Open Access Protocol

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
Atraumatic versus traumatic lumbar
puncture needles: a systematic review
and meta-analysis protocol
Siddharth Nath,1 Jetan H Badhiwala,2 Waleed Alhazzani,3 Farshad Nassiri,2
Emilie Belley-Cote,4 Alex Koziarz,1 Ashkan Shoamanesh,4 Laura Banfield,5
Wieslaw Oczkowski,4 Mike Sharma,4 Demetrios Sahlas,4 Kesava Reddy,1
Forough Farrokhyar,6 Sheila Singh,1 Sunjay Sharma,1 Nicole Zytaruk,3
Magdy Selim,7 Saleh A Almenawer1

To cite: Nath S, ABSTRACT

Badhiwala JH, Alhazzani W, Strengths and limitations of this study
Introduction: Lumbar puncture is one of the oldest
et al. Atraumatic versus
and most commonly performed procedures in ▪ This review will comprehensively examine the
traumatic lumbar puncture
needles: a systematic review
medicine, used to diagnose and treat disease. evidence on clinical outcomes of atraumatic
and meta-analysis protocol. Headache following lumbar puncture remains a lumbar puncture needles and compare them with
BMJ Open 2017;7:e014478. frequent complication, causing significant patient the traumatic needle type.
doi:10.1136/bmjopen-2016- discomfort and often requiring narcotic analgesia or ▪ An extensive literature search, developed in con-
014478 invasive therapy. Needle tip design has been proposed sultation with a librarian with experience in sys-
to affect the incidence of headache postlumbar tematic reviews, will be conducted to include
▸ Prepublication history for puncture, with pencil-point ‘atraumatic’ needles studies from multiple different databases,
this paper is available online. thought to reduce its incidence in comparison to without language or publication type restrictions.
To view these files please bevelled ‘traumatic’ needles. Despite this, the use of ▪ Heterogeneity is expected to exist in the primary
visit the journal online atraumatic needles and knowledge of their existence outcome; however, we plan to explore it by con-
(http://dx.doi.org/10.1136/ remains significantly limited among clinicians. This ducting several prespecified subgroup analyses.
bmjopen-2016-014478). study will systematically review the evidence on ▪ The current systematic review protocol does not
atraumatic lumbar puncture needles and compare them plan to evaluate economic analysis. We intend to
Received 27 September 2016 with traumatic needles across a variety of clinical
Revised 7 November 2016 examine cost in future studies but not in the
outcomes. current meta-analysis methodology.
Accepted 13 December 2016
Methods and analyses: We will include published
randomised controlled trials (RCTs), observational
studies and abstracts, with no publication type or
language restrictions. Search strategies will be Ethics and dissemination: Research ethics board
designed to peruse the MEDLINE, EMBASE, Web of approval is not required for this study as it draws from
Science, ClinicalTrials.gov, CINAHL, WHO Clinical Trials published data and raises no concerns related to
Database and Cochrane Library databases. We will also patient privacy. This review will provide a
implement strategies to search the grey literature. 3 comprehensive assessment of the evidence on
reviewers will thoroughly and independently examine atraumatic needles for lumbar puncture and is directed
the search results, complete data abstraction and to a wide audience. Results from the review will be
conduct quality assessment. Included RCTs will be disseminated extensively through conferences and
assessed using the Cochrane risk of bias assessment submitted to a peer-reviewed journal for publication.
tool and eligible observational studies will be examined Trial registration number: CRD42016047546.
using the Newcastle-Ottawa Scale. We will examine the
outcomes of: headache and its type, intensity, duration
and treatment; backache; success rate; hearing
disturbance and nerve root irritation. The primary
outcome will be the incidence of postdural puncture INTRODUCTION
headache. We will calculate pooled estimates, relative
Lumbar puncture is one of the most
risks for dichotomous outcomes and weighted mean
For numbered affiliations see differences for continuous outcomes, with
common invasive procedures in modern
end of article. corresponding 95% CIs. Statistical heterogeneity will medicine. It is used routinely by physicians
be measured using Cochran’s Q test and quantified not only to diagnose disease through sam-
Correspondence to using the I2 statistic. We will also conduct prespecified pling or imaging, but also to treat by deliver-
Dr Saleh A Almenawer; subgroup and sensitivity analyses to examine if ing chemotherapy or anaesthesia, or
dr_menawer@hotmail.com covariates exist and to explore potential heterogeneity. reducing intracranial pressure.1 2 It remains

Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478 1

Open Access

one of the most widely practiced procedures, being per- versions will be made available on PROSPERO, with cat-

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
formed by various specialists, such as neurologists, neuro- alogued version history.
surgeons, paediatricians, anaesthesiologists, emergency
physicians, radiologists and internists. While the safety Literature search
profile of lumbar puncture has improved drastically since A detailed search will be conducted of the MEDLINE,
its inception in the 1800s, it remains plagued by often EMBASE, Web of Science, ClinicalTrials.gov, CINAHL,
debilitating patient complications, the most prominent of WHO Clinical Trials Database and Cochrane Library
which is headache. Postlumbar puncture headaches are databases through 5 January 2017, without language or
classically described as postural in nature and frontal– publication type restrictions. We will use keywords and
occipital in location. These headaches can severely medical subject heading terms related to needle type
impact patient well-being, often requiring hospitalisation (atraumatic or traumatic) and clinical outcomes. Search
and potentially, invasive therapy.3 strategies will be developed in consultation with a librar-
Headache is postulated to occur as a result of leakage ian with expertise in systematic reviews.12 The search
of cerebrospinal fluid (CSF) from the site of the dural strategy employed for the MEDLINE database is pro-
tear created during the puncture. This leakage results in vided in table 1. The search will be supplemented by
a decrease in CSF pressure, causing traction of manually screening the references of relevant articles,
pain-sensitive structures in the cranium. Multiple factors, reviewing the proceedings of pertinent meetings, and
such as needle gauge, needle tip design, patient position contacting clinical experts in the field.
and operator experience, have been proposed to affect
the incidence of headache following lumbar puncture. Study selection
Of prime importance is the needle tip design. Spinal Three investigators will independently evaluate studies
needles can be broadly characterised as being atrau- for eligibility. Disagreements between the reviewers
matic or traumatic depending on their tip configur- concerning the decision to include or exclude a study
ation.4 Traumatic or ‘conventional’ needles are the most will be resolved by consensus, and if necessary, con-
commonly used. They feature a bevelled tip, designed to sultation with a fourth reviewer. Our inclusion criteria
puncture through tissue, with an opening at the tip to shall be:
facilitate collection of CSF or injection of therapeutics. ▸ Study design: randomised controlled trials (RCTs;
In contrast, atraumatic needles are blunted, with a including cluster RCTs and pilot studies), controlled
pencil-point tip and a side port for injection or collec- (non-randomised) clinical trials, prospective and
tion. Theoretically, atraumatic needles dilate the dural retrospective cohort studies, and abstracts, with no
fibres, splaying them during the procedure, and follow- publication type or language restrictions,
ing removal of the needle, allow them to gradually ▸ Population: patients of any age group and demo-
return to their original position, in contrast to traumatic graphic undergoing lumbar puncture as a part of
needles, which tear and damage the dural tissue.3 5 their clinical care,
Despite development of atraumatic needles early in ▸ Intervention: lumbar puncture with atraumatic
the 20th century, their use in clinical practice and knowl- needle,
edge of their existence among clinicians remains signifi- ▸ Control: lumbar puncture with traumatic needle,
cantly limited. In fact, as few as 2% of clinicians surveyed ▸ Outcomes: clinical outcomes such as incidence of
reported using atraumatic needles.6 This review will aim postdural puncture headache (PDPH; primary
to systematically examine atraumatic needles in compari- outcome), any headache, backache, hearing dis-
son to the traumatic type. This systematic review and turbance, nerve root irritation, traumatic tap, sever-
meta-analysis protocol has been developed in accord- ity of PDPH, duration of PDPH, number of patients
ance with the Preferred Reporting Items for Systematic requiring intravenous fluids/controlled analgesics
reviews and Meta-Analysis Protocols (PRISMA-P)7 guide- or blood patch, failure rate, success on the first
lines and has been registered with the International attempt and number of attempts required to obtain
Prospective Register of Systematic Reviews (PROSPERO; CSF.
registration number: CRD42016047546). For studies published more than once, we will include
only the report with the most informative and complete
data. We will exclude studies that did not evaluate atrau-
METHODS matic needles, studies without clinical outcomes such as
We shall conduct a systematic review and meta-analysis in vitro studies, review articles, letters, and correspon-
in accordance with the Preferred Reporting Items for dences or comments (figure 1).
Systematic Reviews and Meta-Analysis (PRISMA),8 9 the
Meta-analysis of Observational Studies in Epidemiology Data management and collection
(MOOSE) consensus statement10 and the Cochrane Literature search results will be exported from all rele-
Handbook for Systematic Reviews of Interventions.11 vant databases as .ris files or .ciw files containing the
This protocol will be amended and updated in conjunc- complete reference. EndNote X7 software will be used
tion with the PRISMA-P guidelines,7 and updated for reference management. Reviewers will develop and

