Soares A
    668. Soares T
    669. Solórzano C
    670. Sorio GL
    671. Sorley K
    672. Sosa-Rodriguez T
    673. Souza CM
    674. Souza BS
    675. Souza AR
    676. Spencer AJ
    677. Spina F
    678. Spoors L
    679. Stafford L
    680. Stamford I
    681. Starinskij I
    682. Stein R
    683. Steven J
    684. Stockdale L
    685. Stockwell LV
    686. Strickland LH
    687. Stuart AC
    688. Sturdy A
    689. Sutton N
    690. Szigeti A
    691. Tahiri-Alaoui A
    692. Tanner R
    693. Taoushanis C
    694. Tarr AW
    695. Taylor K
    696. Taylor U
    697. Taylor IJ
    698. Taylor J
    699. te Water Naude R
    700. Themistocleous Y
    701. Themistocleous A
    702. Thomas M
    703. Thomas K
    704. Thomas TM
    705. Thombrayil A
    706. Thompson F
    707. Thompson A
    708. Thompson K
    709. Thompson A
    710. Thomson J
    711. Thornton-Jones V
    712. Tighe PJ
    713. Tinoco LA
    714. Tiongson G
    715. Tladinyane B
    716. Tomasicchio M
    717. Tomic A
    718. Tonks S
    719. Towner J
    720. Tran N
    721. Tree J
    722. Trillana G
    723. Trinham C
    724. Trivett R
    725. Truby A
    726. Tsheko BL
    727. Turabi A
    728. Turner R
    729. Turner C
    730. Ulaszewska M
    731. Underwood BR
    732. Varughese R
    733. Verbart D
    734. Verheul M
    735. Vichos I
    736. Vieira T
    737. Waddington CS
    738. Walker L
    739. Wallis E
    740. Wand M
    741. Warbick D
    742. Wardell T
    743. Warimwe G
    744. Warren SC
    745. Watkins B
    746. Watson E
    747. Webb S
    748. Webb-Bridges A
    749. Webster A
    750. Welch J
    751. Wells J
    752. West A
    753. White C
    754. White R
    755. Williams P
    756. Williams RL
    757. Winslow R
    758. Woodyer M
    759. Worth AT
    760. Wright D
    761. Wroblewska M
    762. Yao A
    763. Zimmer R
    764. Zizi D
    765. Zuidewind P
    (2021) Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK
    The Lancet 397:99–111.
    https://doi.org/10.1016/S0140-6736(20)32661-1

Decision letter

  1. Jos WM van der Meer
    Senior and Reviewing Editor; Radboud University Medical Centre, Netherlands

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The investigators expanded their healthcare worker (HCW) asymptomatic screening programme by offering weekly testing to all staff at Cambridge University Hospitals. They also rapidly rolled out vaccination of the staff with the BNT162b2 mRNA vaccine (Pfizer-BioNTech). This provided the opportunity to investigate whether the vaccination protects against infection (i.e., transmission), or just abrogates COVID-19 disease. This is particularly important, because the vaccine has been widely administered to HCWs, and nosocomial infections have blighted hospitals throughout the pandemic. Published clinical trials data have so far only addressed whether this vaccine prevents symptomatic COVID-19. Here, the investigators show that asymptomatic SARS-COV-2 infection is also markedly reduced in a real-world setting, with four-fold fewer infections 12 days post-vaccination onward. The findings are of great interest, supporting the role of single-dose BNT162b2 vaccination to reduce SARS-CoV-2 transmission. The findings are also applicable to non-healthcare settings, may guide healthcare policies.

https://doi.org/10.7554/eLife.68808.sa1

Author response

[Editors' note: we include below the reviews that the authors received from another journal, along with the authors’ responses.]

Reviewer #1:

This short communication describes an opportunistic analysis of results of regular screening PCR tests for SARS CoV2 among healthcare workers some of whom had received the Pfizer vaccine and shows a reduction in positive results most marked from 12 days post dose. The allocation of vaccine was (presumably) non-random so there is some potential for bias which is not acknowledged (although this might have favoured frontline workers first who might be expected to be more exposed).

We thank the reviewer for highlighting this important potential source of confounding in our observations. We had not previously discussed confounding due to the stated word limits for ‘Correspondence’ articles. We have now included a description of the potential limitations of the study in a brief paragraph on limitations:

“HCWs working in COVID-19 clinical areas were prioritised for vaccination first, and a small number of clinically vulnerable HCWs were also given priority. Otherwise, vaccine allocation was uniformly at random. As asymptomatic infection was examined, the date of infection could have been earlier than the test date. These would all likely lead to an underestimate of the vaccine’s effect (bias towards the null). This study could only examine the short-term impact of single-dose BNT162b2 vaccination.”

It would also be helpful to know what the uptake rate of regular screening tests for HCWs at this hospital was at the time of the study and the range of time between vaccination and testing among those included in the analysis.

We thank the reviewer for this useful suggestion. Uptake at the time of study was ~50% (4,408 – 4,411 weekly tests from a population of ~9,000 HCW who worked at least one shift in each week [lines 45-46]). We have now included median and interquartile range values for the number of days since vaccination in each of the post-vaccine groups (lines 48-50).

Reviewer #2:

This is an interesting and important analysis. However, I don't think the evidence presented is sufficient to claim that, "…we provide real world evidence for a high level of protection against symptomatic SARS-CoV-2 infection after a single dose of BNT162b2 vaccine.". This might be overstated. There is no data to support an argument that the findings last over time and without a second dose of the vaccine at 3/52.