2 Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478

 Open Access

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
Table 1 Search strategy for the MEDLINE electronic database using the Ovid interface
Database Search terms
MEDLINE 1 Spinal puncture/
1946–present 2 (Spinal adj2 (puncture* or tap or taps)).mp. (mp=title, abstract, original title, name of substance word,
subject heading word, keyword heading word, protocol supplementary concept word, rare disease
supplementary concept word, unique identifier)
3 Lumbar punctur*.mp.
4 Dural punctur*.mp.
5 Spine punctur*.mp.
6 ((Spine or spinal or lumbar or subarachnoid) adj2 block*).mp.
7 Spinal drain*.mp.
8 Spinal fluid drain*.mp.
9 Cerebrospinal fluid drain*.mp.
10 anesthesia, spinal/
11 Anesthesia, obstetrical/
12 Anesthesia/
13 An?esthe*.mp.
14 Myelography/
15 Myelography.mp.
16 (Spinal epidural adj2 (combined or block* or an?esthes* or technique* or procedure* or method*)).mp.
17 (Continuous spinal adj2 (combined or block* or an?esthes* or technique* or procedure* or method*)).mp.
18 or/1–17
19 Atraumatic needle*.mp.
20 Sprotte.mp.
21 Whitacre.mp.
22 ((Non cutting or noncutting or noncutting or pencil point* or pencilpoint*) adj3 needle*).mp.
23 Pencan.mp.
24 Gertie marx.mp.
25 Zimmon.mp.
26 Traumatic needle*.mp.
27 Quincke.mp.
28 Cutting needle*.mp.
29 Knife needle*.mp.
30 Standard needle*.mp.
31 Conventional needle*.mp.
32 Greene.mp.
33 (Green adj2 needle*).mp.
34 Spinal needle*.mp.
35 Lumbar puncture needle*.mp.
36 Tuohy.mp.
37 Crawford.mp.
38 Eldor.mp.
39 Hustead.mp.
40 Weiss.mp.
41 Wagner.mp.
42 Cheng.mp.
43 Crawley.mp.
44 Foldes.mp.
45 Bell.mp.
46 Brace.mp.
47 Huber.mp.
48 Scott.mp.
49 ‘Needle through needle’.mp.
50 or/19–49
51 18 and 50

pilot screening questions and forms based on the eligi- published in English, the full article will be translated
bility criteria. Prior to data abstraction, full articles of all into English. In addition, a medical expert fluent in
eligible studies will be retrieved. For studies not the original language of the article will be involved in

Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478 3

Open Access

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
Figure 1 Flow chart of study selection.