We thank the reviewer for highlighting this important study limitation. We have now included a brief paragraph on study limitations, which emphasises that “this study could only examine the short-term impact of single-dose BNT162b2 vaccination”. We have also re-worded the first sentence of the concluding paragraph, so as not to overstate the study conclusions: “Our findings provide real-world observational evidence of high level short-term protection against asymptomatic SARS-CoV-2 infection after a single dose of BNT162b2 vaccine, at a time of predominant transmission when the UK COVID-19 variant of concern 202012/01 (lineage B.1.1.7) accounted for >90% of local infections, and amongst a population with a relatively low frequency of prior infection (7.2% antibody positive).”.

Also, there is no evidence presented that I can see to suggest that the single dose vaccine is effective against the B.1.1.7 strain. Perhaps the authors could comment on the % of SARS-CoV-2 infections that were B.1.1.7?

We thank the reviewer for this useful suggestion. We have now included a sentence on the rate of B.1.1.7 lineage detection in samples from local community at the time of study “at a time when the UK COVID-19 variant of concern 202012/01 (lineage B.1.1.7) accounted for >90% of local infections”. This is based on publicly available COG-UK sequencing data, which found that from 228 sequenced samples from Cambridgeshire from between 18/01/21 and 31/01/21, 207 (90.8%) were identified as lineage B.1.1.7. https://β.microreact.org/project/kqW9xZtPytsXUpbyhVFCZM-cog-uk-2021-02-25-uk-sars-cov-2/

Reviewer #3:

Thank you for the opportunity to review this timely manuscript. As the COVID vaccines start to make their way out of the highest-risk populations and into more generally at-risk groups information regarding immediate efficacy and arresting of spread is essential for shaping vaccine rollout policies. I'd like to offer some comments on these preliminary findings:

1a. The authors' presentation of the three different groups (unvaccinated, those with fewer than 12 days post-vaccine and those with more than 12 days post vaccine) is accompanied by p-values which are testing the rate of positive PCR tests over all by group and then under various subsets. I do not see the statistical validity of these tests. This is not a random sample or a randomized trial. Is there evidence to suggest that the HCW were randomly selected or received vaccines that were not purposive in some way? Applying p-values to this data suggests that there is some potential for replicability where I do not think such a condition actually exists based on this design. If there is some evidence to suggest that the vaccination of HCW was random then it should appear in this document. The absence of randomization of any kind opens the door to argument about potential (and unexplored) confounding.

Many thanks for these comments. We have discussed at length our approach and data with two leading statisticians in Cambridge, who as a result have now been included as full authors. Shaun Seaman works at the MRC Biostatistics Unit, and his particular interests are in methodologies for drawing causal inferences from observation data and for handling selection biases. Professor Richard Samworth is the University of Cambridge Professor of Statistical Science and the Director of the University Statistical Laboratory. Both agree with our approach, and have been instrumental in writing this rebuttal. In response to the points raised:

a. P-values and confounding. P-values are frequently used in observational studies such as ours. It is standard practice when analysing observational studies to present p-values for differences between non-randomised groups and then to discuss potential confounding and how such confounding might be expected to bias the results. A p-value may be less than 0.05 because `treatment' or `exposure' (in this case, vaccination) has a causal effect on the outcome (in this case, testing positive), because of random chance (which is unlikely when p<0.05), or because of confounding bias. We had not yet discussed confounding due to the stated word limits for ‘Correspondence’ articles, however have now included a paragraph explicitly discussing limitations of our study. HCWs working in COVID-19 clinical areas were prioritised for vaccination first, and a small number of clinically vulnerable HCWs working elsewhere in the hospital were also given priority. Apart from that, vaccine allocation was uniformly at random. This would be likely to lead to an underestimate of the vaccine’s effect on asymptomatic infection, and we now directly reference this potential bias in lines 66-71 of our paragraph on study limitations.

In addition, the dates of infection would have been earlier than the dates of test, meaning that some positive tests in individuals who had been vaccinated >12 days may have been due to infections that occurred <12 days after vaccination. This would cause a bias towards underestimating the benefit of vaccination.

b. Randomisation and potential for replicability. It is not being claimed that the data originate from a random sample or randomised controlled trial. However, for our study, the data could be considered to be random in at least three different ways. First, Cambridge University Hospitals (CUH) could be thought of as randomly chosen hospitals from a population of similar hospitals. Had we randomly chosen different hospitals, we would have observed different numbers of positive tests in the three vaccination groups. Second, we could think of the CUH staff as being a random sample from the population of possible staff that CUH might have employed. Again, had the staff been different, so would have been the numbers of positive tests in each of the groups. Third, we can imagine hypothetically re-running January 2021. Again, the numbers of positive tests in each of the groups would then be different. This conceptualisation of the sample as coming from a hypothetical infinite population is standard in statistics. See, for example, Professor David Spiegelhalter's book `The art of statistics' pages 91-93, or Professor George Barnard's discussion (JRSSB 1971 vol 33 page 377):

"Another way to bring the standard theory to bear is to view the population from which we draw our samples as itself a sample from a super-population. This model of the survey process – which can be a perfectly valid model – perhaps does not receive the attention it deserves in the paper we have under discussion. My limited experience of the use of surveys in the social sciences suggests, indeed, that the super-population model is usually the appropriate one. One is rarely, for example, concerned with the finite de facto population of the U.K. at a given instant of time; one is more concerned with a conceptual population of people like those at present living in the U.K. If one adopts the super-population model, then since a simple random sample of a simple random sample is itself a simple random sample, the problems of inference can be dealt with along classical lines."

1b. Adding to that, it looks like the authors looked at all of the PCR screenings made over a two week period. This means that they looked at the whole population of screenings done during this time period and not a random sample of those screenings. Any differences observed between these groups are different but cannot be considered "statistically different" because there is no real potential for variability inherent in THIS design. These are not a random sample of PCR screenings -- they are all the screenings that happened in the time frame. Saying it's a random sample in time is also false because the authors did not actually randomly select these weeks -- they purposively chose them to see what was going on between the groups.