data management. Three primary reviewers will inde- assess headache; treatment of headache; number of
pendently screen the titles and abstracts obtained by the patients in atraumatic and traumatic groups; age of
search against the predefined eligibility criteria. Full patients in atraumatic and traumatic groups; body mass
reports will be obtained for all references that appear to index of patients in atraumatic and traumatic groups;
meet the eligibility criteria, or where there is any ambi- number of patients in atraumatic and traumatic groups
guity. Reviewers will then screen the full text and deter- given prophylactic intravenous fluids; number of patients
mine whether these articles meet the eligibility criteria. in atraumatic and traumatic groups instructed bedrest;
Where necessary, we will contact authors of relevant characteristics of headache reported; number of patients
studies to obtain additional information and resolve with postural headache in atraumatic and traumatic
questions about eligibility. Discrepancies will be resolved groups; number of patients with non-postural headache in
by discussion and consensus, and if necessary, consult- atraumatic and traumatic groups; number of patients with
ation with a fourth reviewer. Reasons for excluding postural and non-postural headaches in atraumatic and
studies will be recorded. traumatic groups; severity of headaches in atraumatic and
Data from selected studies will be abstracted independ- traumatic groups; duration of headaches in atraumatic
ently by the three primary reviewers and verified for accur- and traumatic groups; number of patients with backache
acy by the fourth reviewer. We will gather information in atraumatic and traumatic groups; number of patients
from eligible articles using data abstraction forms that treated with epidural blood patch for headaches in atrau-
include fields for: study first author; year of publication; matic and traumatic groups; success on the first attempt
journal of publication; language; study design; included with atraumatic and traumatic needles; number of trau-
centres; included countries; number of patients; number matic taps with atraumatic and traumatic needles; failure
of men and women; inpatients or outpatients; recruitment rate of atraumatic and traumatic needles; number of
period; eligibility criteria; method of randomisation; attempts required to obtain CSF with atraumatic and trau-
purpose of lumbar puncture; specialty of clinician per- matic needles; crossovers (atraumatic-to-traumatic and
forming lumbar puncture; patient position; atraumatic traumatic-to-atraumatic); serious complications (eg,
and traumatic needle-specific type; atraumatic and trau- hearing disturbance, nerve root damage, etc); ease of use
matic needle gauge; procedure for follow-up; scale used to of needles as reported by authors.

4 Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478

 Open Access

Risk of bias in individual studies backache, hearing disturbance, nerve root irritation,