Please see our comments above about randomisation. We performed an observational study and are not claiming that the data came from a random sample. We chose the core two weeks of study (18/1/20 – 31/1/20) for the following reasons:

i. Prior to these two weeks, only a very small proportion of tests were from the ≥12 days since vaccination group because mass vaccination began on 8th January 2021. Exclusive examination of data earlier to 18/1/20 would have limited our ability to assess the effect of vaccination when an effect may be expected according to the phase III clinical trial.

ii. After these two weeks, a relatively small proportion of tests were from the unvaccinated group, because most people had been vaccinated by then.

iii. The prevalence of infection over the two weeks of study was stable in the unvaccinated group, facilitating simple comparison.

In the text, we have already stated “Here, we analyse data from the two weeks spanning 18th to 31st January 2021, during which: (a) the prevalence of COVID-19 amongst HCWs remained approximately constant; and (b) we screened comparable numbers of vaccinated and unvaccinated HCWs.” Furthermore, as we demonstrate, our analysis is robust to the choice of weeks. As discussed above (‘Randomisation and potential for replicability’), it is possible to regard the data as a random sample from a larger population. One limitation of our study includes the sampling from a single location (Cambridge), which could restrict the generalisability of our results. Nonetheless, we think it is very unlikely that vaccine efficacy will differ substantively between hospitals.

We have now included full week-by-week data for the extended six-week period of study (28/12/2020 – 07/02/2021) as an appendix to the manuscript to enable a full reanalysis of our data by any interested parties, and to highlight why the 2-week period used for primary analysis was chosen. As discussed in response to reviewer #3 point 5, we have also removed this analysis from the main manuscript.

In terms of the reviewer’s comment about examining the whole population of screenings as opposed to a random selection of them, it was necessary to examine the full sample as the proportion of positive tests in the vaccinated group in particular was very small (4/1989 tests). For any reliable estimate of a small proportion, a large sample size is required.

1c. What I would recommend instead of presenting these results as if they were inferential is to put 95% Clopper-Pearson intervals (or some other exact method as the authors choose) and describe these populations rather than comparing them as if they were randomly assigned and present these results as observational and descriptive rather than making them seem experimental and inferential.

In reviewer #3 point (3), the reviewer suggests that there was "no real potential for variability in this design". If this were true, then there would also be no justification for presenting confidence intervals. Both p-values and confidence intervals are statements about what would happen over repeated samples of the data. If confidence intervals are meaningful, then so are p-values. A 95% confidence interval is a random interval that would include the true proportion in (at least) 95% of repeated samples. A p-value is the proportion of the repeated samples where the difference between the proportions in the two (or more) groups is as large as the difference that is seen in the observed data. For our study, repeated sampling could be interpreted in at least three different ways (see ‘Randomisation and potential for replicability’ in our response to reviewer #3 point 1). Whilst we agree with the reviewer’s request to include a discussion of the limitations of the study, we disagree that p-values are inappropriate, and we fail to understand why confidence intervals can be appropriate if p-values are not.

To address the reviewer’s point, we have re-analysed all of the data providing 95% confidence intervals, using Wilson’s method (please see results below). This illustrates that both p-values and intervals calculated using an exact method lead to similar conclusions.

26/3,252 (0·80%, Wilson’s interval 0.55 – 1.17%) tests from unvaccinated HCWs were positive (Ct<36), compared to 13/3,535 (0·37%, Wilson’s interval 0.22 – 0.63%) from HCWs <12 days post-vaccination and 4/1,989 (0·20%, Wilson’s interval 0.08 – 0.51%) tests from HCWs ≥12 days post-vaccination (p=0·023 and p=0·004, respectively; Fisher’s exact test, Figure).

A marked reduction in infections was also seen when analyses were repeated with: (a) inclusion of HCWs testing positive through both the symptomatic and asymptomatic arms of the programme (56/3,370 (1·71%, Wilson’s interval 1.28-2.15%) unvaccinated vs 8/2,018 (0·40%, Wilson’s interval 0.20 – 0.79%) ≥12 days post-vaccination, 4·3-fold reduction, p=0·00001); (b) inclusion of PCR tests which were positive at the limit of detection (Ct>36, 42/3,268 (1·29%, Wilson’s interval 0.96 – 1.74%) vs 15/2,000 (0·75%, Wilson’s interval 0.46-1.23%), 1·7-fold reduction, p=0·075); and (c) extension of the period of analysis to include six weeks from December 28th to February 7th 2021 (113/14,083 (0·80%, Wilson’s interval 0.67 – 0.96%) vs 5/4,872 (0·10%, Wilson’s interval 0.04 – 0.24%), 7·8-fold reduction, p=1x10-9).

2. The authors (p2) appear to be presenting analyses where they noticed the largest differences between unvaccinated and 12+ days post-vaccination accompanied by p-values to suggest that these differences possess scientific significance. It is unknown how many different tests the authors carried out and it is a well-known statistical fact that the number of erroneous but significant results is directly related to the number of tests carried out on a dataset. The authors are cherry-picking significant findings and presenting them as if this were a scientific study driven by hypotheses that could actually be tested. It looks, to me, like they used all the data they could and found all the significant differences they could find and present them here. There's no discussion of practical significance -- the fact that less than 1% of health care workers who aren't vaccinated are potentially asymptomatic COVID positive. There's no discussion of the false positive or false negative rate of the PCR test that could be impacting all three groups and there's no justification such a large sample size was explicitly necessary for this investigation. The Fisher's Exact test is easily fooled into producing a significant result simply by inflating the sample size. The authors excessive sample size could be enough to create significance when in fact it may not really be there at all. To me, for example, the fact that almost the same number of people tested positive after being vaccinated as did without the vaccine is a bigger problem than talking about whether they were testing positive relative to when they had the vaccine. Chasing after significance has led to missing this important fact about the vaccinated HCW that are still potentially spreading COVID even though they may think they're safe.