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
Three reviewers will independently perform quality assess- traumatic tap, severity of PDPH, number of patients
ment. The Cochrane risk of bias assessment tool11 will be requiring blood patch, number of patients requiring
used to assess the risk of selection, performance, detec- intravenous fluids/controlled analgesics, failure rate and
tion, attrition, reporting and other biases among included success rate on first attempt, with atraumatic and trau-
randomised trials. If there is insufficient information pro- matic needles, will be analysed by calculating the relative
vided to make a judgement, we will categorise the study as risk (RR) with corresponding 95% CI. For continuous
having an unclear risk of bias and the original study outcomes, such as the number of attempts with atrau-
authors will be contacted for further information. Trials matic and traumatic needles and the duration of PDPH,
with one or more high-risk components will be judged as the weighted mean difference with 95% CI will be calcu-
having an overall high risk of bias. The Newcastle-Ottawa lated. In addition, pooled estimates of all incidences will
Scale13 will be used to assess observational studies on selec- be calculated across all studies for the atraumatic group,
tion, comparability and exposure. Studies that receive at then separately for the traumatic group. Random-effects
least one star in every domain will be categorised as being meta-analyses for all outcomes of interest will be per-
of high quality. Disagreements during quality assessment formed using the DerSimonian and Laird model.15
will be resolved through discussion, consensus and, if Weights will be calculated using the inverse-variance
necessary, consulting a fourth reviewer. method. The number needed to prevent harm (NNPH)
will be calculated using the following equation:11
Definition of outcomes
Our clinical outcomes will include multiple incidences  
 1 
after lumbar puncture. Headaches will be categorised NNPH ¼ 
into the following groups: PDPH ( primary outcome), ACR  ð1  RR)
and any headache. PDPH will be defined as headache ful- The threshold of type I error for statistical significance
filling the International Classification of Headache shall be α=0.05. All statistical analyses will be conducted
Disorders (ICHD) III criteria,14 which is described as an using Comprehensive Meta-Analysis V.3.3.070 (Biostat,
orthostatic headache occurring within 5 days of a lumbar Englewood, New Jersey, USA).
puncture, caused by CSF leakage through the dural punc- Between-study heterogeneity will be evaluated using
ture. This headache is usually accompanied by neck stiff- Cochran’s Q test and measured by the I2 statistic, with I2
ness as well as subjective hearing symptoms and remits values exceeding 25%, 50% and 75% being judged as
spontaneously within 2 weeks or after sealing of the leak low, moderate and high heterogeneity, respectively.16
with an autologous epidural blood patch. ICHD III diag- Publication bias will be evaluated visually by funnel plot
nostic criteria for PDPH are as follows: (1) headache has analyses and quantified by Begg and Mazumdar’s17 and
developed in temporal relation to the low CSF pressure Egger’s tests.18
or CSF leakage, or led to its discovery; (2) dural puncture Prespecified subgroup analyses will be conducted to
has been performed; (3) headache has developed within examine if covariates exist and to explore potential
5 days of dural puncture; (4) headache is not better heterogeneity for the primary outcome (PDPH). We
accounted for by another ICHD-III diagnosis.14 The plan to examine subgroups stratified by age, gender,
outcome of ‘any headache’ will encompass PDPH and all patient position during puncture, needle gauge,
headaches not fulfilling the previously discussed criteria. quality of studies, prophylactic use of intravenous
Furthermore, we will classify PDPH based on intensity fluids, postpuncture prophylactic bedrest, purpose of
using the visual analogue scale (VAS) as follows: mild lumbar puncture, continent of publication, older
(VAS 1–3; responds to over-the-counter analgesics and versus newer studies and specialty of physician per-
bedrest), moderate (VAS 4–7; requires controlled or forming lumbar puncture.
intravenous analgesics, or intravenous caffeine) and We will conduct sensitivity analyses to examine
severe (VAS 8–10; requires epidural blood patch). We will whether differently manufactured atraumatic or trau-
also evaluate the duration of PDPH, incidence of back- matic needles influence the incidence of the primary
ache, need for intravenous fluids/controlled analgesics outcome (PDPH). In addition, we will perform trial
or epidural blood patch, success rate on the first attempt, sequential analyses and calculate adjusted RR
number of traumatic taps ( presence of blood in CSF on values.19 20 Quality of evidence will be assessed using the
visual inspection), failure rate of needles, number of Grading of Recommendations Assessment, Development
attempts required to obtain CSF, incidence of hearing dis- and Evaluation (GRADE) approach.21 Evidence for all
turbance and incidence of nerve root irritation. outcomes will be judged for the domains of risk of bias,
consistency, precision, reporting bias and directness.
Data synthesis Evidence will be ranked as being of high quality
We will compare all outcomes for atraumatic versus trau- (further research is very unlikely to change confidence
matic needles across randomised trials using the in the effect size), moderate quality (further research is
intention-to-treat principle. Dichotomous outcomes, likely to have an important impact on confidence in the
including the incidence of PDPH, any headache, effect size/may change the effect size), low quality

Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478 5

 Open Access

(further research is very likely to have an important REFERENCES

BMJ Open: first published as 10.1136/bmjopen-2016-014478 on 31 March 2017. Downloaded from http://bmjopen.bmj.com/ on April 15, 2020 by guest. Protected by copyright.
impact on confidence in the effect size/likely to change 1. Horlocker TT, McGregor DG, Matsushige DK, et al. A retrospective
review of 4767 consecutive spinal anesthetics: central nervous
the effect size) or very low quality (any estimate of effect system complications. Perioperative Outcomes Group. Anesth Analg
size is very uncertain).21 Results from the analysis will be 1997;84:578–84.
2. Corke BC, Datta S, Ostheimer GW, et al. Spinal anaesthesia for
summarised using the GRADEPro software. Caesarean section. The influence of hypotension on neonatal
outcome. Anaesthesia 1982;37:658–62.
3. Thomas SR, Jamieson DR, Muir KW. Randomised controlled trial of
atraumatic versus standard needles for diagnostic lumbar puncture.
DISSEMINATION BMJ 2000;321:986–90.
4. Calthorpe N. The history of spinal needles: getting to the point.
Results from this study are expected to significantly Anaesthesia 2004;59:1231–41.
change clinical practice. By comprehensively examining 5. Cruickshank RH, Hopkinson JM. Fluid flow through dural puncture
sites. An in vitro comparison of needle point types. Anaesthesia
the evidence on atraumatic needles in comparison to 1989;44:415–18.
traumatic needles for lumbar puncture, this review is 6. Birnbach DJ, Kuroda MM, Sternman D, et al. Use of atraumatic
expected to inform clinicians on which needle to use spinal needles among neurologists in the United States. Headache
2001;41:385–90.
for procedures in order to minimise patient complica- 7. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for
tions. The results from this review will be submitted to a systematic review and meta-analysis protocols (PRISMA-P) 2015:
elaboration and explanation. BMJ 2015;349:g7647.
peer-reviewed journal for publication and will be widely 8. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for
presented at conferences and seminars. reporting systematic reviews and meta-analyses of studies that
evaluate healthcare interventions: explanation and elaboration. BMJ
2009;339:b2700.
Author affiliations 9. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for
1
Division of Neurosurgery, McMaster University, Hamilton, Ontario, Canada systematic reviews and meta-analyses: the PRISMA statement. BMJ
2 2009;339:b2535.
Division of Neurosurgery, University of Toronto, Toronto, Ontario, Canada
3
Department of Medicine, McMaster University, Hamilton, Ontario, Canada 10. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
4 observational studies in epidemiology: a proposal for reporting.
Division of Neurology, McMaster University, Hamilton, Ontario, Canada Meta-analysis Of Observational Studies in Epidemiology (MOOSE)
5
Health Sciences Library, McMaster University, Hamilton, Ontario, Canada group. JAMA 2000;283:2008–12.
6
Department of Clinical Epidemiology and Biostatistics, McMaster University, 11. Cochrane handbook for systematic reviews of interventions. 5.1.0
Hamilton, Ontario, Canada ed. The Cochrane Collaboration, 2011.
7 12. Sollenberger JF, Holloway RG Jr. The evolving role and value
Department of Neurology, Beth Israel Deaconess Medical Center, Harvard
Medical School, Boston, Massachusetts, USA of libraries and librarians in health care. JAMA 2013;310:
1231–2.
Contributors SN, JHB, WA and SAA conceived and designed the protocol. 13. Wells G, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale
SN, LB and SAA developed the search strategy and piloted it in all relevant (NOS) for assessing the quality of nonrandomised studies in
meta-analyses. [Webpage]. 2013. http://www.ohri.ca/programs/
databases. SN, JHB, WA and SAA developed the review protocol, selection clinical_epidemiology/oxford.asp
criteria, risk of bias assessment strategy, and data management and synthesis 14. Headache Classification Committee of the International Headache
methodology. SN, JHB, WA, FN, EB-C, AK, AS, LB, WO, MS, DS, KR, FF, SSi, Society (IHS). The International Classification of Headache
SSh, NZ, MS and SAA critically revised and commented on the intellectual Disorders, 3rd edition (beta version). Cephalalgia 2013;33:629–808.
content of the manuscript. 15. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Trials 1986;7:177–88.
Funding This research received no specific grant from any funding agency in 16. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring
the public, commercial or not-for-profit sectors. inconsistency in meta-analyses. BMJ 2003;327:557–60.
17. Begg CB, Mazumdar M. Operating characteristics of a rank
Competing interests None declared. correlation test for publication bias. Biometrics
1994;50:1088–101.
Ethics approval Not required. 18. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis
detected by a simple, graphical test. BMJ 1997;315:629–34.
Provenance and peer review Not commissioned; externally peer reviewed.
19. Thorlund K, Engstrom J, Wetterslev J, et al. User manual for trial
Open Access This is an Open Access article distributed in accordance with sequential analysis (TSA). Copenhagen, Denmark. 2011.
the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, 20. Wetterslev J, Thorlund K, Brok J, et al. Trial sequential analysis may
establish when firm evidence is reached in cumulative
which permits others to distribute, remix, adapt, build upon this work non- meta-analysis. J Clin Epidemiol 2008;61:64–75.
commercially, and license their derivative works on different terms, provided 21. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging
the original work is properly cited and the use is non-commercial. See: http:// consensus on rating quality of evidence and strength of
creativecommons.org/licenses/by-nc/4.0/ recommendations. BMJ 2008;336:924–6.

6 Nath S, et al. BMJ Open 2017;7:e014478. doi:10.1136/bmjopen-2016-014478