Many thanks for offering the opportunity to clarify our data further. In terms of the points made:

1. Choice of weeks. As described in the response to reviewer #3 point 2, the study period analysed was rationally chosen in order to include comparable numbers of vaccinated and unvaccinated HCWs, over a period in which the prevalence of COVID-19 amongst HCWs remained approximately constant. Furthermore, our analysis was robust to examination of an extended six week period chosen using similar principles. To further address this point and provide full data for analysis by others, full week-by-week data is now presented in the appendix (see also response to reviewer #3 point 3).

2. Practical significance. The practical significance of the results is demonstrated by the magnitude of the risk ratios, which are large. The proportion of asymptomatic screening tests is expected to be lower than the proportion of symptomatic tests. Even if fewer than 1% of unvaccinated HCWs tested positive over the course of the period of study, a 75% reduction is clearly extremely valuable in preventing nosocomial transmission to vulnerable patients and other staff i.e. from the point of view of infection control within the hospital, the difference between 0.2% and 0.8% is highly clinically significant. We have added a sentence “Nevertheless, the 75% reduction in asymptomatic infection offered by vaccination is likely to be crucial to control nosocomial transmission of SARS-CoV-2.” to the Discussion section to highlight this.

3. False positive and false negative rate. Errors in the test represent a form of measurement error, which, unless the sensitivity and specificity are different in vaccinated and unvaccinated individuals, is likely to cause the estimated relative risk to be slightly biased towards the null hypothesis (i.e. the size of the effect will be underestimated).

4. Sample size. We have discussed the necessity for testing the full, large sample size in our response to reviewer #3 point 2. This is required in order to reliably estimate a small proportion of positive tests. When a large sample is available, we fail to see the justification for discarding part of this sample just to reduce the power of a statistical test to detect small effects. We agree that when the sample size is large, a null hypotheses of no effect can be (correctly) rejected when the effect size is non-zero but small. That is one reason why it is important to present the estimate of the effect size, as we have done here.

5. Reviewer’s suggestion that ‘almost the same number of people tested positive after being vaccinated as did without the vaccine’. The basis for this statement is unclear to us. As stated, 26/3,252 (0·8%) tests from unvaccinated HCWs were positive, compared to 13/3,535 (0·4%) from HCWs <12 days post-vaccination and 4/1,989 (0·2%) tests from HCWs ≥12 days post-vaccination. As discussed above, an effect on symptomatic infections in the original phase III clinical trial was seen from 12 days onwards. The most important comparison is therefore between the proportions of positive tests in the unvaccinated and ≥12 days post vaccination groups. Furthermore, the most important comparison is between proportions of positive tests, not total numbers.

6. ‘Vaccinated HCW are still potentially spreading COVID’. We agree that the vaccine is not a panacea and have already suggested that vaccinated HCWs are substantially less likely to become infected spread SARS-CoV-2, as opposed to being completely protected. To emphasise this point, we have added the sentence “This protection was by no means complete, suggesting that continued asymptomatic HCW screening, social distancing, mask-wearing and strict hand hygiene remain vital.” to the end of the discussion.

3. The authors present additional analyses where they extend the time period to include more unvaccinated health care workers in an effort to further inflate the apparent difference between vaccinated and unvaccinated HCW. If there was no demonstration of efficacy of the vaccine then there might be a chance for the null to be true in a situation like this. As it was, we expect the vaccinated group to have less incidence of positive testing, so this contrivance is simply a manufacturing of evidence without any real grounding as to why it's an important "finding." Adding in all the rest of the data since the beginning of the vaccination rollout reinforces my earlier point that this truly is now the whole population and should not be accompanied by a p-value for any reason. These findings are just not statistically significant -- they are what they are. If there's any real practical difference that should be discussed here. In addition, the numbers used in the extended time period do not make sense. They do not indicate which groups are being compared (113/14k what? 5/4.8k what?). This whole thing just doesn't make sense to me and looks like it's just another way to produce a significant-looking p-value without having a real experimental design first.

Our intention in analysing the extended six-week period was merely to determine whether our conclusions were robust across a longer period of time. The weeks included in this analysis met at least one of the two major criteria for selection (i.e. constant prevalence of infection in the unvaccinated group, and/or sufficient tests in the ≥12-weeks post vaccination group to assess the proportion of positive tests).

Nevertheless because this has caused confusion, we have removed the six-week analysis, and have given further details in the form of a week-by-week data over this period in an appendix.

Furthermore, vaccine efficacy against asymptomatic infection was absolutely not pre-supposed. Indeed, to try to answer that key question is exactly why we performed the study! To clarify this point, we have changed the statement “A vital outstanding question is whether this vaccine prevents asymptomatic as well as symptomatic SARS-CoV-2 infection, or converts infections from symptomatic to asymptomatic.”.

Please see our other responses above, which already address other points in this section.

4. I do not agree with the authors' conclusions regarding the value of their "evidence." There is nothing in here to suggest that the vaccination has actually reduced the number of asymptomatic carriers in this population. The rate of positive tests in those vaccinated does not appear to be practically different in the vaccinated and unvaccinated and the only way the authors could make this look like evidence was to manufacture significant p-values through purposive comparisons and sample size inflation. The authors do not rely on sound scientific and statistical methodology to demonstrate their results and present their observational findings as if they were the result of a designed experiment.

This is extremely problematic. The pandemic has been raging for more than a year and vaccination is our best chance to controlling the spread and potentiating a return to social normality. There are a tremendous number of problems with vaccination; barriers to vaccination are rooted in great mistrust of the health care system and anti-vaxxers are compelling in their arguments because they can look at manufactured research like this and easily demonstrate its flaws and fallibility. They can easily make the point that if the vaccine really worked then why would they have to make things like this up to show it? They can then suggest that it's evidence of a hidden agenda and we've lost the fight. Fabrication of evidence and manufacturing significant findings to create the illusion of science is not going to help move the vaccine into the population desperately hoping for a return to normal but skeptical about how that will happen when they can't trust the people who should be trying to help them. This research which was supposed to show that the vaccine is working does so in a way that looks like it's really prestidigitation made to fool people into doing something they should want to do but don't.

Please see our responses to similar points above. For the avoidance of doubt, our study design was as follows:

a. Record all screening test results during the relevant period;

b. Measure the proportions of positives in the different groups;

c. Compare the proportions of positives between the groups.

It was therefore an extremely simple, transparent study, which would be inherently difficult to manipulate. We reject in the strongest possible terms the reviewer’s suggestions (both implicit and explicit) that we have fabricated evidence, manufactured significant findings or engaged in any sort of prestidigitation. These are categorically untrue.

Response to second round for reviews:

Reviewer #1:

In Jones et al., the authors are investigating whether single dose BNT162b2 mRNA COVID-19 vaccine protects from asymptomatic SARS-CoV-2 infection in a HCW cohort. They claim that there is a 4-fold reduction in risk of asymptomatic SARS-CoV-2 infection in vaccinated HCW at >12d post single dose vaccination compared to unvaccinated HCW. The study was carried out in a HCW cohort with documented prior SARS-CoV-2 infection at 7.2% (by positive serology). The B.1.1.7 variant may have accounted for about 90% of infections.

This is an extremely important question.

Many thanks for these comments, and for noting that our study addresses an extremely important question.

Major concerns

1. Prior infection history. A recent Letter published in the Lancet [Charlotte Manisty, et al. Antibody response to first BNT162b2 dose in previously SARS-CoV-2-infected individuals. Lancet. 2021 Feb 25] showed that HCW with a prior history of mild or asymptomatic SARS-CoV-2 infection made a substantially enhanced response following single dose vaccination with BNT162b2 such that the single dose prime was effectively acting as a boost. In this study, the authors report a 7.2% frequency of prior infection in the HCW cohort being studied. However, the authors do not comment on the frequency of prior natural infection across the three groups studied (unvaccinated, <12d post vaccination & >12d post vaccination). This is important because this could be an important potential confounding factor in the analysis of the study. Prior immunity from natural infection or boosted immunity following single dose vaccination may protect from SARS-CoV-2 infection thus impacting on frequency of PCR positive tests across the three groups. Prior infection history (as documented by serology) needs to be taken into account in the analysis. This data is almost certainly available for the HCW cohort being reported here.

Many thanks for this interesting suggestion. Our manuscript was originally submitted on Feb 21, prior to the publication by Mainsty et al. Nevertheless, we have now compared the frequency of natural infection across the three groups studied. The frequency of prior SARS-CoV-2 infection (as documented by a positive serological test for SARS-CoV-2 prior to 28/12/2020) was similar across all groups (7.1%, unvaccinated; 5.6%, <12 days post-vaccination; 5.7%, ≥12 days post-vaccination) suggesting that this did not confound our observations. We have added this data to our manuscript (lines 75-78).

2. B.1.1.7 variant. The authors should show the frequency of B.1.1.7 variant / WT SARS-CoV-2 across each of the three groups studied (as documented by actual study PCR data during study follow up and not a local average). This data should be available for the HCW cohort being reported here. It is picked up by the PCR test as the ∆69/70 deletion results in failure to detect the S gene target while the ORF1ab and N are not affected. This is how PHE was first alerted to the emergence and spread of the B.1.1.7 variant in the UK. There is no biological evidence presented in this study to show that single dose BNT162b2 protects against the B.1.1.7 variant.

Many thanks for this interesting suggestion. We are certainly aware of the potential for some PCR assays to estimate the proportion of ∆69/70 viruses by S gene drop-out. In fact, only a fraction of the Lighthouse labs and wider PHE regional laboratory network routinely use triple target PCR testing – this reflects the different platforms used between laboratories in the network, an intentional design feature which provides resilience in the UK’s testing capacity. As such, the S gene is not targeted by the platform used in the Cambridge COVID-19 Testing Centre (Lighthouse lab). PCR to identify the B.1.1.7 variant was therefore not routinely available during the study period.

We considered it highly likely that infections amongst HCWs in Cambridge would be representative of viruses circulating in the Cambridge community (and indeed, UK) during the study period. Nevertheless, we have now been able to confirm this by examining Nanopore sequencing data for isolates from asymptomatic HCWs. Over the 2-week period used for the primary analysis, 29/43 viral isolates have been successfully sequenced, of which 24/29 (83%) were from the B.1.1.7 lineage. Across the three groups, 15/19 (79%) sequences from positive unvaccinated HCWs were B.1.1.7, compared to 6/7 (86%) from HCWs <12 days post-vaccination and 3/3 (100%) from HCWs ≥12 days post-vaccination (no significant difference between groups).

We have added data on the overall percentage of sequences that were the B.1.1.7 variant to the manuscript (lines 82-83). We have also added two main authors (Dr. Ben Warne and Professor Ian Goodfellow) who played major roles in the sequencing effort, and additional authors from COG-UK to the group authorship.

3. Lack of follow up beyond 18d. The study can only comment on short term impact of first dose vaccination as the latest time point studied is d18. This is a major deficiency of the study as the second dose of BNT162b2 has been delayed by 12 weeks in the UK and the results described here do not address any later time points. There will have been a proportion of HCW that did not have a COVID-19 vaccine (about 35% of HCW in Leicester for example) and this group could be included in the analysis (for comparison at later time points).

We thank the reviewer for highlighting this point. We previously addressed this same point in response to reviewer #2 point #1 (in their original review). We included a paragraph on study limitations, specifically emphasising that “this study could only examine the short-term impact of single-dose BNT162b2 vaccination”, and re-worded the first sentence of the concluding paragraph: “Our findings provide real-world observational evidence of high level short-term protection against asymptomatic SARS-CoV-2 infection after a single dose of BNT162b2 vaccine…”

Unlike in Leicester, >85% of HCWs participating in our screening programme have now been vaccinated. Combined with the rapid decline in incidence of SARS-CoV-2 infection in the local community, in the last week of data we present (starting 01/02/2021), there were only 3/4282 positive tests overall, offering little scope for analysis of later time points amongst our cohort.

Reviewer #3:

Thank you for the opportunity to take a look at the revised manuscript. I have read through the authors' responses to both the others' and my reviews and I have some additional comments:

1. Thank you for taking a look at what I had to say about your article and addressing many of the statistical issues I commented on in my initial review. The authors are to be commended for seeking out professional advice and the addition of two professional statisticians to the manuscript says a lot about the authors' commitment to scientific rigor.

We are grateful for this comment, and agree with this point.

2. I would like to thank the authors for the comment regarding p-values and confidence intervals. The authors are, of course, correct that p-values and confidence intervals "come from the same place" whereby changing from one incorrect form of inference to another is not really making a real change. However, the confidence interval provides considerably more information to the reader where the p-value provides no information at all. A 95% confidence interval gives the reader a rough idea of how large or small an effect one may expect to see in repeating a study. A p-value may tell you the chance of seeing a result as extreme in magnitude of occurring in replication but evidence based practice should be more about bang for one's buck. Including the confidence interval, while still inappropriate for inferential purposes in this study, provide greater description of the data and that is an invaluable addition to this paper. Thank you for choosing to make the article more valuable by adding in this very useful information.

We are grateful for this comment, and agree with this point.

3. Following from my previous comment, I see that the authors chose to use Wilson's approach for producing their intervals. My guess is they chose this approach because the Clopper-Pearson approach (an exact approach) gives a wider interval and my guess is that led to "loss of significance." Considering their decision to use other exact tests makes me wonder why they went with an approximation here.

In their prior review, reviewer #3 requested “What I would recommend instead of presenting these results as if they were inferential is to put 95% Clopper-Pearson intervals (or some other exact method as the authors choose)…”

We chose to use Wilson’s exact method at the suggestion of our professional statistical colleagues. Shaun Seaman works at the MRC Biostatistics Unit, and his particular interests are in methodologies for drawing causal inferences from observation data and for handling selection biases. Professor Richard Samworth is the University of Cambridge Professor of Statistical Science and the Director of the University Statistical Laboratory. Both are now included as full authors. They referenced the very highly cited manuscript by Brown, Cai and Dasgupta 2001 “‘Interval estimation for a binomial proportion”. Here, the authors suggest: “The Clopper-Pearson interval is wastefully conservative and is not a good choice for practical use… better exact methods are available…” “For larger n, the Wilson, the Jeffreys and the Agresti–Coull intervals are all comparable”.

We have now calculated Clopper-Pearson intervals, and the results are very similar. For example, for our primary outcome 26/3,252 (0·80%) tests from unvaccinated HCWs were positive (Wilson interval 0.55 – 1.17%; Clopper-Pearson interval 0.52 – 1.17%), compared to 4/1,989 (0·20%) tests from HCWs ≥12 days post-vaccination (Wilson interval 0.08 – 0.51%; Clopper-Pearson interval 0.05 – 0.51%).

4. The authors make a point about "estimating a small proportion" requiring a large sample size. This really is not the case at all. Any sample size is sufficient for estimating a proportion of any size. Yes, some samples may give an estimate that will never be exact, but this does not mean that the estimator itself is problematic. Consider a population with 100 individuals and one of them is sick so the true P = 0.01. If we select a sample of size n=5 we will never get a good estimate of the truth from our sample. There are a total of (100C5) possible samples or 75287520. Of those, 71523144 result in an estimate of p=0 and the remaining 3764376 produce an estimate of p=0.2. Clearly, both resulting estimates are wrong but, on average the estimate comes out to be 1% which is the true population proportion. What the authors really mean is that if they want a really PRECISE estimate of their small proportion then they need a really large sample size. An unbiased estimator doesn't get more unbiased as the sample size increases; it just gets more precise, which is what the authors need if they want to show that their really small differences are statistically significant.

We agree with the reviewer that to estimate precisely the small proportions of positive cases in each group, a large sample size was required. This is why we included all screening test results in our analysis, rather than a random sample. We agree that we should have used the term precise rather than reliable in our original response, and thank the reviewer for pointing this out.

5. The authors' response regarding clinical significance is greatly appreciated. I am a statistician and not a clinician, so when I see two percentages that are considerably small it is hard for me to know that they are actually important. I appreciate the clarification as it speaks to the relevance of the data from a practical perspective, something I noted was missing in the original draft. The authors do use percent change to create an inflated perceived improvement which I think is perhaps an opportunity to sensationalize a finding that should already be important if a change from 0.8% to 0.2% is already highly significant from a clinical point of view. I just do not understand why, if these results are so clinically important, why the authors need to paint this vivid picture of statistical significance. It makes me wonder, and a little concerned that this was ok with the statisticians that they added to the paper. The use of ratios and percent change are contrivances of modern epidemiology to create effect when they may simply not be there otherwise. If the practical significance is there be honest about it.

Ultimately, the authors' work is very important for many descriptive reasons. My lack of clinical knowledge gives me the inclination to take the authors at their word regarding the practical importance of their findings. However, my considerable statistical knowledge and experience has me wondering.… if these results are so important, why do the authors have to abuse statistical methods to sell them? If they're really this big of a deal, shouldn't they be able to sell themselves on their own merit?

It is certainly not our intention to inflate the perceived improvement, and we agree that absolute differences can be important. We have made precise estimates of the small proportions of positive cases in each group, as acknowledged by the reviewer in point #4, which facilitates assessment of the change in %infection between groups. To address this point, we have changed “Nevertheless a 75% reduction in asymptomatic infection…' to 'Nevertheless a four-fold reduction from 0.8% to 0.2% in asymptomatic infection…”.

By analogy, if the prevalence of COVID-19 infection in two regions of the UK were (for instance) 0.8% vs 0.2%, the reviewer would presumably agree that such a difference would be exceptionally important.

To be completely transparent, we have presented both proportions (as %) and absolute numbers in every instance in the text, together with all the raw data in the appendix. We believe that both provide valuable information and context, and have made great efforts to report the significance of our findings accurately instead of sensationalising them. We therefore do not feel it is at all fair to say that we have 'abused statistical methods'. We hope that on reflection the referee will accept this and retract this unjustified accusation.

If the authors are willing to report their findings descriptively, using 95% Wilson intervals to show effect (as they agree this is important in their response) and eschew p-values then I will re-consider my decision. If the authors present their findings cleanly, without resorting to ratios and percent differences which sensationalize the findings and inflate their perceived importance then I will be happy to accept their work for publication. Finally, I think it's important for all of the authors to understand that statistical methods are about quantifying uncertainty and not about coming up with the best estimate of the truth. The beauty of an observational study is in its opportunity to provide us with an idea of how things are going; from that perspective ANY sample size can provide sufficient information to inform future research. It's only when we're blinded by the need for statistical significance that we start looking for ways to "find results" rather than answer real questions.

As in our original rebuttal, we believe that p-values also have merit alongside confidence intervals, whilst accepting the reviewer's point that additional information is provided by confidence intervals. We have therefore now included both, using identical wording to the statement we provided in our original rebuttal. We note that the reviewer accepted this statement as “an invaluable addition to this paper” (reviewer #3 point 2 above; lines 53, 54, 55, 62, 63, 65).

For the reasons stated above, we believe it is clearly helpful to include ratios (fold reductions) to estimate the size of the effect. This is entirely consistent with the approach taken by other studies in the medical literature. Again, we absolutely reject the reviewer’s suggestion that we have attempted to “find results”. This is simply untrue.

If the editors wish us to remove these ratios, we would be willing to do so, however we believe this would make life unnecessarily difficult for the reader.

I do hope that my comments are helpful and give the authors an opportunity to decide how to best convey the story their data are telling.

https://doi.org/10.7554/eLife.68808.sa2

Article and author information

Author details

  1. Nick K Jones

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Clinical Microbiology & Public Health Laboratory, Public Health England, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Project administration
    Contributed equally with
    Lucy Rivett
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4475-7761
  2. Lucy Rivett

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Clinical Microbiology & Public Health Laboratory, Public Health England, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Project administration
    Contributed equally with
    Nick K Jones
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2781-9345
  3. Shaun Seaman

    Medical Research Council Biostatistics Unit, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Richard J Samworth

    Statistical Laboratory, Centre for Mathematical Sciences, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Ben Warne

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Cambridge Institute of Therapeutic Immunology & Infectious Disease, University of Cambridge, Cambridge, United Kingdom
    3. Department of Medicine, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Chris Workman

    Occupational Health and Wellbeing, Cambridge Biomedical Campus, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Mark Ferris

    Occupational Health and Wellbeing, Cambridge Biomedical Campus, Cambridge, United Kingdom
    Contribution
    Investigation, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  8. Jo Wright

    Occupational Health and Wellbeing, Cambridge Biomedical Campus, Cambridge, United Kingdom
    Contribution
    Data curation, Investigation, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  9. Natalie Quinnell

    Occupational Health and Wellbeing, Cambridge Biomedical Campus, Cambridge, United Kingdom
    Contribution
    Data curation, Investigation, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  10. Ashley Shaw

    Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    Contribution
    Supervision, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  11. Cambridge COVID-19 Collaboration

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Department of Medicine, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Resources, Data curation
    Competing interests
    No competing interests declared
    1. Amy Amory
    2. Stephen Baker
    3. Emma Bateman
    4. Aklima Begum
    5. Moushima Begum
    6. John Bradley
    7. Michael Brennan
    8. Helen Burn
    9. Caroline Crofts
    10. Afzal Chaudhry
    11. Yasmin Chaudhry
    12. Daniel J Cooper
    13. Sharon Dawson
    14. Gordon Dougan
    15. Renny Feather
    16. Louise Free
    17. Katie Friel
    18. Claire Gildea
    19. Iliana Georgana
    20. Lizz Grimwade
    21. Ravi Gupta
    22. Susan Hall
    23. Sophie Hannan
    24. James Hayes
    25. Aleksandra Hosaja
    26. Myra Hosmillo
    27. Rhys Izuagbe
    28. Aminu Jahun
    29. Lidia James
    30. Jill Jardin
    31. Nathalie Kingston
    32. Sara Lear
    33. Paul A Lyons
    34. Patrick H Maxwell
    35. Sue Mott
    36. Sarah Mugavin
    37. Joyce Mwiya
    38. Sharon Peacock
    39. Ravi Prakash Nallattil
    40. Kazeem Oloyede
    41. Willem H Ouwehand
    42. Elle Page
    43. Marina Perez
    44. Tim Raine
    45. Matthew Routledge
    46. Caroline Saunders
    47. Kenneth GC Smith
    48. Dominic Sparkes
    49. Maria Stafford
    50. Charlotte Summers
    51. Despiona Tatsi
    52. James ED Thaventhiran
    53. Sharon Thomas Johnson
    54. M Estée Török
    55. Mark Toshner
    56. Lesley Turner
    57. Kate Wall
    58. Karis Watson
  12. Ian G Goodfellow

    Division of Virology, Department of Pathology, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9483-510X
  13. Paul J Lehner

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Cambridge Institute of Therapeutic Immunology & Infectious Disease, University of Cambridge, Cambridge, United Kingdom
    3. Department of Medicine, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Conceptualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9383-1054
  14. Rob Howes

    Cambridge COVID-19 Testing Centre and AstraZeneca, Anne McLaren Building, Cambridge, United Kingdom
    Contribution
    Data curation, Investigation, Methodology, Project administration, Writing - review and editing
    Competing interests
    Dr Howes was employed by AstraZeneca PLC during the period of study and preparation of this manuscript.
  15. Giles Wright

    Occupational Health and Wellbeing, Cambridge Biomedical Campus, Cambridge, United Kingdom
    Contribution
    Supervision, Investigation, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  16. Nicholas J Matheson

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Cambridge Institute of Therapeutic Immunology & Infectious Disease, University of Cambridge, Cambridge, United Kingdom
    3. Department of Medicine, University of Cambridge, Cambridge, United Kingdom
    4. NHS Blood and Transplant, Cambridge, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing - review and editing
    Contributed equally with
    Michael P Weekes
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3318-1851
  17. Michael P Weekes

    1. Cambridge University NHS Hospitals Foundation Trust, Cambridge, United Kingdom
    2. Department of Medicine, University of Cambridge, Cambridge, United Kingdom
    3. Cambridge Institute for Medical Research, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Project administration
    Contributed equally with
    Nicholas J Matheson
    For correspondence
    mpw1001@cam.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3196-5545

Funding

Wellcome Trust (108070/Z/15/Z)

  • Michael P Weekes

Wellcome Trust (207498/Z/17/Z)

  • Ian G Goodfellow

Wellcome Trust (210688/Z/18/Z)

  • Paul J Lehner

Medical Research Council (MR/P008801/1)

  • Nicholas J Matheson

NHS Blood and Transplant (WPA15-02)

  • Nicholas J Matheson

EPSRC (EP/P031447/1)

  • Richard J Samworth

Medical Research Council (MC_UU_00002/10)

  • Shaun Seaman

EPSRC (EP/N031938/1)

  • Richard J Samworth

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was funded in part by Wellcome Senior Clinical Research Fellowships (Grant numbers 108070/Z/15/Z to MPW, 207498/Z/17/Z to IGG), a Wellcome Principal Research Fellowship to PJL (210688/Z/18/Z), an MRC Clinician Scientist Fellowship (MR/P008801/1) and NHSBT workpackage (WPA15-02) to NJM, EPSRC grants to RJS (EP/P031447/1,EP/N031938/1), and an MRC grant to SS (MC_UU_00002/10). The sequencing costs were funded by the COVID-19 Genomics UK (COG-UK) Consortium, which is supported by funding from the Medical Research Council (MRC) part of UK Research and Innovation (UKRI), the National Institute of Health Research (NIHR), and Genome Research Limited, operating as the Wellcome Sanger Institute. Funding was also received from Addenbrooke’s Charitable Trust and the NIHR Cambridge Biomedical Research Centre. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. We also acknowledge contributions from all staff at CUHNFT Occupational Health and Wellbeing, the COVID-19 vaccination programme and the Cambridge COVID-19 Testing Centre.

Ethics

Human subjects: This study was conducted as a service evaluation of the CUHNFT staff testing and vaccination services (CUHNFT clinical project ID ID3682). As a study of healthcare-associated infections, this investigation is exempt from requiring ethical approval under Section 251 of the NHS Act 2006 (see also the NHS Health Research Authority algorithm, available at http://www.hra-decisiontools.org.uk/research/, which concludes that no formal ethical approval is required).

Senior and Reviewing Editor

  1. Jos WM van der Meer, Radboud University Medical Centre, Netherlands

Publication history

  1. Received: March 26, 2021
  2. Accepted: April 7, 2021
  3. Accepted Manuscript published: April 8, 2021 (version 1)
  4. Version of Record published: April 23, 2021 (version 2)

Copyright

© 2021, Jones et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 4,642
    Page views
  • 537
    Downloads
  • 19
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Epidemiology and Global Health
    2. Medicine
    Lucy Rivett et al.
    Research Article Updated

    Significant differences exist in the availability of healthcare worker (HCW) SARS-CoV-2 testing between countries, and existing programmes focus on screening symptomatic rather than asymptomatic staff. Over a 3 week period (April 2020), 1032 asymptomatic HCWs were screened for SARS-CoV-2 in a large UK teaching hospital. Symptomatic staff and symptomatic household contacts were additionally tested. Real-time RT-PCR was used to detect viral RNA from a throat+nose self-swab. 3% of HCWs in the asymptomatic screening group tested positive for SARS-CoV-2. 17/30 (57%) were truly asymptomatic/pauci-symptomatic. 12/30 (40%) had experienced symptoms compatible with coronavirus disease 2019 (COVID-19)>7 days prior to testing, most self-isolating, returning well. Clusters of HCW infection were discovered on two independent wards. Viral genome sequencing showed that the majority of HCWs had the dominant lineage B∙1. Our data demonstrates the utility of comprehensive screening of HCWs with minimal or no symptoms. This approach will be critical for protecting patients and hospital staff.

    1. Epidemiology and Global Health
    2. Medicine
    3. Microbiology and Infectious Disease
    Edited by Diane M Harper et al.
    Collection

    eLife has published the following articles on SARS-CoV-2 and COVID-19.